The outdated org chart: who owns that now?

Start here

One piece of work, three moments.

Take the monthly report your commercial team lives on. It has always been made by Anna, an analyst who has been there six years. Nothing about it changed on the chart. Move through the three moments and watch which of the three answers goes first.

Anna builds it. Two days of her month, every month, and everybody knows that.

Who makes itAnna, two days a month

Who checks itHer team lead, half an hour

Who answers if it is wrongAnna

A boring, healthy arrangement. Three clear answers and a name for each.

The third row is the one that goes first, and it goes without a decision. Nobody handed the accountability to anyone. It simply stopped having a name while the work carried on arriving.

Drawn from several conversations rather than one company.

You have seen this

A normal week, looked at closely.

None of these is a crisis. That is exactly why nobody escalates them, and why they are still true a year later.

A colleague at work with her agent beside her — **Tuesday, and three people say yes** You ask in the team meeting who owns the weekly forecast. Three hands go up, and all three are being honest. One set it up, one runs it, one presents it. Nobody has ever had to decide which of those is ownership, and now everybody assumes it is one of the other two.

Someone reviewing work alongside their agent — **“That used to be Anna”** You ask who checks the numbers before they go out. The answer starts in the past tense. Anna moved to another team in January, the checking went with her for a while, and then it stopped moving and simply stopped.

Two colleagues going through work together with their agents — **The handover that nobody wrote down** Two teams both improved their half of the same flow. Both versions are better. Between them sits a step that used to be a conversation and is now two systems assuming the other one handled it. A customer finds it in about four months.

The outdated org chart is a company where the work moved and the boxes did not. Only 13 in a hundred people who use AI say they are rewarded for changing how the work is done when the result does not arrive straight away.1

Read this bit if nothing else

Microsoft surveyed 20,000 people who use AI at work, in ten countries including the Netherlands. Only 13% say they are rewarded for reinventing how work gets done when the results are not immediate.1
45% say it feels safer to stick to current goals than to redesign the work, and 65% fear falling behind if they do not adapt. Those are the same people.1
Only 26% say their leadership is clearly and consistently aligned on AI. One in four.1
Microsoft calls it the Transformation Paradox: people are ready to change how they work, and the metrics, incentives and habits around them keep rewarding the old way.1
Three questions fix most of this, and they take one meeting. They are further down.

Something like this has probably happened

You asked who owns something and got three names or none
Somebody left and you discovered what they had quietly been carrying
Two teams improved the same flow and something fell between them
You are about to redraw the chart and you would rather redraw the right thing

What you take away

Three questions that show you where ownership has gone missing
The research on why people wait, with the numbers and the sample
Why rewriting the chart before you look at the work makes it worse
What to change first when the honest answer is that nobody knows

In plain terms

An org chart describes who reports to whom. It never described who owns what.

That was fine for a long time, because ownership followed the reporting line closely enough. The person who did the work reported to the person who checked it, and the two questions had the same answer.

Then tasks started moving without people moving. Somebody sets up an agent, it produces something every week, and it keeps producing it after that person's role changes. Nothing about the chart is wrong. It simply stopped answering the question you actually need answered.

Nobody hands over accountability. It just stops having a name, and the work keeps arriving anyway. Paul Musters

Microsoft put three questions in this year's Work Trend Index that every company will have to answer once agents do part of the work. They are worth reading slowly, because no chart on any wall answers any of them.1

1

Who reviews how the agent performed?

Not who set it up. Who looks at what it produced this month and says whether that was good enough.

One name

2

Who may change the workflow it runs?

And who approves that change. If the honest answer is anyone with access, write that down, because that is the finding.

One name, plus who signs

3

How does a local win get captured and spread?

Somebody in your company solved something clever last month. Name the route by which the rest of the company gets it.

A route, not a person

Three questions, and in most companies the third one has no answer at all. Good things happen and they stay exactly where they happened, which is the quiet reason a company of two hundred keeps solving the same problem four times.

The measured part

People are ready. The system around them pays for the old way.

Microsoft surveyed 20,000 people who use AI at work across ten markets this spring, including the Netherlands. Three answers from that survey explain why nobody redraws anything.1

Fear falling behind if they do not adapt quickly

65%

Say it feels safer to focus on current goals than to redesign the work

45%

Say they are rewarded for reinventing the work when results are not immediate

13%

The same 20,000 people, three questions. Wanting to change, feeling it is safer not to, and knowing it does not pay.1

Read those three as one sentence and you have the whole thing. Two thirds are afraid of being left behind. Nearly half find it safer to keep their head down. And thirteen in a hundred believe this company would back them if they changed the work and it took a while to show. Microsoft calls it the Transformation Paradox, and it is a fair name.

