“Don’t let parallel brilliance become org entropy”

—Nathaniel Whittemore, The AI Daily Brief

At AI-native companies, every job is becoming a startup.

A BCG study of the companies furthest ahead with AI found just under two-thirds have removed procedural permission gates, and about half pushed decisions further down the hierarchy. Three-quarters now judge success by the quality of judgment, not the volume of activity.

Inside these companies, work looks different:

• Work is shifting from executing to framing the problem and setting the objective.

• Roles are blurring as people assemble around problems rather than org-chart positions.

• People take on more ambitious work, because a bold attempt costs less than ever.

• And instead of approval chains, companies build-in governance, access, versioning, and conventions into the system, so everyone can move faster with less risk.

And it seems to be paying off. Another BCG Institute study found that companies redesigning work and talent around AI, rather than layering AI tools on top of old approaches, outperform their peers by more than 11 percentage points a year in industry-adjusted shareholder returns.

And it isn’t confined to product development. Across these companies, people in every function are asking some version of the same question: “what can we do now that wasn’t possible before?”

PwC’s 2026 Hopes and Fears survey of nearly 50,000 workers shows who these people are. They are the “front-runners” with scarce skills and strong AI capability. They’re far more likely than average to say they’re rewarded for using AI effectively and for challenging how work gets done. They’re also less focused on avoiding mistakes. A second group, which PwC calls “AI insurgents”, is the people outside scarce roles using AI to gain a strong advantage. The rest of the workforce, the engine room, is rewarded for something else entirely.

Two very different mindsets, inside the same company. About a third of workers are already behaving like intrapreneurs, the rest are keeping the business running. Both are necessary. Most organisations have no way to connect them effectively.

But the front-runners won’t wait forever for their ideas to land. Almost a third of them (29%) say they’re very or extremely likely to change employer in the next year.

So every job may be a startup, but most companies aren’t built to run hundreds of them at once. Without the right systems, the parallel brilliance of their people turns into organisational entropy.

The problem with innovation labs

This isn’t the first time companies have tried to connect explorers with the engine room. The usual solution has been the innovation lab: a dedicated team, set apart from the business, free to experiment.

Most labs fail. They fail because they were structurally disconnected from the organisation that has to judge, adopt, resource and scale what they find through experimentation. The lab itself was never the problem, the handoff from exploration into coordinated organisational action was.

Josh Woodward, VP of Google Labs, made a similar point in a recent interview with Lenny Rachitsky on Lenny’s Podcast. Labs fail when they aren’t connected to the rest of the organisation. They build something clever, everyone admires the demo, then nothing happens. The company simply can’t absorb it.

If the failure point is often the handoff, it’s worth asking what’s being handed off. Labs produce things: demos, features, prototypes. But adoption of new things doesn’t happen because a thing is good, it happens when people change what they do.

So maybe the problem isn’t the experimenting. Maybe it’s that we only know how to experiment with one kind of thing.

Towards a broader understanding of innovation

It’s not just labs. When most companies say exploration, they mean product. Innovation teams run design sprints. Hackathons end in demos. The whole vocabulary assumes there’s a thing that gets produced at the end: a feature, an app, something you can put in front of a user.

Design Thinking principles have shaped the vocabulary of that world. First you empathise, then define the problem, then ideate, then prototype, then test. It works well when there is an artefact to iterate on.

But most of what makes a company work isn’t an artefact. How we price, how we sell, how we onboard new hires, how we make decisions or structure our compensation plans. We innovate in these areas constantly. We just don’t call it innovation.

And because we don’t, we overlook the creative talent working within them. Nobody has a clear way to start their ideas, test them or decide what to do with them.

This isn’t just speculation. Larry Keely and the innovation consultancy Doblin studied more than 2,000 successful innovations to understand where breakthroughs actually come from. In their book, Ten Types of Innovation: The Discipline of Building Breakthroughs, they explain that innovation is much bigger than new products or technologies. They sort innovation into ten types: profit model, network, structure, process, product performance, product system, service, channel, brand and customer engagement. Only two are about the product itself.

The most interesting finding is that the strongest innovations rarely rely on one type. They combine several. A new profit model with a new channel, and a new way of engaging customers, rather than competing on product performance alone.

Which means the most valuable innovations are cross-functional by nature. They need pricing, sales, operations, marketing to move together. Exactly where the handoff beaks.

AI just gave everyone a lab

Before AI, testing a new approach, running an experiment or conducting research was expensive, both in time and money. For this reason, exploration lived where the most obvious gains could be measured and validated: product performance. Not anymore.

Now a sales rep can test five pitch angles on real prospects in a week. A finance manager can model a new pricing structure in a day. An ops lead can play out a different workflow without touching a single existing process. The cost of trying something new has collapsed everywhere, not just in engineering.

Anthropic’s research shows how early we are in this is transformation. Even in computer and math, where use is the highest, AI covers only about a third of what it theoretically could. For all other functions, the gap is wider. The capacity is already there. The exploration has barely begun.

So companies have given every function a lab’s worth of capacity, but, like a product lab, few have a method for turning all that creative output into differentiated value.

And therein lies the problem. It’s likely one reason 95% of enterprise GenAI pilots reportedly fail to deliver measurable P&L impact. Exploration gets judged with the tools of exploitation: a business case, a budget cycle, a roadmap. Most pilots never stood a chance.

The creative spark happens in every function

Everyone is capable of being creative. For many of us, school and corporate life beat it out, but look closely and you’ll see it in every function. The account manager who sees a new pattern in the way customers articulate their feedback, the finance person who spots a new way to close the month, the sales person who identifies a unique niche channel for finding leads.

Until recently, most of these ideas died in someone’s head. But AI has changed that. A spark can become a draft, a model, a working mock-up in a single afternoon.

Making something is now cheap. Making sense of it isn’t. The idea arrives looking finished, without the story of why it matters, who has to act on it or what happens next. So the work of judging it falls on whoever receives it. Multiply that across the entire organisation and the so-called “slop grenades” start flying.

Not all of it is slop. Some of it is a bold attempt.

And attempts are good. AI let’s people step outside their lane, and that’s often where the better ideas are. In a randomised trial with 791 professionals at P&G, people without AI stayed in their lane: R&D professionals proposed technical solutions, commercial professionals proposed commercial ones. Given AI, both groups produced balanced solutions whichever side they came from, and the quality of work improved.

AI lowers the cost of accessing adjacent expertise.

The trouble is that most organisations weren’t built for this volume. AI has radically increased our capacity to generate and explore new possibilities. Our capacity to evaluate them, coordinate around them, and turn the best ones into action hasn’t kept pace.

So new ideas either get ignored, or get pushed through the planning machine too early. They get evaluated using tools built for exploitation, not exploration. So the ideas die there too, just more slowly.

This short video below shows why AI has moved the bottleneck to alignment and coordination, and why the handoff from exploration to exploitation is harder than ever.

We get more ideas, more drafts, more demos. The divergent zone explodes. What doesn’t expand is what expert facilitator, Sam Kaner, calls the “groan zone”: the uncomfortable stretch where a group looks at all that output and decides what it means, and whether anyone will act on it. That’s the new bottleneck.

Every innovation is a plot. Some produce an object.

Part of the problem is how we’ve been taught to be creative. Most creativity training runs on one engine: divergent thinking. You generate lots of ideas by recombining what you already know. Brainstorming, ideation workshops, and most design education are built on it.

, author of Storythinking and Primal Intelligence, and his colleague Mike Benveniste at Ohio State’s Project Narrative, argue that this engine has a problem. Decades of training have made people better at divergent thinking tests, without the expected gains in actual innovation. More ideas do not necessarily mean more breakthroughs.

And now AI can generate and recombine existing ideas at a scale humans never could. So we’ve increased our capacity for the part that was never enough on its own, and the genuinely creative ideas end up downed out by the slop.

Angus Fletcher and Mike Benveniste propose a second engine for creativity training: narrative cognition.

That’s the language most companies are missing. Every innovation is the plot of a story, and outside product, there often isn’t an object at all.

A new pricing model is a story. If we raise the entry price, small customers will likely churn, but the mid-market expands. If sales changes how it qualifies prospects, finance sees a different revenue curve. You can’t prototype this. You can only play it forward.

Angus Fletcher and Mike Benveniste describe three ways to train narrative cognition:

1. World building: something unexpected happens. Instead of dismissing it, ask: what if this is the new normal? If so, what becomes possible?

2. Perspective shifting: temporarily borrow someone else’s causal model: how another person believes cause and effect works.

3. Action generating: create a collision and simulate what happens. Introduce a new actor into an existing environment, or an exiting actor into a radically different one, and ask what actions would follow.

The more we innovation as a narrative, the easier the handoff between exploration and exploitation becomes, and the more creative ideas turn into throughput and differentiated value. Because the plot already includes the people who have to act on it. Sales, marketing, finance and ops aren’t an afterthought, they’re characters in the story from the start.

The dual operating system of innovation

If every function can now explore more broadly and more boldly than ever before, the lab can’t be a separate place. It has to be a loop. And to design that loop, you have to start with how creativity actually works.

In his interview, Josh Woodward was explicit about where his team’s best ideas come from. On walks, over lunch. In the middle of the night. While on holiday.

Anyone who loves their craft will attest to this. Creative solutions to complex problems often come when you least expect it, yet traditional organisations expect you to solve problems between the hours of nine to five, Monday to Friday, while seated at your desk, in front of your computer. This isn’t how the creative process works.

Creativity moves in waves, both within an individual and within a team. Someone gets a spark and goes deep - they build, test, narrow in. Others nearby see it too and volunteer to support the investigation. Someone else needs to step out, to wander, to “fuck around and find out” as my co-founder and I like to say.

You can’t call a meeting for Monday and say: “now we will brainstorm an effective solution to this problem”. The best you can do is align on the problems worth exploring, then ride people’s curiosity when it lights up.

John Kotter, emeritus professor of leadership at Harvard Business School, describes a structure built for exactly this. He calls it a “dual operating system”. MIT’s Edgar Schein described something similar as “parallel structures”.

The traditional hierarchy keeps running the business, while a network of people drawn from across the hierarchy explores what’s next. John Kotter calls them a “volunteer army”: the people who had the spark at 3am on a Saturday and feel compelled, almost by an invisible force, to investigate, to play it forward and see what happens to the business if it works.

The crucial detail is who sits in the network. They’re not a permanent “lab team” but rather volunteers from every level of the organisation who keep their day jobs. Which makes them the same people who will have to adopt, resource and scale what they find. And the membership is fluid. Initiatives form around a specific opportunity, run, then dissolve, and people join and leave as their energy and the work dictate.

Exploration is a different kind of work

In order for the dual operating system to be effective, the network and the hierarchy need to operate with different rules.

Complexity science and the Cynefin framework help make this distinction. Most of what the hierarchy handles is the complicated work: the known unknowns that are discoverable through analysis. For complicated work, experts primarily use deductive and inductive reasoning: they apply what’s known, generalise from what’s worked before, plan a solution and execute it. The management mode is to sense-analyse-respond.

What the network explores is complex work: the unknowns are unknown, and cause and effect only become coherent in hindsight. The answers only appear as you act. So the volunteers reason abductively, noticing what emerges and making their best guess at what explains it. They probe-sense-respond.

That’s why tools built for complicated problems kill the exploration of complex ones. They ask people to predict the outcome of something that can’t be predicted yet.

So how should people act when they can’t predict what will happen?

Saras Sarasvathy, professor at the University of Virginia’s Darden School of Business and one of the most cited researchers on entrepreneurship, answers this question. Her research found expert entrepreneurs have a unique way of making decisions under high degree of uncertainty. She calls it “Effectuation”, and four of its principles describe how a healthy network, and the intrapreneurs in it, behave.

• First, they don’t start with a goal and work backwards. They start with what they have and work forwards, letting the goal take shape as they go.

• Second, they use surprises instead of trying to stay on the planned path.

• Third, they think in terms of affordable loss, not expected returns. They know how much time, money, and reputation they can risk in order to reduce their uncertainty.

• And fourth, they build with whoever commits. Not a hand-picked team of the “right” people, but volunteers drawn in by the creative spark, who shape the direction by showing up.

This is how probing works in practice. Small bets, capped losses, volunteer partners. And when a probe reveals something real, the work changes character. It moves from complex to complicated, from probe-sense-respond to sense-analyse-respond. From exploration to exploitation.

That transition is the moment most companies miss. Partly because it’s invisible.

When a probe starts working, nothing visible changes. It’s still the same people, same side project, same half-finished idea. So it either gets pushed into the planning machine too early and dies being scrutinised with a business case. Or it lingers as a side project nobody scales and nobody kills. Until eventually the volunteers running it lose interest or leave.

That transitions needs to be marked clearly with a decision. Made deliberately and out loud. It’s the moment the group says: “we know enough, this is no longer an experiment.” Or just as important: “we’ve learned what we needed to, and we’re stopping.”

That’s what a decision gate is for.

The governance loop

That’s where governance comes in. Not governances as control, but governance as connection. The point is to make sure exploration actually reaches a decision.

Here is what this governance loop looks like:

1. Everything starts from shared assumptions: this turns the loop into learning rather than a sequence of disconnected experiments. Every plan rests on beliefs about customers, the market and how people will behave. Very often these assumptions are invisible, so nobody knows when they’ve been proven wrong. With an explicit layer, a probe becomes “which belief did we test?”. And the AI signal detection gets something concrete to watch for. An anomaly is simply reality departing from an assumption.

2. Signals come in from everywhere: operational planning sets trip-wires. Strategic foresight anticipates scenarios (i.e. plots) to test. AI flags anomalies nobody was looking for.

3. Probing tests narratives and prototypes: small experiments, in any function. A new pitch, a pricing test, a changed handover. The goal is signal, not success: which belief held, which didn’t.

4. Experimental evidence goes to a decision gate: a group of people with good judgment and expertise get together and answer two questions: what does it mean? and will we act on it?

5. The decision gate sends the decision where it belongs: scale what works into operational planning. Update shared assumptions, so foresight and planning build on what was just learned. Escalate when immediate direction should change.

The manager’s role in the dual operating system

In the network, managers take part on equal footing with the volunteers. Rather than routing information and having veto power, as they do in the hierarchy, they act as coaches and facilitators. Most importantly, they give legitimacy to what the network commits to. They keep it aligned with the rest of the organisation and help decide where each idea will have the most impact.

AI is tuning this into something closer to a set of new jobs. In an episode of the AI Daily Brief entitled “Why agents make every job a startup”, host Nathaniel Whittemore names three:

• Intrapreneur Coaches: they support the people doing the exploring with judgment, and pacing. They help people know when to push, when to pause and when to let an idea go.

• Experiment Portfolio Managers: they treat experiments as a portfolio: fund, scale, merge, kill. This is the decision gate as a job, someone whose work is deciding which probes earn more resources and which have taught all they can.

• Internal Agent Product Managers: they curate the company’s agent ecosystem: the tools, data, agents and policies everyone explores with. If AI is what makes probing cheap, someone has to own the quality of what people probe with, including security and privacy guardrails.

None of these is a classic manager. All three exist to help exploration reach a decision.

Innovation is a plot. Every chapter ends with a decision.

A plot is a chain of decisions. Someone acts, something happens, someone decides what it means and what comes next. Probing plays the story forward, but the decision gate is where each chapter ends.

