By Andreas Pettersson, Founder, Leaders ADAPT
EOS was created before AI. That is not a criticism of the people who built it. It is a date, and it is why AI for EOS companies has become the question I get asked most.
These days AI is a massive culture change, and EOS needs to evolve so it does not miss it.
I am a big fan of EOS. I ran it for years at Arcules, the company I founded and scaled to 150 people, and we run it at Leaders ADAPT today. I also install AI for leadership teams as a fractional AI executive. So the question usually arrives in one line: we run EOS, now where does AI go?
This guide is my answer on AI for EOS companies. It maps AI to each of the six components, says who should own it, shows what goes wrong and what results are realistic, and draws the line on what stays human.
Quick answer: In a company running the Entrepreneurial Operating System (EOS), AI fits best in preparation and tracking: scorecard pulls and exception flags, Rock status collection, Level 10 Meeting pre-reads and to-do capture, issue capture, process documentation and first drafts of the vision. It should not own people decisions, the solve in issue solving, accountability consequences or strategic commitments. Those stay with the leadership team.
Leaders ADAPT works independently of every framework owner. We are not affiliated with, certified by or endorsed by EOS Worldwide, which owns the EOS name and tools, or by Scaling Up, Pinnacle Business Guides, FranklinCovey or anyone else who owns a business operating system. Framework and software names on this page are descriptive. We reproduce no proprietary tool, and the AI map below is our own.
Where does AI fit in each of the six EOS components?
AI in EOS means using AI for the preparation, tracking and documentation work around the six components, while every decision, people call and commitment stays with the leadership team.
EOS organizes a company around six components: Vision, People, Data, Issues, Process and Traction. Here is the map I use with clients. I tested it against how we actually run the system, and I stand behind every row.
| Component | What it covers | Where AI fits | What AI never does | Go deeper |
|---|---|---|---|---|
| Vision | The Vision/Traction Organizer, core values, long range target, one year plan | Drafting help: first drafts from offsite notes, checks that the plan matches the vision | Choose the direction, the values or the targets | Visionary vs Integrator |
| People | The Accountability Chart, right people in right seats | Never. At most it books the meeting | Judge fit, decide seats, deliver the verdict | Leadership team development |
| Data | The scorecard and measurables | Scorecard pulls from source systems, exception flags | Explain why a number moved | The weekly KPI scorecard |
| Issues | The issues list and IDS | Capture and triage: collect, group, remove duplicates, suggest a ranking | The solve | The Level 10 Meeting |
| Process | Core processes, documented and followed | Documentation and retrieval: draft process docs from recordings, answer how we do things | Decide the process or excuse someone from it | The operations manual |
| Traction | Rocks, the L10, to-dos | Rock status collection, L10 pre-reads, to-do capture, scoring the meeting | Set Rocks, hold people to them | Quarterly planning |
People gets one word: never. The whole point of that component is judgment about humans, made by humans who know them.
Here is what a Monday looks like once the Traction and Data rows are in place. On Sunday night the scorecard pulls itself from the CRM, the accounting system and the helpdesk, and every number that missed its goal is flagged. Rock owners got a short status request on Friday, and their answers are already summarised into a one page pre-read. The issues raised in Slack and email during the week sit in one list, grouped and deduplicated.
The leadership team walks in having read one page. The reporting sections stay inside their time boxes, and IDS gets its full hour. After the meeting, the transcript becomes draft to-dos with owners, and a person confirms them before they go out.
Nothing in that Monday decided anything. It only cleared the table so the team could.
If I had to pick one row to start with, it would be Traction. My favourite single AI use inside EOS is a second opinion on the weekly meeting. We rate the L10 ourselves, AI rates it from the transcript against the meeting rules, and the gap between the two scores tells us what to fix. The method is in score your meetings with AI.
Who should own AI in a company running EOS?
EOS Worldwide has a clear public line on this. A post on its blog in June 2025, Use AI or Fall Behind: Why Integrators Must Lead the Charge, argues that the Integrator should drive AI adoption.
I see it differently, and this is the position across our whole series.
The CEO owns AI. The same way the CEO has to own EOS and be its visible supporter inside the organization. And the Integrator has to be fully bought in. Both of them, or it does not happen.
Not because the Integrator lacks the skill. Because AI is not a tool rollout. It is a culture change, and culture changes run from the top.
Then the CEO delegates the implementation to the whole C suite, not only the Integrator. In my words: "AI is a cultural change, if you have one person sandbagging it's never gonna go anywhere." Once the decision is made, you go full speed, like any other change management.
One more thing the CEO has to hear early. AI does not solve the leadership problem. It makes everyone more effective and more efficient.
That is a big deal. It is not the same deal.
None of this makes the Integrator less important. AI for Integrators is real, and the seat gets more hours back than any other, because so much of it is chasing, collecting and preparing.