26%

say their leadership is clearly and consistently aligned on AI. One in four, in a survey of people who use it every day

1

67%

of AI's reported impact comes from the organisation around a person, against 32% from their own mindset and behaviour

1

15×

growth in active agents inside Microsoft 365 in one year, and 18 times in large enterprises

1

That middle number is the one worth arguing with your own instincts about. Two thirds of whether AI does anything for a person sits in the company around them: the culture, the manager, how people get developed. It is not the person and it is not the tool.

One caution about all of these. They are self-reported answers, collected from the same person at the same moment, and Microsoft says plainly that they show association rather than cause. Read them as a consistent picture of how it feels inside companies right now, which is exactly what you need here.

And almost nobody writes any of it down

Microsoft also asked whether agent workflows, human handovers and quality standards are documented and repeatable. Even among the most advanced AI users in the survey, the answers are low.1

Written down at team level

Agent workflows, handovers and quality standards

Advanced26%

The rest19%

1 in 4Even among the most advanced AI users in the survey.

Written down across the company

The level at which a good idea can actually travel

Advanced25%

The rest14%

14 in 100Which is your answer to the third question.

Share saying agent workflows, human handovers and quality standards are documented and repeatable, comparing the most advanced AI users in the survey with everybody else.1

Fourteen in a hundred. That is your answer to question three, and it explains why a clever fix in one team stays in that team until somebody moves desks.

Where it comes from

People were being sensible, one task at a time.

Redrawing the chart is expensive and redrawing the work is not visible

A reorganisation is a project with a name, a slide and a date. Ownership drifting away from a weekly report is none of those. So the first gets attention and the second gets noticed the week a customer asks a question nobody can answer.

The person who would fix it has nothing to gain

Somebody in your company can see exactly where the ownership went. Changing it means raising a question that makes work for four people and makes them look like the person who complicates things. Thirteen in a hundred believe they would be backed for that.1 The other 87 are making a reasonable calculation.

Everybody can see where it went. Nobody is paid to say so. Paul Musters

Leadership disagrees quietly rather than loudly

One in four people say their leadership is clearly aligned on AI.1 That does not mean three in four leadership teams are fighting. It means they have never had to be specific enough to disagree, so everybody keeps their own version and the org chart keeps describing whatever the last reorganisation decided.

Which level is the chart drawn for?

Six questions, answered anonymously, three minutes. Where you sit, what it costs, and what the next level looks like.

Do the Culture Level scan

How to see it

Three names, or the honest gap where one should be.

Pick one thing your company produces every week that AI now largely makes. A report, a summary, a first draft that goes to a customer, a set of numbers somebody acts on. Then write three names on a piece of paper.

1

Who owns it

One person whose job it is that this exists and is good. If three people could claim it, that is the same as nobody, and the meeting where you settle that takes ten minutes.

2

Who checks it

Before it goes out. Ask that person what they actually look at, and expect a narrower answer than you hoped for. If the answer starts with a name in the past tense, you have found it.

3

Who answers if it is wrong

When a customer or a board member asks. One name, not a team. This is the row that goes missing first and it goes without anybody deciding.

Do it for three pieces of work and you will have a better picture of your organisation than the chart gives you. Do not fix anything in the same meeting, because the first instinct is to appoint somebody and the right answer is often to change the work instead.

One thing I have learned running this. Start with something that is going fine. Pick a piece of work that is causing trouble and the meeting turns into who is to blame, and then everybody stops answering honestly for the rest of the exercise.

On the ladder

Before you redraw anything, find out who does what now.

The temptation after this exercise is to redesign the chart. That usually produces a new chart with the same problem, because it describes reporting lines again and the work has already moved somewhere else.

An Operating Profile in use. Personality type and AI level in one profile, with the agents that fit it.

A job title tells you what somebody was hired for. It does not tell you that the person two levels down is running four workflows end to end while somebody senior is still checking output line by line. Those two need completely different things from you, and the chart shows them as one box above the other.

Two readings per person: the way somebody works, and how far along they are with AI. What comes out is a map of who actually carries what, which is the thing you wanted when you reached for the chart. Then you redraw, once, with something real underneath it.