And this is where most organisations lose the plot. The group meets to pass judgment, they weigh the evidence, they argue, they agree. The probe crossed from complex to complicated, or gets killed, or an assumption everyone relied on turns out to be wrong. Then the meeting ends, and the decision dissolves into notes nobody looks at, or a Slack thread only three people saw. The next probe builds on the assumption that just got invalidated, because nobody outside the first room was told. No decision was recorded.

Startups rarely have this problem early on. The founders are in every conversation. They remember how the last chapter ended. An intrapreneur in a large firm doesn’t have that luxury. She’s one of hundreds of people exploring things at once, across functions, each with their own rhythm.

So if every job is now a startup, and every employee an intrapreneur, then the company needs what founders carry in their heads: a shared record of what was decided, by whom, and whether it held.

That’s what we are building with Afoot. When a group makes a decision, Afoot captures it and asks the people who made it to confirm it. Once confirmed, it becomes a commitment both people and agents can act from.

The network can see which bets are open, which held and which were killed. The hierarchy can see what has crossed the gate and is now theirs to plan. And everyone can see which assumptions changed along the way.

AI has given every function a lab. But without a shared record of what those labs decided, the parallel brilliance of volunteers turns into organisational entropy.

Labs don’t fail for lack of ideas. They fail for lack of a plot.

Write the next chapter with us

Afoot goes live soon and the waitlist is growing. Right now, we’re looking for a few design partners: companies where people are already exploring, where good ideas keep getting lost between the meeting and the plan, and that want the coordination infrastructure built around how they actually work.

If that’s you, email me at vanessa@afoot.net, or join the waitlist to follow our journey.

Join the waitlist

In late 2024, Gartner predicted that by the end of 2026 one in five organisations would use AI to flatten their structure, cutting more than half of their middle management positions.

This is more than a rumour. AI-native companies like Block, Cognizant, Lovable, and Every have begun to put these ideas into practice. And they are seeing results. We should all be paying close attention to what they are doing.

Look closely at these companies and it becomes easier to see why they are getting results from AI, and why others are shelving their AI initiatives entirely. They clearly don’t have it all figured out just yet. Nobody finds the clean version on the first try. But there are some interesting patterns emerging.

The best descriptions of where this is all headed come from the tech companies at the frontier. Jack Dorsey, CEO of Block, and Roelof Botha wrote a widely shared article, From Hierarchy to Intelligence, about how they’re rebuilding Block around AI. Then, , a well known growth operator working at Lovable, wrote her field notes from a year inside a truly AI-native company. And finally, , CEO of Every, wrote “After Automation” about how his company works now that it has automated everything it can.

To make this transformation, you first have to change your assumptions about how a company is supposed to be managed: the assumption that a company is its org chart, that managers keep the people beneath them aligned, that innovation occurs in a designed lab, and that meetings are mainly for passing information around.

This is my attempt at translating what these new beliefs mean for everyone else. The hard part of reading this won’t be the new ideas presented. It will be to let go of the obvious ones you assume are settled.

Here is what this article will cover:

1. What’s changing for the individual

2. What’s starting to break as the system transforms

3. Where the bottleneck in the system is shifting

4. What new roles are emerging as agents join the team

5. What management infrastructure is missing to make it all work

First, what’s changing for the individual

Let’s start by looking at how this shift is impacting employees. There are seven shifts going on at the individual level. The seventh is the most important.

1. The lines between job titles are fading: At Every, people in operations and customer service do work that used to need an engineer, and all engineers talk directly to customers. At Lovable, senior people who used to have dozens of direct reports are happily building things themselves. Traditional functional job titles hold less meaning. The impact of this has been measured too. In a randomised trial with 791 professionals at P&G, people without AI stayed in their lane: R&D professionals proposed technical solutions, commercial professionals proposed commercial ones. Given AI, both groups produced balanced solutions whichever side they came from, and the quality of work improved.

2. Everyone can ask the company what it knows. Jack Dorsey called it a World Model. Lovable calls it the Company Brain. Both are centralised systems that ingest every useful data point from the company’s various disparate systems, every Slack message, document, meeting transcript, and KPI to create a continuously updated digital twin of the organisation. It’s highly relevant context available to everyone, all the time, so everyone operates from the same basic information.

3. Agents do real work and everyone is a manager of agents. At Every, when you tag a colleague in Slack it’s 50/50 whether you’re talking to a person. At Lovable, every agent has a named human responsible for what it produces; a knowledgeable “agent parent”. Directing, reviewing and correcting an agent is a normal part of everyday work.

4. People take on more ambitious work than they could have tried before. The ceiling on what one person can attempt has gone up, so the work that was too expensive to start becomes possible.

5. With fewer permission gates, everyone becomes a founder: When an experiment costs a few tokens and a day, nobody needs a committee to approve it. The controls haven't gone. They’ve just moved. As describes it, this means a non-technical employee can take an idea to production with the governance built in. This used to be called intrapreneurship, and was reserved for a designated group. Not anymore. A tool someone builds for themselves can become a team’s agent, then a shared pattern, then an internal product.

1. It feels good. In Lenny Rachitsky and Noam Segal’s survey of tech workers this summer, 82% said AI makes them measurably more productive and 49% are said they feel amplified (able to do more, and better). The P&G trial measured the emotional side and found that people who worked with AI came away with more excitement, energy and enthusiasm, and less anxiety, frustration and distress, with the biggest lift of all for teams using AI. Working alone with AI matched or beat the emotional boost people get from having a human teammate.

   

The seventh shift: the unique human advantage

The biggest shift for the individual is in the unbundling of their value. With AI agents doing more of the production work, they’re discovering what makes them uniquely valuable — their judgment, taste and intent.

To understand what 'human judgment' means, it helps to be precise about what kind of knowledge judgment actually is.

The cognitive scientist John Vervaeke splits knowledge four ways:

• propositional: knowing that

• procedural: knowing how

• perspectival: knowing what a situation looks like from where you're standing;

• participatory: knowing what it is to play your part in it.

AI absorbs the first two kinds of knowing. Real human judgment exists in the last two.

A diagram in Dan Shipper’s article, After Automation, does the best job of explaining how judgment becomes a unique advantage in the AI era.

Essentially, you have two converging realities. On the left side, LLMs learn from work that has already been done, which opens the floodgates for every individual regardless of title, experience or function to make an attempt at things. To make use of an increasingly intelligent corpus of propositional and procedural knowing.

And on the right, you have the humans who know the live situation. The people who can see it from where they stand, who know what matters right now, and who are part of the thing rather than just looking at it. That is judgment formed from perspectival and participatory knowing. And it is something AI can never do.

Jack Dorsey says something similar in his article, about how people sit at the edge of the company, where intuition, trust and “feeling the room” really matter. Elena Verna explains how people ask themselves: “but is this lovable?” before they commit to a path.

Second, what’s starting to break

While everything described up until now sounds exciting, there are some less welcome by-products showing up as these companies transform. It's still early and correlation isn't causation, but the companies furthest ahead are already seeing where the costs accumulate. Here are a few:

• More gets made, but more of it looks the same. When everyone can produce a decent draft of anything using a tool that is designed to spit out an average of everything that has already been done, someone with real expertise still has to review it, improve it or throw it away. Otherwise we converge on sameness. And sameness is a commodity. Cheap execution makes bad judgment more expensive.

• Work collides and stresses people out. Since starting something is cheap, people start all sorts of things. They don’t wait to be assigned things, and they don’t check with anyone first. So two people pick up the same problem from different ends, and neither finds out until the work is done. Collisions are a major source of uncertainty for people. And heightened uncertainty is one of the leading causes of burnout.

• Managers stop hearing things. A TNG survey found that 60% of respondents ask AI, at least once a week, questions they would have previously asked their manager. When people take their questions to AI, managers lose one of the main ways they find out where things are unclear.

• Being an effective manager is getting harder. It was a hard job even before AI. In a 2025 survey of tech workers, only 26.6% rated their manager as highly effective, and 42.3% rated theirs as ineffective. AI makes it even harder. Managers have less to go on, because the questions that told them what was unclear now go to AI. They have more to hold together, because work moves faster and collides more often. And they're squeezed from above: in TNG's survey, line managers were the group most worried that leadership expects too much from AI, at 36% against 23% of senior executives. This matters because of what an effective manager is worth. People with great managers reported 63% more belonging and 31% less burnout than people with ineffective ones.

  

• People are moving faster and feeling worse. In the survey of tech workers, burnout rose from 44.7% to 55.7% in a year. People worry about being expected to do more for the same pay (51%), an unsustainable pace (46%) and the quality of their work going down (41%). The same fears show up in Sweden. In TNG’s survey, the top concerns were wrong data leading to wrong decisions (51%) and human judgment being pushed aside (50%).

The judgment paradox

There’s a paradox worth stopping for here. If judgment is where human value is moving, how can it be that it is the thing 50% of people surveyed are afraid of losing?

Part of the answer is that we're not all talking about the same thing. At Lovable, people automate themselves on purpose:

What the people at Lovable are automating are the tasks that run on propositional and procedural knowing (i.e. the what and how tasks). What they aren’t voluntarily giving away is the judgment that decides whether the work was any good. Judging AI output mostly means spotting mistakes, questioning assumptions and rejecting sub-par quality.

Unfortunately, almost nobody is measured on that. KPIs are often about speed, volume, output. You don’t get rewarded for slowing down because something isn’t quite good enough. Add the pressure to do more in less time, and the rational move is to hand more of the judgment to the AI. Even when you shouldn’t.

The second reason is more dangerous. AI has separated our ability to produce something from our ability to judge it. A lot of what we now attempt sits outside our expertise. So we send the draft to someone whose judgment we actually trust on the matter, someone with real domain knowledge. The cost of judging lands on the person who didn’t make the attempt in the first place. As explained in the diagram above, the experts are now flooded with things they need to pass judgment on. So who could fault them for outsourcing some of that judgment to AI? If there is little incentive to catch mistakes, why would you risk burnout trying to spot them in someone else’s work?

Lastly, there may be something more sinister going on. Something I suspect many experts are beginning to think about after Meta began installing software on employees’ computers to capture mouse movements, clicks and keystrokes in order to train their AI models. It’s what economists from economists from MIT, Washington University and UCLA wrote about in a paper called Some Simple Economics of AGI. In it they describe “the codifier’s curse”, which explains that our current “human-in-the-loop” equilibrium is unstable. The experts carrying the judgment are, in the course of doing their job, converting their judgment into training data. They are codifying their own obsolescence.

Third, where the bottleneck is shifting

Now let’s step back. Every system has one step that sets the pace for everything else. Speed up the other steps and work just piles up in front of it. That's the bottleneck.

Previously, production was the bottleneck. Making things was the expensive part, so we coordinated in order to produce. But AI has flipped that.

The cost of production has fallen below the cost of coordinating.

Producing is now the easy and cheap part. Deciding what to produce, agreeing on it, and making sure multiple capable people aren’t all solving the same problem in different ways without even knowing it, is not. Coordination, which includes alignment, is the new bottleneck.

The reason coordination has become more difficult than before goes back to Dan Shipper’s diagram, which I have repurposed to make my point.

The numbers used are illustrative. It’s the gap between them that isn’t.

So let’s say execution has gone up by a hundred times. Expert judgment has gone up too, because an expert can now pass judgment on more work than ever before. But that time is finite, so let’s say that’s gone up by ten times.

But coordination hasn’t moved at all. Who decided what, who owns what, what’s still open: that still runs at the speed of people talking to each other, the same as it did twenty years ago.

So both sides of the diagram converge on the one step that didn’t get faster. That’s where the queue forms.

It’s tempting to think the Company Brain or World Model alone solves the coordination problem. It solves the first step. It gives every person information.

But individual access to more information isn’t the same as shared sense-making or committed action. Knowing more does not automatically mean people understand the situation in the same way, agree on what matters most, or know what should happen next. That still occurs in conversations between people.

Elena Verna’s story of work colliding happened in a company where almost everything is visible to everyone. So what was missing wasn’t information. It was agreement.

Fourth, emerging new roles

In their article, Jack Dorsey and Roelof Botha put forward three new roles. None of them are managers.

1. The Directly Responsible Individual (DRI), who owns a single outcome for a set period and can pull in whoever they need.

2. The Player-Coach, who still does IC work and develops the people around them. They replace the traditional manager whose primary job was information routing.

3. The Individual Contributor (IC), a specialists and experts who, because of the world model, can make better decisions without being told what to do.

In an episode of the AI Daily Brief entitled “Why agents make every job a startup”, host Nathaniel Whittemore goes further. He lists nine new jobs AI-native companies are likely to need, grouped into three categories:

1. Technical roles: agent ops engineers keep the fleets of agents running. Context librarians curate what the agents know. Eval engineers build the quality checks their output has to pass.

2. Managerial roles: experiment portfolio managers decide which bets to fund, scale, merge or stop. Intrapreneur coaches help people with judgment and pace. Internal agent product managers look after the company's own agents as products.

3. Coordination roles: Coordination architects design how the organisation stays legible. Information pipeline owners route the right signal to the right place. Orchestration leads broker the conversation when work overlaps.

Those three coordination jobs, which include keeping people aligned, are real work. They are what an effective manager does, and always has been. And today they are a company's greatest point of leverage. Look back at the diagram I adapted from Dan Shipper to understand why.

Everything AI produces and everything the experts know has to pass through this one step of coordination. Done well, it turns a flood of output into actual differentiated value. Done badly, you get commoditised sameness, faster. Whether it’s done well comes down to people.

The judgment, taste and intent that make the difference only show up when people are engaged and committed to their work. In the survey of tech workers, the people who rate their manager as effective are the ones who report being engaged and committed. And the gap is enormous. Workers with extremely effective managers are 8.2 times as likely to be committed to their role as workers with ineffective ones: 61.6% against 7.5%.

So what a company most needs from its people depends on the person with the least support. An engineer has a whole toolchain, and a salesperson has a CRM. The person responsible for getting the best thinking out of a group currently has a calendar and their own memory.

What’s missing is infrastructure that makes managers more effective.

Fifth, the missing management infrastructure

Jack Dorsey and others are right that companies move at the speed of information flow, and that hierarchy slows it down. And the World Model fixes that. It gives people what they need to know to make good decisions.

But managers did more than that. They were coordinating the whole dang thing.

Coordination has two parts: information is one and alignment is the other. But alignment isn’t something you retrieve. It gets made, between people, in conversation. That’s where a group works out what the information means, what matters, and what they’re going to do about it. It’s where good judgment on complex problems is exercised.

As agents handle more of the routine and the complicated, complex problems will increasingly be the domain of people.

If the quality of collective human judgment is what separates differentiated value from commoditised sameness, then the place where that judgment is formed deserves its own infrastructure.

Put the whole picture together. What a company needs hasn't changed. It has always needed information, judgment and alignment. What has changed is where they come from.

Judgment, alignment and information used to arrive through one role, the manager. AI has unbundled them. Now they come from three different places, and each needs its own kind of support.

Take any one away and you get some or all of the problems listed earlier.

• Information and judgment without alignment is the colliding work that Elena Verna wrote about.

• Information and alignment without judgment is what the sameness Dan Shipper describes, well coordinated.

• Judgment and alignment without information is where most companies were before AI: people who agree, but on incomplete, uneven and sometimes wrong information.

In the AI era, differentiated value sits in the middle, where all three meet. The World Model is being built by every AI-forward company. The people with good judgment are already there. The third circle is the one with nothing underneath it.

And it matters, because having judgment alone isn’t enough. A company only benefits from what its people know, notice and decide if it can turn that collective judgment into coordinated action.

This is why the conversations among experts have to produce more than they do today, and why what they produce has to become an enduring company asset. A reliable record that lets increasingly independent people (and agents) act in a coordinated way.

For that, the record has to be reliable. And it’s only reliable if the experts who were in the conversation keep authority over what they meant. AI can interpret what a group of people said. It can’t decide what they are standing behind. The people involved have to confirm their intent before anyone else can act on it.

Afoot listens to the conversation and works out what was decided, who committed to what, and what’s still open. Then the people who were in the room confirm it. And from there, a team of agents does the coordination work. Because the agents maintain the record, they know who is doing what, they route the right information to the right people, and they prevent work from colliding. Their starting point is always the reliable record of human intent.

With that in place, the new roles described earlier can work as intended. The DRI who owns an outcome doesn’t have to chase it, and the Player-Coach doesn’t have to carry everyone’s context.

What’s left for both roles is harder and worth more. As Houda Nait El Barj writes, the more of our day we spend working with AI, the more we need human connection, inspiration and direction.

I’d put it more plainly. The leader’s job becomes orientation and facilitation: helping a group see the same situation, decide what matters, and mean the same thing when they agree. That was always the part of management worth having. We just never gave anyone the time and support to do it effectively.

AI can make each of us more intelligent. But agreeing is still something we have to do with each other.

Build the infrastructure with us

Afoot goes live soon and the waitlist is growing. What we’re looking for now is the right design partners: a few companies that feel the pull towards working this way and want the coordination infrastructure built around how they operate.

If that’s you, email me at vanessa@afoot.net, or join the waitlist to follow our journey.

Join the Waitlist

Other relevant articles

It’s for this reason Afoot asks teams to do one thing in meetings: say what you decided, and who’s doing what. Out loud. Clearly. On the record.

This sounds so obvious it barely counts as advice. And yet, for many, it is like a request to get undressed. People go quiet. They begin to hedge and find reasons to “circle back” next week.

So I’ve been asking myself WHY

Why can’t we make decisions? Why aren’t we willing to clarify our commitments to one another face-to-face? Something that should feel intuitive and completely normal, feels foreign and uncomfortable instead. I don’t think it’s because people are lazy or scared.

I think something in the environment has changed.

A team is made by what they choose

Soren Kierkegaard, the Danish philosopher and one of the foundational thinkers of existentialism, was concerned with how a person actually lives, chooses, and becomes themselves. He described a way of living called the aesthetic life. The aesthete lives for what comes next. He chases novelty and endless possibilities. Every choice gets postponed because choosing would close a door, and the aesthete’s whole policy is to keep the doors open. It’s a philosophy many of us, myself included, lived by for most of our 20s. Kierkegaard’s point was that this person never really becomes anyone.

Because, according to him, a person is made by what they choose.

I’ve begun to wonder whether a lot of us, well past our 20s, are stuck in some version of this now. Abundance and optionality is the code we live by. Portfolio careers and superficial relationships held until something better comes along. Every decision reversible and every commitment provisional. Everything a tentative “yes”.

And I think it’s spilled into our professional lives too.

Sit quietly in your next meeting and watch the room avoid a decision. What you’ll see is the aesthete. The nods that mean nothing. The same topic that comes back week after week as if it’s never been discussed before. There is no disagreement because actual disagreement would be a choice. What you will witness is optionality preserved.

Fake agreement IS optionality preserved.

A life of abundances doesn’t make us more decisive. It makes deciding feel unnecessary. Why pick a path when you can keep all of them?

The tech industry may be the worst offender of all

The aesthete philosophy seems to dominate the tech industry right now. We took Jeff Bezos’ “two-way door”and applied it to everything. If any decision can be undone next sprint, no actual decisions need to be made this sprint. Optionality has become our collective personality. The person who commits looks naive and close minded. The person who keeps the options open looks smart.

I’ve sat in enough tech leadership meetings to know what this mentality produces. It becomes product roadmaps nobody believes, priorities that get “revisited” quarterly, which actually means monthly, which really means whenever someone senior has a new idea or got inspired by the latest Lenny’s podcast. It’s ownership that is assumed instead of accepted.

And then, just when the capital markets were about to force us to converge, to take some modicum of accountability, we were given the most powerful option-generating machine ever built.

AI is terrible for a team that can’t make decisions.

With intelligence on tap, we now have three drafts where there used to be only one, five prototypes built while we sleep. A deck that explores every direction with an appendix for the best-medium-worst case scenario calculated to perfection. Every meeting now ends with more possibilities than it started with. Unfortunately, it also ends with the same amount of commitment as before, which is none.

We were already drowning in options, AI just turned on the tap.

Fake agreement is sensible math

Now, I want to be careful here. Because I’m not trying to turn this into a lecture about character. That is not it. This behaviour makes perfect sense.

As growth and marketing leader for startups, I’ve spent a great deal of time guiding teams through the “groan zone”. A term coined by the expert facilitator Sam Kaner. The “groan zone” is the uncomfortable stretch in any complex project where a group tries to move from ideas to decisions, cognitive overload kicks in, and leaders often skip the hard work of integration. The result is often watered-down compromise instead of real alignment.

Gaining wisdom requires sitting in the “groan zone”. Innovation requires it too. But people will only sit in that discomfort when they are confident that what comes out of it will actually lead somewhere. Nobody volunteers for hard conversations whose outcome disappears when the meeting ends. When decisions don’t stick and accountability is vague, avoiding the friction isn’t a failure, it’s just sensible math.

So that’s the thing. We aren’t shallow people, we are acting quite rationally. We’ve done the math in this environment of abundance and concluded that committing costs more than it pays. So instead, we drift and call it flexibility.

A lack of psychological safety is a part of the problem, but for many capable teams this has much more to do with a lack of infrastructure and incentives than it does with fear of retribution.

Which means the fix isn’t a workshop on decisiveness. Leaders have to deliberately engineer dissent, and make sure it pays well to be decisive and make commitments.

You don’t fix a pricing problem with a pep talk. You change the price.

While AI is giving us endless options and our agents are collapsing the cost of production, the only thing left is the deciding. Which draft? Whose call? Who owns it? What are we actually committing to? Because there is more output to decide about and less time between rounds, the judgment and convergence part is now the expensive bottleneck.

To make matters worse, now we have agents in the room. Humans have the innate ability to sense when a decision didn’t really land, AI agents do not. They act on whatever was written down, confirmed or not. Until recently, the cost of fuzzy, ambiguous coordination was a bit of friction: a few repeat meetings or a slow handoff. Now, poor coordination multiplies fast and can come at an incredible cost. Not least of which is massive token burn.

The cost of doing is collapsing. The cost of deciding is just getting started.

Coordination AI

This is the problem my co-founder and I are building to address, and it needed a name. We call it "Coordination AI”. In plain terms: it’s AI whose job isn’t to make more, but to keep true what the humans decided.

Almost every AI product you use today is built to remove friction. Coordination AI does something rather odd by comparison. It puts one piece of friction back on purpose. The type of good friction added earlier in the process so everything downstream runs smoother and encounters less bad friction.

The good kind of friction

After a meeting (i.e. the highest fidelity coordination moment there is), Afoot proposes what it heard: this was decided, you committed to that but didn’t set a deadline, this got parked for later, this issue needs input from X. Then it asks the people in the room to say: “yes”, “no”, or “not quite, it was this”. Nothing enters the shared record without human judgment.

That pause is the whole product. You cannot have a real commitment without someone accepting it with intention. If you take away that pause what you end up with is a meeting summary. And as we all know, meeting notes are usually one person’s version of reality, or worse, just better-formatted ambiguity. If you keep the reflective pause, you finally change the price of commitment.

With Afoot, commitments are now owned, visible, and explicit. Decisions stay closed and flow to the relevant areas of the organisation immediately (thanks to a team of coordination agents working behind the scenes), instead of waiting on a middle manager to call another status update meeting.

Finally, the hard conversations lead somewhere, so people are willing to have them.

Becoming someone. Becoming a team.

Kierkegaard wasn’t describing a productivity problem. He was describing how a person becomes real. The aesthete stays in possibility forever and never turns into anyone.

Teams are the same. A team that never closes anything isn’t a team yet. It’s a group of people keeping their options open near each other. The moment the team decides, and means it, and can point to a record of that decision months later, it becomes something.

For Kierkegaard, choosing was how you become someone. I think it’s how a team becomes one too. Coordination AI, and Afoot specifically, is just the infrastructure that makes the choosing count.

Afoot will soon be live. Join the waitlist for early access.

JOIN THE WAITLIST

In reality, we primarily interact with each other based on instinct. And that instinct is driven by our biological need to remain included in the group. The risk of social exclusion is what all of us are programmed to avoid. No one is immune to this.

It's for this reason that we offend each other without noticing, we defend our egos without noticing. We adhere to status and hierarchy without noticing. We uphold fake agreement without noticing, even when we know deep down inside disagreement would lead to better outcomes.

In previous eras, our group sizes were small and our pace of work slow, enough so that we could mitigate some of the damage caused by these instinctive threat responses and relational imperfections. Not anymore.

Companies got bigger, asynchronous communication tools that strip away nuance and non-verbal queues took over, execution sped up. Our tolerance for the normal friction of high quality group work began waning. Yet our dependence on the proper functioning of the collective mind did not.

Today, it often seems like we are anything but antifragile. We seem sick with abundance. Too fragile to tolerate contradictions, too weak to handle nuance, increasingly unable to think dialectically.

Or maybe that's too harsh. Maybe we didn't get worse. Maybe the conditions changed.

Dialectical thinking requires friction. You have to tolerate the discomfort of two conflicting things being true at once without rushing to eliminate one of them. Facilitators have a name for that period: “the groan zone”. The uncomfortable part of a discussion where the group hasn't landed yet and every instinct says "just pick something so we can get back to work."

Gaining wisdom requires sitting in that zone. Innovation and progress require it too. But people will only sit in that discomfort when they are confident that what comes out of it will actually lead somewhere. Nobody volunteers for a hard conversation whose outcome disappears when the conversation ends. When decisions don't stick and accountability is vague, avoiding the friction altogether isn't a failure, it’s just sensible math.

We've gotten very good at removing friction. I'm less convinced we've gotten better at living and working collectively. A little friction is the price of a vibrant, high performing team.

It's not surprising that most of the innovation in AI is geared towards optimising the individual. Solo work has no friction, so that's where AI went first. Making each person capable of executing what used to take a team of ten. On the surface, there is nothing wrong with this. Each of us getting more productive is a great thing, but not at the expense of the collective mind.

And here is what all that individual acceleration adds up to: the cost of doing collapses, but the cost of deciding skyrockets.

More output means more decisions, more handoffs, more parallel threads. And when more things run in parallel, coordination starts to break down. Agreements and commitments fade between meetings, ownership blurs and decisions get reopened again and again. Part of the load is real and necessary. Bigger, more complex systems genuinely need more coordination. But a lot of it is self-inflicted. I call this coordination entropy, and it scales with execution volume.

I may be naive, I may be overly sentimental, but to me, teamwork still makes the dream work.

So while AI is accelerating our individual productivity, I think its greatest potential is actually in how it can make us better teammates, better leaders, better human beings. It has the potential to improve the increasingly obvious bottleneck in making work work: coordination.

Because if you really step back and think about it, while our AI agents are out executing our demands, the only thing left for us to do with our time that will have any real value will be to sit in discourse with one another. To talk through trade-offs, take perspective, and co-create ideas that have never existed before. And as much as the self-proclaimed lone geniuses would like to believe otherwise, finding innovative solutions to complex problems is best handled as a group. Creativity is a social act.

Coordination has patterns. AI can read them.

AI can help turn a mediocre team with a fragile dynamic into a high-performing team with a robust and resilient way of working.

A high-quality productive conversation about how to move work forward has a structure. There is a science to it. Human coordination is a system that has predictable patterns, and spotting patterns across large messy datasets is what AI does best. Much better than armies of middle managers, at least.

I spent the better part of a year putting this hypothesis to the test. To see what would happen if I built an ontology based on the best practices in facilitation, group development and organisational design, plus a harness to run LLMs against it, and augmented it with my 15+ years' experience leading teams through uncertainty.

What I found out was pretty impressive. Not only could AI accurately predict potential coordination failures and recommend ways to improve, but the teams and leaders I tested it with were surprisingly receptive to this type of AI-mediated intervention. Which makes sense, when you remember where we started.

Coordination failures happen without us noticing. You can't fix what you can't see, and you can't see it from inside the moment. Watching the replay of how they coordinated changes that.

I saw first-hand how the teams behaviour changed in the flow of work. There was no need for leadership courses, no expensive coaching, no massive organizational restructuring. Only a replay of how the team actually coordinated and specific suggestions on how to improve.

Since those early proof-of-concept tests, I've been joined by an incredible technical co-founder - . A former whizz kid turned VC, back at his craft after a ten-year hiatus.

What a time to be alive!

Coordination infrastructure for high-performing teams

Together John and I are building Afoot: AI that turns human judgment into coordinated action. It processes meeting transcripts, encodes what the room actually decided, who owns what, and what's still open, and carries that context forward so everyone stays on the same page, people are held accountable, and healthy disagreement becomes easier.

How does it do that? With its own team of coordination agents, each one with its own narrow role.

1. The Extraction agent listens to meetings and writes down what happened as one of five possible coordination objects: decisions, alignments, commitments, escalations, and deferrals (i.e. things to revisit later).

2. The Connection agent looks at every new meeting against everything that came before it. So a decision from three months ago and a decision from today get recognised as the same thread. It also maps new coordination objects to existing initiatives, making it easier for everyone to see how work is being coordinated within a given frame.

3. The Prioritisation agent collects everything scattered across meetings (including those you didn’t attend) that actually needs your attention: things to confirm, commitments coming due, issues bumped back for a conversation. It pulls them into one ordered pile and keeps it current. It also keeps track of what others owe you, and tells you when a commitment you were counting on got fulfilled, withdrawn or blocked, so you can adjust accordingly.

4. The Follow-up agent stands back and reads the whole coordination ledger. Which commitments are blocking others? What’s overdue? Then it follows up with the right people for updates and drafts a suggested agenda for a meeting or discussion. This is so issues potentially blocking coordination can be addressed immediately instead of accumulating complexity.

5. The Team Health agent keeps an eye on the softer side of how the team works together. It catches unhealthy team dynamics while they’re still creeping in. Like disagreements that never get resolved, increasingly levels of escalated topics, or open work piling up on one person. When it spots a pattern like that emerging, it flags the change to the leader.

What does it take to get started with Afoot

Afoot is designed based on two fundamental principles:

1. People interact with each other intuitively, and will continue to do so until the end of time. We can’t achieve group flow when distracted trying to adhere to some rigid decision making rubric or feedback framework. Let people be people. Collaboration got us this far. The only thing that has changed is the pace and complexity of the challenges we face, and the pull of the devices making it harder for us to focus on the conversation in front of us. Humans aren’t the problem. The lack of infrastructure is.

2. People are at the limit of their cognitive bandwidth. If Afoot is going to help unlock the power of the collective mind, it cannot require a major learning curve. Record the meeting. Then, when the pile of coordination objects arrives shortly after, either accept, edit, or reject. That's the whole job. If the meeting itself was of any value, a few minutes of reflection afterwards is a small price to pay. The alternative is having the exact same conversation again and again because everyone walked out of the room with a different interpretation of what was discussed.