About emaho

emaho measures one Operating Profile per person: personality type and AI level in a single profile. On that we build a personal set of AI agents that fit how that person works, inside the tools they already use. Fifteen minutes to complete, first profile free, built for companies between 20 and 500 people.

See your own profile, free Find your AI Culture Level

Fifteen minutes per person. No credit card, no strings.

What to do

Stop redrawing. Start mapping.

The instinct after an exercise like this is to open the org chart and move boxes. That produces a new chart with the same problem, because a chart describes reporting and the thing that moved is ownership.

The one thing to stop

Stop treating the chart as the answer to who owns what. It stopped being that answer somewhere around the point where tasks began moving without people moving. Say that out loud in the leadership team, because until somebody does, every conversation about accountability keeps returning to a diagram that cannot settle it.

Three things to start

Write the three names for three pieces of work. Owner, checker, the person who answers. This week, in a meeting you already have, on work that is going fine. You are mapping, not fixing, and the first version will be uncomfortable enough on its own.

Close the gaps one at a time, and prefer changing the work over appointing somebody. If three people could claim ownership, the answer is usually that the work should be split differently rather than that two of them should care less.

Answer the third question for your own company: how does a good thing travel. Pick a route and make it dull. Fifteen minutes in an existing meeting where one team shows what they changed beats any platform you could buy for it, and unlike the platform it will still be running in a year.

One thing that helps all three along. Say in the all-hands that redesigning how the work is done is part of the job, and name somebody who rebuilt something whose numbers have not moved yet. Thirteen in a hundred people believe their company would back them for that.1 You change that number with an example, not with a policy.

Which row went missing?

Send me the piece of work you picked and which of the three names you could not fill in. I will tell you what I usually see behind that and what I would change first. No deck. One message back.

Message me on WhatsApp

What it costs

The chart keeps describing a company that stopped existing.

The visible cost arrives as a question you cannot answer. A customer, an auditor or a board member asks how something was produced and who stands behind it, and the honest answer takes a week to assemble. That conversation is survivable once and expensive twice.

Underneath sits the thing that never travels. Fourteen in a hundred companies say their workflows and handovers are written down at organisation level.1 So a good solution stays in the team that found it, and a company of two hundred solves the same problem in four places without ever noticing.

And the one that arrives with a resignation letter. Somebody leaves, and in the weeks after you discover what they were quietly carrying: three workflows, two judgement calls and the reason a particular number could be trusted. None of that was on the chart, which is why nobody could hand it over.

Questions people ask

The org chart, in questions

What is the Transformation Paradox?

It is Microsoft's name for a company where employees are ready to change how they work while the metrics, incentives and habits around them keep rewarding the old way. In the 2026 Work Trend Index, based on 20,000 AI users across ten markets, 65% feared falling behind if they did not adapt, 45% said it felt safer to stick to current goals, and 13% said reinventing the work is rewarded when results are not immediate.

How should an org chart change when AI does part of the work?

Start by mapping ownership rather than reporting lines, because those two used to have the same answer and no longer do. For one piece of work, name who owns it, who checks it and who answers if it is wrong. Redraw only after you have done that for several pieces of work, or the new chart repeats the old mistake.

Who owns work that an AI agent produces?

Whoever you name, and if you have not named anyone the honest answer is nobody. Microsoft's 2026 report puts three questions to every company running agents: who reviews how the agent performed, who may change the workflow it runs, and how a local win gets captured and spread. Most companies can answer the first two and not the third.

Why does accountability disappear when nobody leaves the company?

Because tasks move without people moving. Somebody sets up a workflow, it keeps producing after their role changes, and no line on the chart ever changed to reflect it. Nobody hands accountability over. It stops having a name while the work keeps arriving.

How many companies have their AI workflows written down?

Very few. In Microsoft's 2026 survey, 14% of respondents said agent workflows, human handovers and quality standards are documented and repeatable at organisation level. Even among the most advanced AI users in the same survey it was 25%. That is why a good solution in one team stays in that team.

Where can I read about the other AI challenges inside organisations?

This is one of 25 AI challenges emaho documents. The outdated org chart sits next to agents nobody owns and the coordination tax.

Is leadership usually aligned on AI?

Not in most companies. Only 26% of the 20,000 AI users Microsoft surveyed said their leadership is clearly and consistently aligned on AI. That rarely means open disagreement. More often the leadership team has never been specific enough to discover that they disagree.

Does the organisation matter more than the individual for AI results?