That’s all Afoot requires of its users: hold higher quality conversations. Record them. Confirm what you understood afterwards. And trust Afoot to do the rest.

How Afoot makes AI agents flow

Since every coordination object, extracted from live dialogue, is confirmed by the people in the room, every agent downstream can now act on that reality. That's what makes it finally safe to let agents move fast with greater autonomy. Because every significant decision an agent takes is traceable back to one or many humans who agreed to be held accountable.

When agents can do more autonomously, this leaves the people to have the discussions, disagreements and debates earlier and more often.

The normal friction of high quality group work stays, and that is the whole point. Afoot simply makes sure something useful comes out of it. The “groan zone” becomes worth it again.

We’ve all been in a discussion where it clicked. Where you left the room thinking: that was worth the time. You know it's possible because you've felt it. It's just so rare it feels like luck.

It doesn't have to be luck. As AI agents will take on more of the production work, what’s left for us is the human part: talking, disagreeing, changing our minds, figuring things out together. Those moments have to be worth showing up for.

We don't need AI to remove the friction of being human. We need it to help us make better use of it.

Build the infrastructure with us

Afoot goes live soon and the waitlist is growing. What we’re looking for now is the right design partners: a few companies that feel the pull towards working this way and want the coordination infrastructure built around how they operate.

If that’s you, email me at vanessa@afoot.net, or join the waitlist to follow our journey.

Join the Waitlist

Real leaders focus on generativity, not legacy.

Generativity is the work of empowering others. Legacy is simply what occurs when the things built by that generativity continue to function after the person is gone. The goal of generativity is to make oneself unnecessary, not to remain meaningful. It requires the creator to relinquish control, to let successors decide what it means, and to let the next generation change or overwrite it to suit new realities.

By setting up the scholarship in our great grandfather’s name, my sister was acting as a steward of his legacy. She wanted to ensure that he was remembered. She believed she would have decades of life ahead to witness its impact from the sidelines. For her, it was a forward-looking investment in the next generation of law students. A profession she believed in, and was absolutely exceptional at.

And therein lies the paradox; you cannot truly build your own living legacy. You can only act generatively, and entrust the actual building of legacy to the people who survive you.

Your legacy is built by those who are touched by your generativity.

What generative leaders do

The generative leader does not seek to dictate what the next generation of leaders will do, they aim to ensure their successors have the resources and infrastructure to figure it out themselves.

Here are a few key characteristics of a generative leader. Some traits to help you spot them in the wild.

1. They don’t push rigid rules and process. Instead they build a platform for participatory decision making. They want their successors to be able to tackle systemic challenges without needing a single visionary at the top.

2. They normalize dissent. They deliberately build a culture where staying silent feels uncomfortable. This allows the remaining team to evaluate new information objectively rather than defending the old regime’s point of view.

3. They view the organization as a complex, evolutionary system. They know trying to control the future from the top down, or from beyond their tenure, is a losing battle against complexity. Instead they build resilient teams where authority is pushed to the edges, and ensure the team has the autonomy and the coordination infrastructure to thrive long after they move on.

The impact of generative leadership

Since her passing three years ago, the stewards of her scholarship, which include the amazing team at Strigberger Brown Armstrong (SBA), our family and her husband, have continued to steward it successfully. And it is thanks to her generative leadership style that we do so every year with great joy. She didn't leave behind a rigid rubric. We each argue for our own picks, with respect and openness. We coordinate as a steering committee with little friction. And the final decision goes to her firm.

We look forward to it every year. Reviewing the candidates is a chance to be generative ourselves. It's also a reason to come together and think about her, and about who she was as a lawyer, a sister, a daughter, a wife and a friend.

When she passed away in 2023, her firm worked with the CCAWR to establish a second scholarship, this one under her own name. She never asked for it and she never would have. It exists because of what she did for the people around her while she still could.

That’s the whole blueprint. And it's an oblique one. You don’t build your own legacy. You can’t. You act generatively, and you trust the people you empowered to decide what it meant.

The Caroline Meyer Riley Memorial Scholarship

The Caroline Meyer Riley Memorial Scholarship is named in honour of Caroline Meyer Riley, who was the great granddaughter of a very distinguished Haitian lawyer Delabarre Pierre-Louis. Caroline in her own right was an exceptional and tenacious lawyer. She was one of the founding partners at Strigberger Brown Armstrong (SBA) and played an instrumental role in shaping the firm’s success. This CCAWR scholarship may be awarded to an individual who demonstrates strong community service and has an interest in law, civil rights or social justice issues. The scholarship award is valued up to $2,000.

Learn more and donate

In my 20 years in leadership, I’ve come to realize that our composure under this threat response is extremely fragile. Becoming anti-fragile in this domain takes tons of uncomfortable practice. Something many of us, sheltered behind our keyboards, aren’t willing to do. It is, in essence, the modern day equivalent of standing up to lions.

It may surprise you how little it takes for people to get dysregulated from a perceived threat. For them to stay quiet in a meeting room in order to preserve the peace, to say they agree with a decision when they really don’t, or to quickly become defensive when challenged on their views. It can be triggered by a subtle glance, tone of voice, or accidental exclusion from an email thread. It happens to all of us, all of the time. No human being is immune to this.

These responses don’t occur in isolation. We can be primed to be more sensitive to these social cues depending on what is going on in our lives outside the office, as well as by how our emotional needs were met as children.

This isn’t a bug. This is a feature of our social brains and it serves a very useful function. Without it we would all be out to serve our own interests and desires, and be completely unconcerned with the consequences our actions have on others.

But having this feature does not mean it should dominate how we behave in every situation. In the modern workplace, a dissenting opinion is rarely a threat to our survival. It is possible to modulate our response to perceived threat, to consciously choose not to identify with the emotions coursing through our veins. By doing so, we can remain open and connected, and capable of contributing our absolute best.

Because the activation triggers are very sensitive, it is up to the leader to deliberately design the conditions for productive dissent. The leader needs to be fully aware that this feature is enabled by default in every member of the team, including themselves, and that in order to bring out the best possible performance of the team, healthy dissent has to be cultivated carefully, strategically, methodically.

Designing the conditions

Dissent is a gift, and like coaxing a seedling from the earth, a leader can use a few practices to help it emerge. But these practices must be applied in the right order to be effective.

To bring out productive dissent in a team, a helpful model to apply is the PCP Model. The PCP Model (perspective, context, permission) is a behavioural influence framework popularized by Chase Hughes, a behavioral profiling specialist with a background in military operations and specialized tactical psychology. It explains how human decisions unfold sequentially. First you change people’s perspective. The change in perspective shifts the context. When the context change has been internalized, it gives people permission to take action.

Many leaders fail at creating the conditions for healthy dissent because they jump straight to step three of the model, often pleading: “please speak up and voice your honest opinions”. While well-intentioned, this is one of the subtle psychological triggers that activate our threat responses. It makes people anxious. And in this anxious state, the natural human reaction is to fawn, resulting in false harmony and fake agreement.

Instead, the first step is to methodically dismantle the fear of social isolation and retaliation.

1. Shape perspective

It’s often said that a great leader is a “repeater-in-chief”. And there is a good reason for this. People naturally suffer from cognitive drift. Under stress or routine, our perception of reality defaults back to personal survival and siloed thinking. We do this to preserve energy. By constantly repeating “why we’re here” and “what we are trying to do”, a leader anchors the team’s perception of reality, aligning individual actions into a cohesive system rather than a series of disconnected tasks.

Beyond broadcasting the strategic fundamentals, leaders must also actively dedicate space to uncover each member’s current perspective and view of reality. This not only makes it easier to synthesize perspectives into a shared worldview, it also builds trust by making people feel genuinely seen and heard.

2. Alter context

Once the team’s perspective is aligned, the unspoken rules of the environment have to be re-written. Perspective dictates what things mean, but context dictates what behaviours are normal and expected. By deliberately altering the context, the leader can remove the social friction of speaking up. This means the path of least resistance and energy conservation is no longer to stay silent, but to voice a critique. Silence should feel unnatural.

There a three key ways to do this:

• Run a pre-mortem: instead of open-ended questions, force the entire room into a failed future: “assume this project has completely failed in six months. Write down exactly what killed it.”

• Dismantle the hierarchy: the mere presence of a leader in the room carries psychological weight. You can alter this context by passing the facilitation role to a junior colleague or by requiring everyone to submit critiques anonymously via sticky notes. And most importantly, the leader must always speak last. When you speak first, you inadvertently anchor the room; when you speak last, you allow alternative perspectives to emerge unfettered.

• Assign the role: formally appoint a “devil’s advocate” or “red team” for the meeting and rotate this duty regularly. When dissent is a formal job assignment, interpersonal friction dissipates. The team member is no longer attacking a colleague’s idea; they’re simply executing their systemic role.
  

3. Grant permission

The final step is to give the team a green light to dissent. Many leaders mistake this as a passive policy. But permission is an active extraction. Since people are hard-wired to preserve peace and seek consensus, the leader has to use their authority to explicitly mandate the dissent the meeting was designed for.

There are a few key ways to do this:

• Replace passive prompts with direct commands. Instead of “what are your thoughts?” say: “let’s come up with three ways this will blow up in our faces” or “let’s find the flaws in this data”.

• Reward dissent: the exact moment the first person voices a dissenting opinion, the entire room will look at the leader for their reaction. This is the moment to give positive praise and acknowledgement for contributing their contrarian point of view. By rewarding the pioneer with praise, you give permission to the rest of the group to follow suit with zero risk.

• Refuse quick and easy agreement: if the team aligns too quickly with no one offering a challenge or alternative path, reject their decision. Force them to articulate at least one fully formed alternative path before making a final decision.

By systematically pulling the levers of perspective, context and permission, a leader can transform dissent from a scary professional risk into a celebrated systemic requirement. You stop having to beg people to speak up and instead build a structure that naturally extracts your team’s best thinking.

Refusing quick and easy agreement is the hardest practice on the list, because it means slowing the room down at the exact moment everyone wants to converge. Most leaders won’t do this consistently and most teams won’t ask them to.

Afoot makes that moment structural. It captures what a team decided and reflects it back for confirmation, which turns agreement into something a team has to actually give rather than something the meeting assumes.

Join the waitlist

Then add to the mix a brilliant sister turned litigator, and a dose of ADHD among all of us and you had a recipe for a household that never backed down from a fight.

Needless to say, I got comfortable taking a punch (the real and metaphorical kind), arguing my position and, most importantly, calming myself down and seeking proper reconciliation.

During these years my sister and I developed a good practice of writing each other apology letters after some of the most intense arguments. It was our way of expressing ourselves calmly and letting bygones be bygones. After reading each other’s letters we would never speak of the disagreement again and return to our typical sisterly love.

Since my sister passed three years ago and my mother has chilled TF out, I’m able to reflect on those years and appreciate that it was in the constant rupture and repair that we grew incredible close as a trio. Feelings were definitely hurt, and some arguments went on for longer than they should have (mostly because we got older and let our egos get in the way), but those moments were fundamental to the strong bonds we developed over the precious 40 years we all got to be on this earth together.

Those early years of conflict management training shaped how I developed as a leader, and they’re often the lens through which I evaluate the leadership skills of others. Not just the fighting part, either. Knowing how to rupture is only half of it. Knowing how to repair, the way my sister and I did with those letters, is the other half, and its typically not something we get coached on after childhood.

And while I’m still learning how to lead effectively in life and work, and always will be, after 20 years in professional leadership roles (and many more in volunteer roles), I can safely say that knowing how to instigate, facilitate and participate in productive conflict is what sets great leaders apart from the rest. And what I would say separates a leader from a manager.

There is no book you can read to learn this skill Prya Parker’s new book “The art of Fighting” is a good start) and reading this post will likely give you little to no value apart from awareness. Despite this, I still think it’s worth saying. And I want to say it because from what I can tell this skill is in decline and beginning to atrophy.

If we can’t figure out how to have an intense exchange without trigger warnings, overly identifying with our emotions and letting them get the best of us, throwing a tantrum and being vindictive, we will struggle to live up to our potential as individuals, teams and as a broader society.

Among the teams I’ve worked with and the founders I advise, many of them can barely handle the discomfort of a mildly heated exchange or awkward conversation. I see them regularly avoiding the necessary confrontation they need to get to the optimal solution. Instead they dance around the issue for months hoping things will naturally work themselves out. It never does. Either fight it out or fizzle out. But I promise you, the latter is far worse for your relationships and your business.

Why we’ve lost the art of fighting

The primary issues I think many people have with this are: nervous system regulation, inflexible egos and an inability to disassociate from their emotions in the moment.

If you can’t roll with the punches without getting defensive, cruel or vindictive, being an effective leader and building a high performing team will be hard, if not impossible.

My advice on how to train this skill: start picking some fights. But pick them on purpose. Not every disagreement is worth having, the ones worth picking are the ones where you’re genuinely unsure of the answer, or where staying quiet is already costing you something.

Push a few buttons on the more contentious issues you are curious about. Play devils advocate without telling the other parties that is what you are doing. Sit in the discomfort of being judged and ridiculed for your opinion. Witness what it looks like when the others in the argument can’t keep calm while you stay zen as a bird.

Like a workout regime, try to do these exercises at least once every few weeks: pick a fight, disassociate from your emotions, stay curious and open to the other person’s perspective despite the cruel way they may be delivering their message, and use language that builds bridges and demonstrates understanding as opposed to defensiveness and disconnection.

Once you can master your nervous system and stay as cool in these heated exchanges, then you can start facilitating productive conflict amongst a group.

Start practicing this yourself as a leader. Find some low stakes arguments to take up (ideally outside the family) and see what you learn about others and yourself in the process.

It’s all about the ruptire and repair

My mom and my sister taught me the rupture half of this a long time ago. It took losing my sister to really understand the repair half, and to see how much of my leadership is still built on those letters we used to write each other.

That’s the whole thing, really. Not whether you can avoid the fight. Whether you know how to have one and still know how to come back from it.

In every team, people are constantly weighing whether to speak up. Most of the time they choose silence, because silence is cheaper. No social risk, no awkwardness, no chance of getting shut down.

That's why the leader's response to disagreement matters so much. When someone pushes back, the leader has a choice. Get curious: "Tell me more." Or get defensive and protect the original call. That moment teaches everyone in the room what's really safe.

Over time, patterns form. Curiosity invites contribution. Defensiveness trains people to self-censor. And slowly, the room gets quieter.

Call that what it is: avoidance. Real alignment comes from having enough trust to face disagreement head-on. That's the human side of coordination, and no tool can solve it.

Tools can make the moment visible, though. A clear statement of what's being decided, made clear and shared back while it's still fresh, gives quiet doubts a chance to surface without putting anyone on the spot.

That's what I'm building with Afoot: structured pauses where dissent has room to breathe before action moves forward.

The future of leadership won't go to whoever runs the smoothest meetings. It belongs to the people who build teams that would rather be honest than pretend to agree.

The humans and the agents depend on it.

Every organization has a theory of how to fix coordination problems.

Most of them are wrong.

The default theory goes like this: things are getting complicated, people are misaligned, decisions aren’t being made. The fix? Add someone. Hire a project manager. Create a coordination role. Put a person in the middle to keep things moving.

It feels logical. It rarely works. And the reason it doesn’t work is mathematical before it’s even managerial.

Adding a person doesn’t add one connection. It adds many.

In any network, the number of potential communication channels is:

n(n–1) / 2

Five people: 10 channels. Ten people: 45. Add one “coordinator” to a team of ten and you don’t just add one link. You add ten new ones. The new manager doesn’t solve the coordination problem. They become a new surface area for it.

If the underlying system doesn’t change, the new hire spends most of their time managing the relationships their own presence created.

This is why adding management to a broken coordination system so often makes things slower, not faster. You haven’t fixed anything. You’ve made the problem more expensive.

Every layer you add is a lossy translation

Even when a manager does exactly what they’re supposed to do (synthesize information, relay priorities, keep people aligned) something goes wrong in the transfer.

A manager takes ten signals from their team and condenses them for their boss. In that condensation, the specific, awkward, important details get smoothed into something that sounds more like progress than it is. The edge cases disappear. The real blockers soften. The thing that actually needs a decision gets buried under the summary.

By the time the CEO’s strategic intent has passed through three layers of management to reach the person executing it, it has mutated into a set of KPIs that nobody finds particularly meaningful.

The distortion is structural. Every human router introduces noise. The signal degrades with every hop.

High-performing organizations don’t solve this by finding better translators. They build infrastructure that carries intent without distorting it, where the original meaning survives the trip.

The deeper problem is what layers do to the people inside them

There’s a more insidious effect that doesn’t show up in org charts.

When you add a layer of management between a person and their work, you’re sending a signal, even if no one intended to. The signal is: we don’t trust you to coordinate your own judgment.

It may be a nice manager. It may be a brilliant manager. But their presence as a gatekeeper changes how people behave. Decisions start requiring sign-off. Ideas need to be “socialized.” Ownership becomes blurry because technically the manager holds the thread.

Slowly, the people underneath stop practicing the skill of leadership. They stop asking “what needs to happen here?” They start asking “what do you want me to do?”

My friend Graziella has been watching this play out in her coaching work right now. Her observation:

burnout in these environments isn’t about workload. It’s about futility. Bandura’s research on self-efficacy backs this up. The behavior doesn’t change because people are weak, it changes because the system has repeatedly signaled that their actions don’t matter. Eventually people stop proposing. That’s not a character flaw. It’s a rational adaptation to a broken environment.

High-performing teams get built by making accountability visible and inescapable, where people can see the work, own the work, and be measured by it. That is an infrastructure problem, not a headcount problem.

The economic case is just as blunt

Managers are usually among the most expensive people in an organization. And every hour a high-output person spends updating a manager is an hour they’re not doing the work they were hired to do.

In a mediocre organization, this is treated as the cost of doing business.

In an ambitious one, it’s treated as a system failure.

The question to ask is: why pay a permanent tax on something that could be engineered away?

Complexity doesn’t get managed away. It gets engineered away. A management layer is a permanent solution to what was originally a temporary lack of clarity. Once it exists, it defends its own existence through more meetings, more check-ins, more process.

The people who’ve already figured this out

There’s a specific type of professional who has lived through this and come out the other side with a particular kind of impatience.

They’ve been in organizations where their best work got diluted by process. Where decisions they made got relitigated three levels up. Where a meeting was called to “socialize” something everyone already knew. Where “who owns this?” was a question that never got a clean answer.

They care too much to accept a system that buries signal in bureaucracy.

What they want is cleaner rules. Visible commitments. Clear ownership. A system that shows who said what, who owns what, and what happens next, so they don’t have to fight for clarity at every turn.

They’re not asking for less accountability. They want more of it, for everyone.

They show up in a few recognizable forms:

• The person who has burned out in a traditional organization and is looking for somewhere their work can actually land.

• The systems-thinker who looks at a tangled Slack channel and wants to redesign the whole thing.

• The craft-obsessed executor who wants to spend their time doing the work, not reporting on it.

• The person who doesn’t wait for a title to take ownership and just sees the gap and fills it.

What all of them share: a high tolerance for truth and a low tolerance for friction.

In most organizations, these people get called “difficult”. They’re the ones asking “wait, did we actually decide that?” and “who is the primary owner here?” They get flagged as impatient, as too demanding, as hard to manage.

They are, in fact, the most valuable people in any organization. They just need a system that matches how they think.

Clarity is a technical requirement

The tension here isn’t really between management and autonomy. It’s between two modes that every organization needs: managing for predictability (decisions, governance, execution) and leading for complexity (shared sense-making, collective judgment, experimentation).

Most organizations only know how to do the first. And under pressure, they double down on it. More layers, tighter control, more reporting. It feels decisive. It looks like leadership. What it actually does is collapse the right side of the loop entirely, the part that produces agility, breakthrough thinking, and resilience in shifting markets.

The infrastructure argument isn’t about removing accountability. It’s about building the operational layer tightly enough that the human layer can actually function. When decisions, ownership, and next steps are clear and visible, people don’t need a manager to hold the thread. They can see the map and navigate themselves.

That’s what separates the organizations that move fast under uncertainty from the ones that freeze. Not fewer rules. Better infrastructure. A system tight enough that the humans inside it can lead, not just execute.

The organizations that will move fastest in the next decade will build coordination into the infrastructure itself, so that intent travels cleanly, decisions stick, ownership is unambiguous, and the people doing the work can see the full picture without asking for a briefing.

That is an engineering problem. Always has been.

And the people who’ve always known this are already looking for each other.

Growth Wise is building the infrastructure layer for teams that are done paying the coordination tax. It takes human coordination, qualifies its structural integrity, and compiles it into durable organizational reality. PM tools track execution. Growth Wise compiles coordination into executable reality.

A month ago, Jack Dorsey and Sequoia published a provocation: what if AI could replace middle management entirely?

The idea centres on what they call a "World Model." A centralised system that ingests every Slack message, document, meeting transcript, and KPI to create a continuously updated digital twin of the organisation. AI uses that picture to infer priorities, surface insights, and guide action. With comprehensive, real-time context available to everyone, the information asymmetry that managers exist to solve would disappear. People closest to the work would know what is happening and what should happen next, without waiting for direction to travel down a chain of command.

The argument contains a real insight. Middle management, as most organisations practice it, is an expensive patch for a bandwidth problem.

But solving the information problem does not solve the coordination problem. Context without commitment is just a well-informed group of people drifting in different directions. The hardest coordination problems are commitment problems.

And commitment is a social act, not a computational one.

Context without commitment is noise

Organisations don’t fail because they don’t know what to do, they often fail because no one is clearly accountable for doing it.

You can give every person in the organisation the entire company’s data, strategy, and real-time signals. It doesn’t matter. Nothing becomes real until someone owns what happens next.

What people agreed to matters more than what the algorithm predicted.

We are pilots, not sensors

At enterprise scale, the World Model makes centralised sense-making the default. The system models the world; humans feed it data from the edges. The AI does the high-status work: sense-making, prioritisation, strategy. Humans execute and refine its outputs. The human gets to pick from a shortlist the machine already curated.

That is not leadership. That is editing.

We reject that inversion. The human is the sovereign pilot, the sovereign creator. By “sovereign,” we mean the one who holds judgment and is accountable for what happens next.

The AI does not route the human; the human uses the AI to structure their intent. AI can and should enhance local judgment, surface patterns humans miss, and accelerate the thinking that precedes every commitment. The question is who remains sovereign in the loop.

Where intelligence lives

The World Model centralises sense-making. One system ingests everything, computes priorities, and distributes direction to the edges. The people closest to the work receive guidance from a system that does not share the consequences of their decisions.

We believe judgment becomes meaningful where the work happens: in the negotiation between people at the point of contact. The system’s job is not to decide. It is to make what was decided visible, durable, and shared.

Complicated -vs- complex

The World Model is strongest on complicated work. It struggles where work becomes genuinely complex.

Complicated work has right answers. It can be decomposed, analysed, and optimised. Complex work has competing priorities, ambiguous signals, and tradeoffs that only make sense in hindsight.

You cannot optimise your way through complexity, you have to coordinate your way through it.

AI will increasingly handle the complicated. The work that remains for humans is the complex work: navigating ambiguity, making tradeoffs, committing under uncertainty.

This is where the “codifiers’s curse” hurts. The World Model improves AI agent performance on complicated tasks. Humans verify the output. The better the agent gets, the less the human matters. Then suddenly you are a quality check on a machine that is getting better at the thing you’re checking. That is a shrinking role.

The opposite trajectory is possible. As AI takes over the complicated, humans are freed for the work that actually requires them: coordinating with each other, through the uncertainty, to decide what matters and what happens next.

Why you can’t skip the human part

Humans aren’t pre-wired for complex coordination. We figure it out by working with the same people repeatedly until patterns form and shared understanding develops. That shared understanding is the real infrastructure of trust. It’s what lets a team of ten move faster than a team of a hundred.

The World Model assumes you can replace that with a database and a context window. But context is not trust.

When a system assigns priorities without shared understanding, it feels like a suggestion from a stranger. People treat it that way. Without the moment where people align, decide, and take ownership, the work has no weight. It doesn’t carry forward.

The best systems help you make sense of the situation and make sure decisions are clearly made, owned, and followed through. Neither works without the other.

Creators, not editors

We don’t believe a small group of people are creators (I.e. the thinking class). We believe most people are capable of meaningful judgment. But the systems we’ve built often reduce them to editors. AI will amplify that pattern. It will either concentrate judgment further, or make it easier to distribute it.

People don’t lack judgment, most systems just don’t ask for it. A system that treats humans as verification layers will get people who behave like verification layers.

The answer is not to suppress judgment. It is to structure it.

The choice

AI is forcing an organisational design choice that most haven’t recognised yet. There are two default paths.

• Path one: reinforced hierarchy. AI increases visibility and control. More dashboards, more oversight. Humans become more legible to power.

• Path two: centralised World Model. AI becomes the sense-making centre. Humans become inputs and execution nodes. The system decides. Humans approve.

Both paths move judgment away from the people closest to the work. And if you don’t choose, the system will default to centralisation. That is the real risk. Organisational design can quietly demote human judgment below the algoritm, while pretending to augment it.

To preserve the role of human judgment where it actually matters, we need a different path, a third path.

Not more middle management or centralised intelligence, but coordination as infrastructure. A system where decisions are made explicit, ownership is clear, and commitments are durable. Where what gets decided in rooms actually carries forward into the work.

Because the better we coordinate, the more independently we can act. Shared understanding and record of people’s commitments and decisions gives people the confidence to exercise judgment without constantly returning to the centre for direction. That is what makes distributed agency possible.

AI expands the space of possibilities. But it does nothing to ensure anything actually becomes real. That part still requires humans. The question is whether we build systems that honour that fact or not.

Most systems optimise for knowing what to do, but the harder problem is making sure it actually happens.

As AI becomes embedded in how organisations operate, the real question is not how much work machines can take on, but where human judgment belongs.

Ultimately, the advantage will go to organisations that know how to balance automation with human judgment. Knowing what to optimise and what must remain shaped by human judgment, imagination, and people thinking together.

There is a third path.

Afoot is built for organisations that have chosen it. Where human judgment is structured, not suppressed, and where what gets decided together actually survives into execution.

Afoot goes live soon and the waitlist is growing. Right now, we’re looking for a few design partners: companies where people are already exploring, where good ideas keep getting lost between the meeting and the plan, and that want the coordination infrastructure built around how they actually work.

If that’s you, email me at vanessa@afoot.net, or join the waitlist to follow our journey.

Join the Waitlist

Jack Dorsey and published something worth reading last month. “From Hierarchy to Intelligence” is Block’s public statement of intent to restructure itself as an AI-coordinated company, replacing layers of management with a shared world model that gives every person in the organization the context they need to act.

at Superesque wrote a sharp breakdown of what Block is actually building and why it matters for founder-led teams. Read it. It’s one of the clearest accounts I’ve seen of where this is all going.

One thing he touches on deserves more attention.

The argument Dorsey and Botha are making (that AI should replace what management does: routing information, maintaining alignment, coordinating work) is the same argument I’ve been making from a different angle.

The coordination tax isn’t a people problem. It’s a system design problem. Every layer you add between a decision and the person executing it is a translation layer. Every translation layer introduces distortion. The signal degrades. The intent mutates. By the time strategy reaches execution, it has become something else.

Sangeet Paul Choudary calls this “coordination without consensus.” AI reduces the translation costs of turning one team’s outputs into another team’s inputs. You don’t need a manager in the middle to relay, interpret, and repackage. The infrastructure does it cleanly, at speed, without the loss.

Afoot is built to solve exactly this problem, at a different scale than Block but with the same underlying logic: coordination should be an infrastructure problem, not a headcount problem.

The more interesting question is what gets coordinated.

Block’s world model is ambitious. It aggregates millions of data points — transactions, decisions, code, designs, plans — into a continuously updated picture of what the organization knows and does. The AI intelligence layer reads that picture and acts on it.

But there’s a layer that picture doesn’t capture cleanly. The layer where humans are actually in a room together, negotiating reality.

The decisions that matter most don’t happen in task trackers or code repositories. They happen in conversations, where assumptions collide, where commitments get made, where the ambiguity that no document can fully resolve gets worked through in real time. That is the richest coordination moment any organization has. And it is also the most consistently under-captured.

Most organizations have already patched this with AI notetakers. The transcript exists. The summary gets generated. But a summary of what was said is not the same as a structured record of what was decided, who owns it, and what happens next. The notetaker captures the conversation. It doesn’t qualify the coordination. It doesn’t tell you whether a real commitment was made or whether people just seemed to agree. That distinction is where the signal still gets lost.

Ambiguity doesn’t announce itself. It just shows up later as a missed deadline, a relitigated decision, or a project that has to start over because nobody wrote down what was actually agreed. The cost compounds quietly.

That is the gap worth closing.

The convergence Dr. Tim Rayner identifies — between what Block is building at enterprise scale and what forward-thinking founder-led firms are building now — is real. The architecture is the same. Build a shared knowledge base. Give people access to it. Let AI do the coordination work that humans were never well suited for anyway.

The difference, as he rightly points out, is what role people play inside it.

At enterprise scale, the risk is that people become appendages. The system models the world; humans feed it data from the edges.

At the scale of a founder-led team, the opportunity is different. The knowledge base is comprehensible. The people inside it can actually inhabit it rather than just serve it. The AI amplifies what they know rather than routing around them.

There’s a more precise way to say what the bottleneck actually is, and where Afoot sits in this picture.

The goal isn’t primarily knowledge accumulation for improved automation. That framing, while part of it, misses where the real friction lives. The bottleneck has shifted upstream. Execution is getting faster every month. The bottleneck is now the moment before execution: the work people do together to create shared reality.

Coordination, properly defined, is that work. Understanding, decisions, commitments, ownership, priorities, handoffs. The moment where a group of people negotiates what is true, what matters, and who is responsible. That moment is where collective judgment actually occurs. It’s where sensemaking happens. It’s where the direction of work gets set.

And it is still, in most organizations, almost entirely unstructured.

Most world models start from the bottom up: aggregate everything, build the picture, hope coordination falls out of it. Afoot starts from the highest-fidelity moment. Synchronous dialogue sits at the top of any coordination hierarchy. Structure that layer reliably, and everything else (documents, plans, async updates, downstream systems) connects in as supporting context rather than load-bearing foundation. The constraint is the architecture.

AI notetakers capture what was said. Project tools track what gets done. Neither touches the quality of what happened in between: whether a real decision was made, whether the commitment is durable, whether the people in the room actually reached shared understanding or just moved on.