By roughly two to one. Microsoft tested 29 factors against whether AI is delivering anything for a person. Organisational factors such as culture, manager support and talent practices accounted for 67% of reported impact, against 32% for the individual's own mindset and behaviour. Microsoft notes these are associations rather than proven causes.

What is the test for an out of date org chart?

Pick one thing your company produces every week that AI now largely makes. Write down who owns it, who checks it and who answers if it is wrong. Three names, no teams. If any of the three is missing or contested, the chart is describing a company you used to be.

Should we reorganise because of AI?

Map before you redraw. A reorganisation produces a new set of reporting lines, and the problem here is that ownership has drifted away from those lines rather than that the lines are wrong. Most companies get more from fixing three specific pieces of work than from a new chart.

Why do people avoid redesigning how the work is done?

Because it is not rewarded and it is visibly risky. In Microsoft's 2026 survey only 13% said reinventing work with AI is rewarded when results are not immediate, while 45% said it feels safer to focus on current goals. Both groups are making a reasonable calculation about their own position.

How do you spread a good AI workflow across a company?

Pick a dull, repeated route and use it. A fifteen-minute slot in an existing meeting where one team shows what they changed works better than any platform, and it answers the question most companies cannot: how does a local win travel. Only 14% report having anything documented at organisation level.

Close by

Living in the same gap

Each one is a piece of the same question: who carries this now that a machine does part of it.

01

Nobody owns the agent

The same three names, applied to the things that act on their own. Five lines that give one agent an owner in ten minutes.

Read this one 11

Coordination cost

What the missing handover costs in days. Teams finish 21% more work while the queue for review nearly doubles.

Read this one 23

Managers absorb it

Where the unowned work actually goes. It lands on the layer that has least room, and they are the last to raise it.

Read this one

All 25 AI challenges, and what to do about each

Twenty-five things that break inside a company once people start using AI, each with the research behind it and the level where it starts to bite. This one starts to bite at Engine, and 2 of the others start there too.

Open the full set

Your next step

Map who carries what, before you redraw anything.

A chart shows reporting lines. What you need is who actually runs which work with AI, per person. Fifteen minutes each, and then you redraw once with something real underneath it.

Take one report and ask who makes it nowAnna's name is on the chart. Find out whose hands it actually passes through this month.
Map who runs what, per personOne profile each shows who runs whole workflows and who follows one, which is rarely what the chart says.
Redraw once, with the real pictureRedrawing twice in a year costs more than waiting six weeks and doing it properly.

Start with your own profile, free Ask me anything first Plan a 30-minute call

First profile free · no credit card · built for companies of 20 to 500 · you decide what your team gets to see

Not ready to put your team in anything yet? Start with the level of the company instead. The Culture Level scan is six questions, three minutes, and asks nothing of you.

Paul MustersFounder of emaho

Paul Musters

Fifteen years of leadership and team development in Dutch scale-ups. That practice now sits in software: one Operating Profile per person, with agents that actually fit. He writes these pages from what he runs into with clients, not from a research summary.

LinkedIn · paul@emaho.world · WhatsApp

Sources and numbers used on this page

Microsoft, 2026 Work Trend Index Annual Report, published 5 May 2026. Survey by Edelman Data x Intelligence among 20,000 knowledge workers who use AI at work, across ten markets including the Netherlands, 2,000 per market, fielded 18 February to 7 April 2026, alongside anonymised Microsoft 365 telemetry. Source of: 13% say they are rewarded for reinventing work with AI even when results are not met; 45% say it feels safer to focus on current goals than to redesign work with AI; 65% fear falling behind if they do not adapt quickly; 26% say their leadership is clearly and consistently aligned on AI; organisational factors such as culture, manager support and talent practices account for 67% of reported AI impact against 32% for individual mindset and behaviour, across 29 measured factors; the number of active agents in Microsoft 365 grew 15 times year on year and 18 times in large enterprises; among the most advanced AI users 26% say agent workflows, human handovers and quality standards are documented and repeatable at team level, 29% at function level and 25% at organisation level, against 19%, 17% and 14% for other respondents. The three questions about reviewing agent performance, authority to change workflows and spreading a local win are Microsoft's own, from the same report. Microsoft names the pattern the Transformation Paradox and states that survey items are self-reported and show statistical association rather than causal effect.

Numbers are quoted as published. The three moments in the timeline and the three scenes are composites drawn from client situations rather than transcripts, and the names in them are not real.