Afoot is built for that layer. The goal is to recognize where judgment happens, give it structure, and make the output of collective sensemaking durable enough to survive into execution.

The encoded knowledge matters. The more important thing is that the moment of coordination, where humans decide together, produces something real that the rest of the system can act from.

Better conversations. Cleaner commitments. A system that lets both humans and machines act from the same structured understanding.

That is the coordination architecture worth building toward.

Afoot exists to help ambitious organizations innovate and execute at a high standard by ensuring that ideas, decisions, and commitments survive into reality.

For a long time, a strange thing has sat at the center of modern work. We’ve treated it as normal.

A group of people gets into a room. They talk through priorities, tradeoffs, risks, decisions, dependencies, tensions, and next steps. They negotiate reality together in real time. They do the messy human work of deciding what matters.

And then the meeting ends.

After that, one person is expected to go translate whatever just happened into something the organization can use. They write the notes. They summarize the decisions. They assign the actions. They clean up the ambiguity. They interpret what people meant. They decide what to leave out. They update the project management tool. They turn a living conversation into a static record.

That person is often a manager.

And we’ve been using humans to do a job that is both too mechanical and too lossy for humans to do well at scale.

Meetings are where organizational reality gets negotiated

The interesting AI shift isn’t that it will replace all thinking. It’s that it will make us more aware of which kinds of work humans should still be doing.

One answer is clear.

Humans should spend more time doing what only humans can do well together: making tradeoffs, negotiating meaning, aligning around uncertainty, resolving tension, committing to action, deciding what matters.

That work happens most powerfully in synchronous conversation.

Most meetings are terrible. But when people are in the same room, physical or digital, and have to respond to each other in real time, something happens that no document or workflow tool can replicate.

Ambiguity gets surfaced. Assumptions collide. People challenge each other. Misunderstandings become visible. A decision can actually become shared.

That’s a higher-fidelity coordination moment than a status update in a tool or a polished strategy memo. A strategy document matters. A plan matters. A procedure matters. But they are downstream artifacts. The conversation is where the reality changes.

The bizarre inefficiency at the center of work

Once you see that, the current workflow looks ridiculous.

We gather the expensive humans. We do the expensive human work. Then we ask one of those humans to do clerical translation after the fact.

That person becomes the bottleneck between human coordination and organizational action.

They are expected to remember: what was actually decided, what people only seemed to agree on, what was deferred, what was assigned, what was implied, what still felt unresolved, and which part of the discussion mattered enough to log.

That is not a neutral act. It is interpretive. It is filtered. It is biased. It is incomplete.

The notes are never the meeting. The project management update is never the conversation. The official recap is always a compressed, partial rendering of what actually happened.

And yet many organizations still rely on that rendering as if it were reality.

This is one reason so much work feels fuzzy. People aren’t lazy. Teams aren’t stupid. We are simply forcing human beings to manually re-encode shared reality after the richest moment of shared sense-making has already passed.

That is terrible system design.

AI changes the threshold of what feels absurd

In a world where AI can draft documents, generate prototypes, summarize research, analyze transcripts, and route information between systems, the manager-as-transcription-layer model starts to feel even more irrational.

You can feel the absurdity more sharply.

AI can write code, synthesize a market landscape, and generate ten versions of a board memo in seconds. But after a meeting, we still expect someone to manually type out who is doing what by when. We still expect a person to open a project management system and reconstruct the discussion from memory. We still expect middle managers to spend energy converting live coordination into admin artifacts.

That doesn’t just feel inefficient. It feels primitive.

The more capable AI becomes, the more intolerable this clerical burden will feel. And many organizations are still running managers as translation middleware — not leaders, not facilitators of judgment, just translation middleware.

That is the layer AI will absorb.

What AI should actually do

The right frame isn’t AI replacing human judgment. It’s this: humans should do more of the work only humans can do well, and machines should do more of the work humans should never have had to do in the first place.

After a real coordination moment, AI should be able to capture what was decided, identify what was agreed, spot what remained partial, register who owns what, and turn the output of the conversation into something that other humans and machines can actually work from.

That is not replacing the meeting. That is not replacing leadership. That is not replacing accountability.

It is removing the clerical translation layer that currently sits between conversation and execution.

Humans negotiate reality together. AI turns that negotiated reality into structured, usable downstream artifacts. That makes sense.

This is not about making people speak like machines

There is an important tension here.

You can’t build systems like this without some ontology. The machine needs structure. It needs categories. It needs a way to distinguish a real commitment from vague agreement. It needs to know the difference between a decision and a partial one. It needs some way of recognizing ownership, alignment, and next steps.

That can sound rigid.

But the point is the opposite of rigid. The machine should carry most of the framework burden so that humans can remain relatively natural.

People should not have to leave every meeting and fill out a ritualized template to prove they coordinated. They should not need to remember which workflow category applies to which discussion. They should not have to behave like clerks for the system.

What they do need, over time, is to become slightly more explicit about a few very basic coordination signals.

• What did we decide?

• What are we aligned on?

• Who owns this?

• What happens next?

• By when?

That is not bureaucracy. That is the minimum viable structure of coordinated action. When those things never become explicit, much of what organizations call coordination is just socially acceptable ambiguity.

Better AI should make better meetings more valuable

There’s a tempting fantasy that AI will reduce the need for conversation. I think the opposite happens.

As AI makes drafting, producing, analyzing, and executing easier, the real bottleneck shifts upstream. The scarce work becomes judgment, tradeoffs, clarity, alignment, priority, commitment.

The more execution accelerates, the more valuable high-quality human coordination becomes.

The meeting of the future should not be a place where people passively report status so that someone can later type it into a system. It should be a place where people do the highest-value work: sense-making, decision-making, alignment, conflict resolution, real commitment formation.

If AI can reliably capture the outputs of those conversations, people can spend less time on clerical aftercare and more time actually thinking together.

That is a more interesting future of work than “AI summarizes your meeting.”

The manager’s role does not disappear. It changes.

The mechanical parts of management become harder to justify as human labor.

The manager who mainly exists to collect updates, restate priorities, translate decisions into tasks, chase follow-up, and manually maintain the system of record — that role is standing on unstable ground.

The manager who can set direction, create clarity, hold tension, facilitate judgment, surface tradeoffs, protect focus, and help groups reach durable commitments — that role becomes more valuable.

Less transcription. More facilitation. Less clerical reconciliation. More structured human coordination.

AI doesn’t replace management. It strips management down to the parts that are actually human.

Why this matters beyond productivity

Every time a meeting ends and one person manually reconstructs what happened, the organization loses signal.

It loses nuance. It loses dissent. It loses uncertainty. It loses the exact form of the commitment. It loses the difference between what was said and what was assumed. It loses the shape of the conversation that produced the outcome.

That loss compounds. Then we wonder why teams revisit decisions, why ownership feels muddy, why strategy doesn’t translate into execution, why meetings multiply, why coordination feels exhausting.

The problem isn’t that people failed to think. It’s that the outputs of thinking were never captured cleanly enough to travel.

My bet

My bet is that this old system will start to feel increasingly absurd.

We’ll look back and wonder why, after a room full of people did the hard work of making sense together, we handed the burden of translation to one person with a notes document and a project management tab.

We’ll wonder why we tolerated so much loss between conversation and action. We’ll wonder why managers spent so much time encoding reality instead of helping shape it.

The organizations that move first won’t be the ones that use AI to generate more internal sludge. They’ll be the ones that use AI to protect the highest-value human work and remove the clerical friction around it.

Better conversations, clearer commitments, cleaner transfer of reality between groups, and a system that lets both humans and machines act from the same structured understanding.

AI as the semantic layer between human coordination and machine execution.

That’s the future worth building toward.

Growth Wise is a Decision Reliability platform. It instruments the human coordination layer so organizations can see which decisions are closing, which delegations are landing, and where coordination is breaking down.

I spoke with a product ops leader who has run operations at three very different companies: a 6,000-person fashion retailer, an IPO-track automotive marketplace, and a 200-person SaaS company. Different industries, different tools, different team sizes, completely different cultures.

The coordination failures were identical at all three.

That’s the part worth sitting with. When the same ops discipline, applied by the same person across three radically different organizations, keeps finding the same gaps, the gap is systemic. Coordination debt accumulates from the product development model itself, not from the particular people or tools involved.

The first thing that always breaks

At every company, the first fix was the same: making sure people understood what they were expected to do within the development cycle. Not which tools to use. Not which ceremonies to attend. Whether they knew their role in the sequence, when their input was needed, what “done” looked like at their stage.

Everything else varied. That stayed constant.

Process compliance is not coordination

At the 200-person SaaS company, teams followed the process. PRDs were written on time. Stakeholders were tagged in Notion. Every box on the checklist was checked. The coordination was still broken.

PMs and engineering managers were tagging each other in 300 to 400 Notion notifications a day instead of sending a five-minute Slack message asking for fifteen minutes to walk through something. The tagging was process compliance. The conversation would have been coordination.

This is a pattern I see repeatedly. A decision can look closed on paper — the PRD was approved, the ticket was created, the stakeholder was notified — but if the people involved never had the conversation where they surfaced the assumption that will blow up in sprint two, the closure is cosmetic. The rework is already baked in.

Where cycles actually break

Delivery, the middle of the product development cycle, is typically the easiest part to fix. The tools are mature. Engineering teams know how to ship.

The hard parts sit at both ends.

At the beginning: choosing what to build. With AI now generating more customer data, user feedback, and market signals than any team can manually process, the curation problem has scaled. Product teams have always struggled with prioritization. The volume of input they need to filter has multiplied, and the frameworks for doing that filtering haven’t caught up.

At the end: closing the loop on whether what was built actually worked. He called it “set and forget” — teams ship a feature and immediately move to the next one. Nobody circles back to assess whether the thing they built solved the problem it was meant to solve. That missing feedback loop means beginning-of-cycle decisions get made against stale assumptions. The cycle compounds.

The handshake signal

At the 200-person company, he tracks one specific leading indicator: time to confirm scope.

The product-engineering handshake — the moment when product and engineering agree on scope for a planning cycle — should happen by week two. When it slips to week four, something upstream is already broken.

The diagnosis forks from there. If scope was unclear, that’s a product problem. If scope was clear but engineering didn’t engage until late, that’s an engineering engagement problem. Either way, the delivery miss is locked in by the time the handshake slips. Waiting for the sprint review to discover it means finding out six weeks too late.

DORA metrics will register the same problem later as extended lead time. The handshake signal catches it weeks earlier.

Proactive ops versus reactive ops

He described two versions of himself.

Reactive: he shows up after deadlines have been missed. He asks why. He traces the cause. He recommends a fix. Teams experience him as the auditor. He called himself “the bad cop.”

Proactive: he catches the signals before the miss. The handshake hasn’t happened by week two. A cross-functional dependency hasn’t been confirmed. A stakeholder hasn’t been looped in. He nudges the team before anything breaks. Teams experience him as the enabler.

The difference between those two modes comes down to one thing: whether he has visibility into coordination signals early enough to act on them. Most ops leaders don’t. They’re buried in the operational work itself, producing reports, running ceremonies, chasing status updates. The reactive mode prevents the proactive mode from ever starting.

What AI actually changed

He uses AI extensively, but not to automate coordination. He uses it to automate the grunt work that was consuming his signal-reading time. Reviewing PRDs for completeness. Translating Jira data into the financial language his CFO needs. Assessing whether a roadmap item maps to stated strategy.

With that time recovered, he spends more of his day reading the human signals that AI can’t parse: which PM hasn’t talked to their engineering counterpart this sprint, which team’s energy has shifted since the reorg, which handshake is three days late and why.

AI made him more of a people-reader, not less. The tool handles the artifacts. The ops leader handles the relationships.

The scrum master gap

The dedicated scrum master — the person who sits in every ceremony and tracks whether process is being followed — is a role in decline. He was blunt about it.

But the needs that role was created to serve are only growing. Tracking whether action items from a meeting actually get done. Making sure the agreement from Tuesday’s planning session survives contact with Thursday’s priorities. Facilitating the conversation between the PM and the tech lead who disagree on scope but haven’t said so directly.

Those functions matter more as organizations add cross-functional complexity. Right now they’re falling through the cracks. The scrum master title is going away. The coordination work it was supposed to do is piling up with nobody assigned to it.

The pattern across all of this is the same. The product development model, and specifically the coordination layer inside it, is the structural variable that determines whether a team ships well or ships late. Tools change. Team sizes change. Industries change. The coordination patterns repeat.

An ops leader who can see those patterns early stops being the auditor and becomes the person who makes the system work.

This is the gap Growth Wise instruments. Not the sprint velocity, not the delivery metrics — the coordination layer upstream of all of it. The signals that tell you a miss is coming before anyone has missed a deadline.

Anthropic published a research paper in March 2026 mapping AI’s theoretical capability against every occupation in the U.S. economy. The number for management occupations: 91.3%. That means large language models can theoretically handle over nine out of ten tasks that managers currently perform.

The instinctive response to that number is panic. Or dismissal. Both miss what’s actually happening.

The 91.3% figure describes management. It does not describe leadership. And the distinction between those two words, which most organizations use interchangeably, is about to become the most consequential sorting mechanism in corporate life.

Kotter drew the line in 1990

John Kotter’s framework is 36 years old and has never been more relevant. Management, he wrote, is about coping with complexity. It produces order. Planning, budgeting, organizing, staffing, controlling, problem-solving. These are systematic, repeatable activities that turn ambiguity into process.

Leadership is about coping with change. It produces movement. Setting direction, aligning people, motivating, inspiring. These are adaptive, relational activities that turn process into purpose.

Peter Drucker landed in the same place from a different angle. Management is doing things right. Leadership is doing the right things. In a knowledge economy, Drucker argued, the management function becomes almost irrelevant because any effective knowledge worker is exercising judgment, not following procedure.

Abraham Zaleznik pushed the distinction further. In a 1977 Harvard Business Review article that still gets assigned in business schools, he argued that managers and leaders are psychologically different types. Managers are “once-born,” people who followed a stable path, identify with the existing social order, and see themselves as conservators of the organization. They minimize risk and conflict. Leaders are “twice-born,” people shaped by some period of personal struggle or inner conflict that detached them from the status quo and gave them the ability to challenge it.

That psychological framing matters here because it predicts what happens when AI removes the mechanical work. The once-born manager, whose identity is wrapped up in maintaining order, loses the activity that defined the role. The twice-born leader, whose instinct is to challenge and redirect, gains time and headroom that the coordination overhead used to consume.

All three — Kotter, Drucker, Zaleznik — were writing before AI existed as a practical tool. They were making a conceptual argument. AI is turning it into an operational one.

What management actually consists of

Break the management function into its component tasks and the automation case becomes obvious.

• Planning and organizing. Setting objectives, creating schedules, allocating resources. A system that can read every meeting transcript, track every commitment, and measure whether decisions closed with enough structural specificity to execute has more planning visibility than any human manager. It doesn’t forget. It doesn’t get distracted by the loudest voice in the room. It measures whether the planning session produced durable outcomes or just the feeling of progress.

• Directing and coordinating. Assigning work, managing handoffs, tracking dependencies. This is the core of middle management: making sure the right people are doing the right things in the right sequence. A system that measures Delegation Flow (the probability that delegated work will actually execute, based on whether the delegation left the room with an owner, a next step, and a deadline) replaces the manager’s “checking in” function with a structural signal that’s available immediately after the meeting.

• Controlling and monitoring. Tracking progress, measuring performance, course-correcting. The status meeting exists because managers need to know what’s happening. A coordination observability system makes the status meeting redundant: it shows which decisions are holding, which are generating rework risk, which topics keep resurfacing as zombie items, and which escalations left without a carrier. The manager’s role as information aggregator gets replaced by a dashboard that’s more accurate and more timely than any human summary.

• Reporting and communication. Status updates, stakeholder alignment, upward reporting. Most of what managers report to their managers is a curated summary of coordination state. The system produces that summary automatically, with structural evidence instead of narrative spin.

Growth Wise, the decision reliability platform I’m building, already does most of this. It scans meeting transcripts for closure signals (was a decision explicitly stated, was an owner named, was a next step defined, was a deadline set). It grades every outcome as Achieved, Partial, or Absent. It tracks Delegation Flow, Rework Risk, Coordination Debt, and Decision Churn across every meeting. It tells you which meetings are producing real coordination and which are performative theater. These are management functions. And a system handles them better than a person because a system doesn’t have selective memory, doesn’t avoid uncomfortable truths, and doesn’t optimize for its own career survival when reporting upward.

What leadership actually requires

Now subtract all of that from the manager’s job description. What’s left?

Deciding what matters. Choosing direction when the data is ambiguous. Aligning people who disagree on where to go. Building trust in an environment where trust has eroded. Having the conversation that everyone is avoiding. Knowing when the process is wrong and having the courage to override it. Developing people by understanding what they need, not what they say they need. Holding the tension between urgency and sustainability.

None of these are systematic. None are repeatable in the procedural sense. All of them require judgment, empathy, relational skill, and the willingness to be wrong in public.

Kotter would call this the leadership function. Drucker would call this doing the right things. AI can’t do any of it. AI can tell you that Delegation Flow dropped from 80% to 38% across the last three planning sessions. It cannot tell you that the drop happened because two senior engineers lost trust in the product direction after the reorg, and the way to fix it is a 45-minute conversation over coffee where you listen more than you talk.

The thought experiment

Here’s the provocation. If a system can handle 91.3% of management tasks, and the remaining tasks are all leadership tasks, then the manager role as traditionally defined is a transitional form. It exists because organizations needed a human to do both the mechanical coordination work and the adaptive leadership work, and there was no way to separate them.

AI separates them.

The coordination work (planning, monitoring, directing, reporting) gets absorbed by systems that do it faster, more consistently, and without political distortion. The leadership work (direction-setting, alignment, trust, development) stays with humans, but now those humans are freed from the mechanical overhead that used to consume 80% of their week.

This changes who should be in the role. The current selection criteria for managers over-index on organizational skills, process discipline, and stakeholder management. These are management skills. The people who get promoted are often the people who are best at the mechanical function. The new selection criteria would over-index on judgment, relational intelligence, and the ability to create psychological safety. These are leadership skills. They’re rarer, harder to develop, and historically undervalued because they were bundled with the mechanical work that organizations actually needed to keep running.

What this means for Growth Wise specifically

A friend recently pointed out that Growth Wise, taken to its logical extension, becomes a mechanism for eliminating the management function entirely. He’s right, and I’ve been thinking about that observation since.

Growth Wise instruments the coordination layer. It makes visible whether decisions are closing, whether delegations are landing, whether the same unresolved topics are cycling back week after week. Every one of those visibility functions used to live inside a human manager’s head. The manager sat in the meetings, tracked the commitments in a notebook or a spreadsheet, followed up with people who hadn’t delivered, and reported the state of play to the next level up. Growth Wise does all of that, structurally, with evidence, without the cognitive load.

But Growth Wise doesn’t tell you whether you’re building the right product. It won’t surface the fact that your best engineer is disengaged because she feels her technical judgment has been overridden three times in a row. It can’t read the room and realize that the cross-team conflict isn’t about scope; it’s about two leaders who fundamentally disagree on company direction and are using the planning process as a proxy war. And it will never tap you on the shoulder and say: slow down, stop optimizing for velocity, and listen.

The tool absorbs the management function. The human who used to perform that function is now freed to do the leadership work that was always more important, was always harder, and was always getting squeezed out by the coordination overhead.

The apprenticeship question

There’s a legitimate objection to all of this. If AI absorbs the coordination work that junior managers used to cut their teeth on, where do future leaders learn? The traditional development path runs through the mechanical function: you track commitments, run standups, manage a small project, and gradually develop judgment by handling progressively larger coordination challenges. Remove the mechanical rung of the ladder and you risk what some researchers call “experience starvation” — a generation of senior people with no pipeline of developed successors behind them.

But this assumes the only way to develop leadership is through management. There’s another path, and organizations that figure it out early will have a serious advantage.

The highest-value leadership skill in a world where AI handles coordination mechanics is the ability to hold a high-quality conversation. Not present at one. Hold one.

Facilitation: the discipline of structuring collective sense-making so that the right people reach the right conclusions together. Setting the agenda so it surfaces the real tensions. Managing turn-taking so the quiet expert gets heard before the loud executive closes the discussion. Keeping the group on the question that matters instead of the question that’s comfortable. Knowing when the room needs five more minutes of silence instead of another slide.

Junior leaders can learn this by being in the room during the moments that matter — not as note-takers, but as process holders. A senior leadership team working through a product direction disagreement or a reorg decision needs someone holding the conversational structure while they concentrate on the substance. That’s a real job, a hard job, and a training ground for exactly the skills that will define leadership once the coordination layer is automated.

The AI system tracks whether decisions closed. The junior facilitator ensures the conversation had the structural conditions to produce real closure in the first place. The senior leaders focus entirely on the dialogue. Three layers, each developing the next one upward.

The uncomfortable implication

If management is a function and leadership is a skill, and AI absorbs the function, then the uncomfortable question is: how many current managers are actually leaders?

Not many. Most organizations have promoted people into management because they were good at the mechanical work. They’re organized, reliable, responsive, detail-oriented. They can run a planning process and keep a project on track. These are valuable competencies, and they’re about to be automated.

The managers who will thrive are the ones who, once freed from the mechanical overhead, turn out to have the leadership skills that were always there but never had room to develop. They’ll use tools like Growth Wise to handle the coordination layer and spend their newly available time on the work that actually requires a human: building alignment, developing people, making judgment calls in ambiguous situations, and having the hard conversations that no dashboard can have for you.

The managers who won’t thrive are the ones whose entire value proposition was the mechanical function. They kept things organized. They tracked things. They reported things. The system does that now. And the leadership skills that would justify their role were never developed because the mechanical work was all-consuming.

This isn’t a prediction about the distant future. Anthropic’s data shows 91.3% theoretical capability today. Gartner predicts that by 2026, 20% of organizations will use AI to flatten their structures, eliminating more than half of current middle management positions. Deloitte’s research on the future of middle management found that by end of 2024, advertisements for middle management roles had already dropped 42% from their 2022 peak. The reorganization of what “manager” means is already underway, and the job market is pricing it in before most organizations have noticed.

The organizations that recognize the management-leadership split and restructure accordingly will develop leaders. The ones that don’t will keep promoting coordinators into roles that coordination software is about to make redundant.

Growth Wise is a Decision Reliability platform. It instruments the human coordination layer so organizations can see which decisions are closing, which delegations are landing, and where coordination is breaking down. The management function, in other words. The leadership function is yours.

A senior leader has five meetings on Tuesday afternoon. Three are informational syncs. She sends a bot to those three. It captures what happened, generates a five-minute highlight reel for each, and she reviews all three in fifteen minutes.

This is no longer hypothetical. Recently, both the Klarna CEO and Zoom CEO Eric Yuan used AI avatars on their quarterly earnings calls. Yuan’s avatar, deployed via Zoom Clips, opened with: “I am proud to be among the first CEOs to use an avatar in an earnings call.” The CEO of Otter is reportedly training his own avatar to share the workload.

The productivity math is obvious. The instinct is right. But most of the conversation stops there — at whether to send the bot, not which meetings can actually survive the swap.

Two types of meetings. One type of bot.

The AI meeting tools on the market right now — Otter, Fireflies, Teams Facilitator, Fellow — are information processors. They record, compress, retrieve, and route. That’s genuinely useful. But every single one of them does only that.

None of them do coordination work. They don’t challenge an assumption mid-discussion. They don’t synthesize competing perspectives into something a group can act on. They don’t read tension in the room. They don’t force commitment. And they cannot bind their human to a decision.

That distinction is the whole ballgame.

At Growth Wise, we classify meetings into types we call Arenas. For example, a Status Sync exists to move information from the people who have it to the people who need it. No stress-testing, no mental model to align, no decision to close. For that Arena, a bot is a structural improvement — because everything that matters in a Status Sync is information, and the bot captures information well.

An earnings call where a CEO delivers prepared remarks to investors? That’s a Status Sync. The avatar is fine.

A decision forum where a cross-functional team has to commit to a roadmap date, stress-test assumptions, and actually close? That’s a different Arena entirely.

Where the bot breaks

When a bot replaces a human in a meeting that requires coordination, three things break.

First, alignment becomes fragile. The absent person receives a summary — a compression of what happened, not participation in it. The result is divergent interpretations that look like agreement until they don’t.

Second, the meeting drifts. Meetings shift Arenas mid-conversation. A Status Sync becomes a Decision Forum when someone raises a real issue. A human pivots with it. A bot continues in information-processing mode. The coordination function just goes missing.

Third, the decision doesn’t close. A bot cannot bind its human to anything. The group believes they decided; the absent person reviews a summary and quietly retains veto power. The decision reopens on Thursday.

When bots start acting, not just attending

Yuan has said Zoom aims to create digital twins of users — avatars that eventually do more than deliver scripted remarks. That next version is bots that don’t just observe meetings but act in them: approving budgets, committing to roadmap dates. When that happens, governance matters in ways the current conversation hasn’t caught up to.

The industry answer is “Governance as Code” — an immutable audit trail proving a decision was recorded. Necessary, but not sufficient. An audit trail tells you what happened. It doesn’t tell you whether the decision was structurally valid.

The Growth Wise model addresses this through Arena adherence. A bot approving a budget during a Decision Forum where that function was invited is doing legitimate work. A bot forcing a roadmap commitment during an Ideation Session — an Arena whose rules suppress convergence — is structurally invalid. Same action, different Arena, different legitimacy.

There’s a second distinction worth making: whether the bot is exercising judgment or transmitting a decision already made. “Budget approved per policy X” is enforcement, not judgment. The governance question for each is different, and conflating them produces frameworks that are either too restrictive or too permissive to be useful.

And then there’s the alignment gap. A bot can achieve action closure perfectly — log the decision, assign the owner, record the timestamp. It cannot achieve aligned closure. It cannot verify that the humans in the room understood the trade-offs. Any governance framework that doesn’t flag that distinction is measuring the wrong thing.

The question that actually needs answering

Yuan’s avatar opened an earnings call. Klarna’s CEO addressed investors through a digital proxy. These are Status Syncs — prepared remarks, one-directional, no coordination required. The bot belongs there.

Organizations that understand that distinction will delegate the right meetings, protect the right meetings, and build governance that knows the difference between an Arena-valid bot action and one that broke the coordination contract. Their decisions will stick.

Organizations that treat all meetings as interchangeable will send bots into Decision Forums and spend Thursday re-litigating what the avatar attended on Tuesday.

The question was never “should we send bots to meetings?” Yuan and Klarna already answered that. The question is which meetings are structurally safe for delegation — and when bots do act, whether they respected the Arena they were operating in.

Read the full article

Most leaders assume teams get better at working together over time. Experience builds intuition. Repetition builds cohesion. Eventually you get something greater than the sum of individual parts.

Research from the Collective Intelligence Labs at the Stockholm School of Economics says that’s wrong. And the failure mode it creates is worse than most people realize.

Researchers Philip Runsten and Andreas Werr studied 50 knowledge-intensive teams across 22 organizations over two months. Half got a structured digital debrief designed to make their coordination visible. The other half just kept working as normal.

The control teams didn’t hold steady. They declined. Observer-rated performance dropped roughly 6.5%. Self-rated performance dropped 4.6%. No disruption, no crisis. Just the quiet erosion of coordination quality that nobody was watching.

The intervention teams improved by 15-21%.

The default trajectory for a team left to its own devices isn’t improvement through experience. It’s decay through inattention.

That’s already uncomfortable. But here’s the part that really got me.

When the researchers segmented teams by performance, a specific cluster emerged: teams ranked just below the top. The “second best.” When the intervention came in and made their coordination visible, they didn’t improve. They got worse. And they resisted the whole process.

Their profile before the intervention told the story. Weaker task understanding than the top performers. Lower integration behaviors — less peer coaching, fewer real challenges to each other’s thinking. Lower psychological safety. And they consistently rated their own performance higher than outside observers did, by a margin that didn’t budge before or after the intervention.

One team member summarized it: “My strongest impression is that we as a team already are very good at many of the things this study wants to demonstrate, which is positive!”

Another said the intervention “created some discontent and irritation.”

The self-overrating gap — 0.36 units between self-assessment and observer rating — was exactly the same before and after. The feedback landed and changed nothing.

Compare that to the top-performing teams, who actually rated themselves lower than observers did. The best teams were humble about their coordination. The second-best teams were confident about theirs. The confidence was unfounded.

This is the catch-22. The teams most at risk are the ones least likely to seek help, least likely to accept feedback, and most likely to believe they don’t have a problem. They’re performing well enough that no one intervenes. They’re confident enough that they resist when someone does.

And they’re succeeding on familiar terrain. Output looks fine because the tasks are within range — routine enough that individual competence carries the performance, even as the actual quality of collaboration thins out underneath.

The moment the task becomes genuinely ambiguous — a novel strategic decision, a cross-functional tradeoff with no clear precedent, a situation where no single person holds enough context — those teams have nothing to fall back on. The integration behaviors were never practiced. The psychological safety to say “I don’t understand” or “I disagree” was never built.

They’ll walk into a high-stakes moment with the same confidence they’ve always had. And the coordination was never there.

This is how organizations produce decisions that, in retrospect, seem inexplicable. Not from obviously dysfunctional teams. From capable teams whose coordination was never made visible.

One more finding worth sitting with: the structured digital debrief achieved performance improvements comparable to facilitator-led interventions that typically show 20-25% gains.

You don’t need a consultant in the room. You need a reliable mechanism for making coordination visible to the team itself.

Read the full article

Reference

Runsten, P. and Werr, A. (2020). Knowledge Integration and Team Performance — The Effect of a Digitally Supported Team Debrief. Collective Intelligence Labs, Stockholm School of Economics.

You’ve been in that meeting. The one where something actually happened. Nobody was performing. Nobody was waiting for their turn to talk. Ideas were building on each other in real time.

What group flow actually is

Researchers define it as an emergent property of a social system, when members are so in sync that the group operates as a single unit and produces outcomes that exceed what any individual could have reached alone. Mihaly Csikszentmihalyi first mapped the individual version: total absorption, loss of self-consciousness, time distortion, intrinsic reward. Keith Sawyer extended it to groups, studying jazz ensembles, improv theater, and business teams to identify exactly what makes the collective state possible.

Sawyer identified ten specific conditions that have to be present for a group to enter that state:

1. The Paradox of the Group Goal: A compelling shared mission, open-ended enough for emergent creativity to surface

2. Deep Listening and Presence: Responding spontaneously to what’s actually being said, not rehearsing your next point while others speak

3. Complete Concentration and Environmental Boundaries: A clear boundary, physical or virtual, that eliminates distraction and shifts the group into a performance state

4. The Autonomy and Control Paradox: Each member feels individual agency while remaining willing to surrender to the collective direction

5. Blending Egos into a Group Mind: No one needs to be the center of attention; ideas build as if the group is a single unit

6. Equilibrium of Skills and Equal Participation: Everyone contributes their expertise at the moment it’s most relevant; no one is spectating

7. Familiarity and the Common Language: A shared knowledge base and unspoken communication style that reduce friction and speed up decision-making

8. Constant Communication and Immediate Feedback: Synchronous, spontaneous, real-time exchange that lets the group course-correct without breaking momentum

9. Momentum through the “Yes, And” Principle: Every contribution gets accepted and built on; negating ideas kills collective momentum

10. The Necessity of Potential Failure: Shared skin in the game — mental, social, or physical risk — creates the high-stakes environment that triggers intense focus

The neuroscience backs it up. When groups achieve genuine flow, their brains literally synchronize, with measurable alignment in neural oscillations, particularly in the beta and gamma frequency bands. Inter-brain synchrony peaks during collaboration and drops during conflict or stress. When jazz musicians improvise together, synchronization across temporal, parietal, and occipital regions increases significantly. The group is not a metaphor. It’s a measurable biological state.

How meetings kill every condition that makes it possible

Every single condition required for group flow is systematically dismantled by the way most organizations run meetings.

Sawyer’s triggers require deep presence: nobody writing their next point while someone else is still talking. Status meetings reward prepared remarks, polished updates, the performance of having things under control. Group flow requires ego dissolution, the willingness to let the group’s direction supersede your individual agenda. Most meetings are structured around individual agendas. Group flow requires equal participation and shared risk. Most meetings concentrate both speaking and accountability at the top of the hierarchy.

Why the system stays broken

The research on why this happens is not flattering. Deci and Ryan’s Self-Determination Theory identifies three non-negotiable human needs: competence, autonomy, and relatedness.

• When deep work is impossible because the day is fragmented, people can’t demonstrate competence through output, so they perform it in meetings instead.

• When calendars are controlled by other people’s invitations, calling a meeting becomes one of the few available acts of autonomy.

• When teams are too large for genuine collaboration, being in the room becomes a proxy for belonging, and being excluded triggers the same threat response as social exclusion.

A McKinsey survey of 1,200 executives found that fewer than half report their decisions are timely, and 61% say at least half the time they spend making decisions is ineffective. Most organizations never fix this because the people with the authority to change the system are the same people whose needs it serves.

Atlassian surveyed 5000 knowledge workers and found even more compelling evidence of the complex:

• Meetings are ineffective 72% of the time

• 78% of people we surveyed say they’re expected to attend so many meetings, it’s hard to get their work done.

• 77% of meetings end with a decision to schedule another meeting

• 54% of workers leave meetings without knowing what to do next or who owns the tasks

• 75% find meetings ineffective for making collaborative decisions

• 72% find meetings ineffective for creating goal clarity
  
  

  wrote a great piece on this recently.

Seeing The System

The Meeting-Industrial Complex

In 2018, Microsoft faced a crisis: surveys showed that entire groups of high-performing employees in their Surface and xBox divisions were miserable. These employees had specialized skills and would be hard to replace if they started quitting. Microsoft was finally making headway in hardware and the wrong employees leaving risked undoing their progress…

Read more

8 months ago · 4 likes · 4 comments · Ryan Moser

The direct conflict

Group flow requires the psychological safety to say “I don’t know,” to build on someone else’s idea rather than compete with it, to fail in front of the group without career consequence. The meetings industrial complex makes those things feel dangerous. You don’t lose self-consciousness in a room where your competence is under constant evaluation. You don’t listen deeply when you’re calculating whether speaking up will help or hurt you.

We’ve known for decades what genuine collective thinking requires. We’ve built organizational systems almost perfectly calibrated to prevent it.

Meetings stay broken because the people who could fix them don’t experience them as broken. And group flow stays rare because we keep trying to get it from a room designed for something else entirely.

Sources: Csikszentmihalyi, Flow; Sawyer, Group Genius; Deci & Ryan, Self-Determination Theory; McKinsey time management research

In 1865, William Stanley Jevons noticed something uncomfortable about steam engines: as they got more efficient at burning coal, total coal consumption went up. The cheaper something is to use, the more of it gets consumed. Efficiency doesn’t reduce demand. It unlocks it.

AI is doing this to cognitive work right now.

The productivity story is real. Drafts that took hours take minutes. Analysis that needed a team now needs a prompt. Execution costs are collapsing. But Jevons predicts what comes next: when doing becomes cheap, you don’t do less. You do dramatically more. When a document takes five minutes instead of fifty, you don’t produce one and reclaim the time. You produce ten. When analysis costs nothing, you run twenty.

Every new output creates a decision that didn’t exist before.

A Harvard Business Review study tracked an eight-month period at a 200-person tech company after they rolled out AI tools. Workers weren’t asked to do more. They just did more — because doing more felt possible. Product managers started writing code. Researchers took on engineering tasks. Individuals absorbed work that would have previously justified extra headcount.

One engineer summed it up: “You had thought that maybe, because you could be more productive with AI, you save some time, you work less. But then really, you don’t work less. You just work the same amount or even more.”

The workday doesn’t get shorter. It gets denser.

AI didn’t reduce cognitive load. It moved it from execution to evaluation, and from producing to deciding. The Harness-Uplevel study found developers using AI tools complete 21% more tasks, but code review times jump 91%. The work didn’t go away. It migrated to the approval layer. Senior engineers now spend more time vetting AI-generated code than they spent writing code before. And the natural pauses in the day — formatting a doc, waiting for a file to load — were also cognitive resets. When AI eliminates the pauses, what’s left is an unbroken sequence of decisions.

And here’s what compounds that: when more things run in parallel, coordination breaks down quietly. Agreements go unrestated. Ownership stays implicit. Context shifts between conversations and nobody marks the change. No one decides to create confusion, it just accumulates. I call this coordination entropy, and it exists in every organization. The difference is pace. When execution was slow, small ambiguities got caught before they spread. A vague decision affected one project. An unclear ownership assumption surfaced in a week. AI removes that buffer. Now the same ambiguity scales across twelve workstreams before anyone notices.

The part that frustrates me: none of this makes meetings less important. It makes them more important. Meetings are where coordination actually happens — where someone finally says “wait, I thought we decided X” and it either gets resolved or gets buried. That signal doesn’t live in Slack, in the task tracker, or in the doc. But most meetings aren’t built for deciding. They’re built for sharing. Status updates, decks, information transfers. As if getting everyone informed is still the hard part.It isn’t.

The hard part now is getting a group of people to evaluate what AI produced, decide what to do with it, and actually close on something. And we are, broadly speaking, terrible at that. Sensemaking as a group is hard. Holding a process while also participating in it is hard.

And the meetings industrial complex has made it worse — turning these moments where real value should get generated into performance. Another hour where everyone demonstrates their presence and nothing actually closes.

Work is going to get more complex, more uncertain, and more cognitively demanding. The HBR researchers call it a “self-reinforcing cycle” — AI accelerates tasks, which raises expectations for speed, which makes workers more reliant on AI, which widens the scope of what they attempt, which expands the quantity and density of work. Nobody mandated it. It just happened.

Structural complexity from AI is coming regardless. Coordination entropy isn’t. That part is a choice — but only for organizations that treat their coordination layer as something worth designing, not just scheduling.

The cost of doing is collapsing. The cost of deciding is just getting started.

This is the problem Growth Wise is built to instrument. Full piece at the link.

Read the full article

I’ve spent 15 years as a GTM and marketing strategist for startup and scaleup SaaS. And for 15 years I’ve encouraged founders to have a bold point of view — something that resonates with some people and alienates others. A position, not a positioning statement.

Now I am that founder, and I need to take my own advice.

With the caveat that I’m always open to changing my mind. This is where I stand today.

The cost of execution is approaching zero. The cost of knowledge is already there. Everyone will soon have the same facts, the same procedural know-how, the same ability to ship fast. Execution efficiency becomes table stakes. What can’t be commoditized is judgment: the ability to look at genuinely ambiguous situations, weigh competing perspectives, and decide.

That’s where humans still have a real shot. But only if we’re willing to do the hard part.

Dave Snowden’s Cynefin framework distinguishes between problems that are complicated and problems that are complex. Complicated problems have right answers, they just require expertise to find them. AI is genuinely good here. Complex problems are different. Cause and effect can only be understood in retrospect. There are no right answers, only bets. The right management approach is to probe, sense, and respond: run experiments, read what emerges, and adapt. This is the domain where most consequential organizational decisions live, and it’s the domain where most teams are least equipped to operate.

When the problem is uncertain, the tradeoffs are real, and the path forward requires actual perspective-taking and debate, AI struggles. It lacks what philosopher and cognitive scientist John Vervaeke calls perspectival knowing and participatory knowing: ways of understanding the world anchored in having a body, a history, a stake in the outcome. They’re social in nature. You can’t train them out of a corpus.

What makes complex problems hard isn’t a lack of information. It’s that getting to a decision requires groups to move through what facilitator Sam Kaner calls the Groan Zone: the uncomfortable stretch between divergent thinking (lots of ideas, competing frames, no consensus) and convergent thinking (shared understanding, a direction, a decision). It’s messy. People get confused and impatient. Conversations seem to go in circles. Someone usually tries to short-circuit it by calling for a vote, or defaulting to HiPPO (highest paid person’s opinion) in the room.

Most groups will do almost anything to avoid the Groan Zone. Anything that lets them feel like progress happened without sitting in the discomfort of not yet knowing.

The conditions for navigating that discomfort are getting worse, not better.

More white-collar work is now done remotely. And while remote work gets framed as a coordination problem, the deeper loss is perceptual. In a room, you’re reading the whole space. Who shifts in their seat when a topic comes up. Who makes eye contact with whom. The pause before someone speaks. Video strips most of that out. You get a grid of faces, degraded audio, and none of the spatial and temporal cues that tell a group how the room is actually feeling or who’s about to speak. The brain works harder to compensate, and still misses things.

At the same time, the teams most likely to produce something genuinely new are diverse ones. Diverse teams make better decisions than homogeneous ones. They also find it harder to reach consensus. More friction, more misreading, more moments where different communication norms create confusion. The Groan Zone is longer and more treacherous when the people in the room don’t share context.

So we’re asking more of our collaboration — more complexity, more diversity, more distributed — while giving it less to work with. Remote removes the cues. Diversity increases the friction. And the communication skills required to navigate all of it are getting weaker.

So we schedule another meeting. And another. Not because we’re inefficient —because the work is genuinely hard and we haven’t finished it yet. Add the politics layered on top: the very human instinct to avoid being the person who made the wrong call, to stay aligned with the group, to not be the one holding up the decision.

Uncertainty is uncomfortable. Social exclusion is worse.

The meeting becomes a place to keep moving without committing, to feel like you’re still in the room. Communication has eclipsed creativity as the central skill of white-collar work — and we did it to ourselves. No wonder we’re drowning in alignment meetings that don’t actually align anything.

Fewer meetings is part of the answer. The bigger part is making the ones we keep count for far more than they currently do. Right now most of them leave no legible record of what actually happened.

Meetings multiply because decisions don’t close. When a decision doesn’t close, the same conversation reconvenes. When ownership isn’t assigned, people keep gathering to re-establish it. Often what’s really happening is that the group hit the Groan Zone and retreated — scheduled another meeting rather than push through to something real. The calendar is downstream of that failure, not the cause of it.

The meeting is where negotiation actually happens. Where someone says “wait, I thought we were doing X” and it either gets resolved or gets papered over. Where accountability is accepted in real time or quietly avoided. Where the group’s actual coordination logic becomes visible. That signal doesn’t exist anywhere else. Not in the Slack thread, not in the task tracker, not in the shared Google doc.

That’s what I’m building as founder. I’m treating meetings as the diagnostic signal they actually are.

I hold this view for a reason that has nothing to do with research.

I’ve spent most of my career being hired to drive growth with virtually no resources, bringing to market products no one had seen before, from brands no one had heard of. With cash runway of maybe six to twelve months. Sometimes less. Whatever choice you made had to work, because there wasn’t a second one.

In those conditions you can’t relitigate decisions. You can’t ask for more data. You can’t wait for full alignment before moving. You have to get a room full of people — many of whom never saw the problem as theirs to solve, engineers who had never thought about go-to-market, product managers who figured sales was someone else’s department — and somehow get them to think together, decide with conviction, and move.

That was never a meeting problem, it was a coordination problem under fire.

The best solutions I’ve ever been part of came from groups that managed to do exactly that. Not because they had more information or more time, but because they got into the Groan Zone together and came out the other side with something real. An idea nobody walked in with. A decision everyone actually owned.

We did it in the time and with the resources most large companies would spend trying to align everyone’s calendar for the meeting and ordering the snacks.

That’s fifteen years of being taught, expensively, that the quality of how a group thinks together is not a soft skill. It’s the variable that determines whether a company survives long enough to find out if it had a good idea.

You know what observability is.

Logs, metrics, traces. You build it into every system you ship because without it you’re flying blind — you find out something broke when someone files a ticket, not when the error rate spikes at 2am. The whole point is to make failure patterns visible before they cascade.

Now think about how your organization runs its decision layer.

Every week, dozens of cross-functional decisions happen across your teams. Scope gets defined, trade-offs get negotiated, commitments get made. Some of those decisions close cleanly: explicit owner, captured rationale, everyone aligned. A lot of them don’t. The meeting ends, people leave with different interpretations of what was agreed, and two weeks later you’re debugging a scope conflict that was baked in at the kickoff.

You have no instrumentation for any of this. You can see that the meetings happened. You can’t see whether the decisions in them actually closed.

The same failure modes. Different layer.

The patterns that show up when you have no visibility into your coordination layer are familiar. Decisions get relitigated across multiple meetings because nobody explicitly closed them the first time. Scope creep accumulates because an assumption changed but nobody propagated the update. Two teams build in conflicting directions because each thought the other was adapting to the decision made three weeks ago.

These are structural failure modes — the kind you’d recognize immediately in a distributed system. Ambiguous state. Missing acknowledgments. Divergent replicas.

In SRE terms: you have observability on your compute layer, observability on your storage layer, and zero observability on the coordination layer that connects your people to your strategy.

What instrumentation on the coordination layer actually measures

The question it answers is not “are people meeting?” It’s “are the decisions that come out of those meetings actually holding?”

Did the decision close with an explicit statement of what was decided? Does it have a named owner? Was dissent surfaced in the room or suppressed? Is the rationale captured well enough that someone who wasn’t in the meeting can reconstruct why?

These are the signals that determine whether a decision propagates cleanly through the organization or unravels the first time a dependency shifts. When decisions consistently fail on one or more of those signals, you get decision churn — the same questions relitigated across meeting after meeting, alignment that looks achieved but dissolves under pressure.

The missing layer

There are tools for tracking what your teams are building. Tools for how they communicate. Tools for who made which calls and when. None of them answer whether the decisions that created those tasks actually closed, or whether the same scope question is cycling for the third time.

Decision Reliability Infrastructure is the layer that sits between your communication tooling and your execution tooling — making visible the coordination work that happens in between.

You wouldn’t ship production code without instrumenting it.

→ Read the full piece