An Integrator who is bought in becomes the person who knows which workflows are worth automating, because they feel the friction every week. Owning the decision and doing the work are different jobs. The CEO owns the first. The Integrator leads a large share of the second.
What goes wrong when companies self implement AI?
Here is what happens when a company self-implements AI without a framework.
Someone gets excited. They start using general AI assistants and coding agents, run long prompts, and get impressive output in week one.
Nobody puts a validation framework in place. Nobody turns the good prompts into reusable skills with rules. Things go off the rails every now and then, and nobody notices until a number in a board pack is wrong.
The extreme version is real. We have walked into an organization where a vibe coder had built the company CRM as a single HTML file with over 12 million lines of code.
That is not an AI problem. That is an operating problem wearing an AI costume.
The quieter danger is worse. Too much information in the hands of people who do not understand the context can send a whole organization in the wrong direction, confidently and fast.
A validation framework sounds heavier than it is. It is a short set of checks: known answers the output must match, a human sign off on anything that leaves the building, and a log of what the AI changed. Without it, nobody can tell a good week from a lucky one.
The fix is boring. About a month of proper training, with the right ontology, rules and framework, and results come quicker after that. Every client I have seen skip that month has paid for it later.
What results can AI deliver for an EOS leadership team?
These are my own client figures, rounded, as of September 2026:
| Result | Context |
|---|---|
| $200,000 to $300,000 in hours saved per quarter | Clients automating email, calendar and reporting work |
| About 95 percent of that work automated | The decisions were kept with the managers |
| Open requisitions closed | A finance team that was hiring stopped, because the AI processes absorbed the load |
The pattern behind those numbers matters more than the numbers. The wins came from the boring middle of the company: inboxes, calendars, status reports, month end packs. None of them came from AI making decisions.
That lines up with the map above. Data, Issues, Process and Traction carry most of the preparation work. That is where the hours are.
AI for management is mostly AI for the paperwork around management. The management itself stays yours.
How does an EOS company start with AI? One AI Rock in 90 days
Most companies running EOS already know how to execute a priority for 90 days. Use that muscle.
The One AI Rock method treats your first AI project as a single company Rock: one workflow, one owner on the leadership team, a measurable finish line at day 90, reviewed every week in the L10.
- Pick one workflow from the Traction or Data row. The scorecard pull or the L10 pre-read are the safest first choices.
- The CEO sponsors it. The Integrator is bought in. One leader owns it.
- Write the rules and the validation checks before anything is built. Turn what works into a reusable skill, not a one off prompt.
- Budget about a month for training the people who will use it.
- Review it like any other Rock: on track or off track, every week.
- On day 90, keep it, extend it or kill it. Then choose the next one.
If you want someone to run that first AI Rock alongside your team, that is what a fractional AI executive engagement is built around.
Before you start, run our one page AI in your operating system checklist, or take the AI readiness assessment to see where your team stands. If you want the longer planning version, the AI roadmap template covers it, and our 9 step AI operating system playbook is the version for companies that do not run EOS.
Do EOS software tools like Ninety or Bloom Growth have useful AI? Our view
This section is our view, labelled as ours, not a product review. I have no strong position on the software, and we have not run a structured test of each vendor's AI features. We also do not quote prices here. Check each vendor's own pricing page.
The four tools most EOS companies look at are EOS One, which comes from EOS Worldwide, and Ninety, Bloom Growth and Strety. Each is built to hold your Rocks, scorecard, issues, to-dos and meetings in one place. Several now market AI features.
Our view is that a dashboard with a chatbot is still a dashboard. Choose EOS software for how well it runs your meeting discipline and how easily your data flows in and out. Treat the AI features as a tie breaker. These are the questions we would ask any vendor:
| Question for the vendor | Why it matters |
|---|---|
| Does it pull numbers from your source systems, or wait for someone to type them? | Most scorecard pain is manual entry |
| Can you export all your data, or reach it through an API? | Your AI workflows will cross several tools |
| Does its AI summarise, or does it recommend decisions? | Summaries are safe. Recommendations about people are not |
| Can you see what the AI changed, and why? | Validation needs an audit trail |
| Does it still work if half the team rarely logs in? | Adoption beats features |
What should AI never do in EOS? What stays human
This is the list I would put on the wall of every leadership team running EOS with AI:
- People decisions. Hiring, exits, seat changes and fit calls.
- Issue solving judgment. AI can frame an issue. The S in IDS belongs to the team.
- Accountability consequences. What happens when a Rock or a to-do is missed again.
- Strategic commitments. The Rocks, the one year plan, the long range target.
- Performance conversations. Feedback lands from a person, or it does not land.
- External promises. Anything said to a customer, a lender, a board or a partner.
AI can prepare material for every one of those. It should decide none of them. We go deeper in what not to delegate to AI.
What AI does not fix in an EOS company
If you have not done good leadership to begin with, it does not matter what process you put on top of it. Introduce AI into that and you will amplify the problem even more.
EOS has the same limit. It helps a mediocre leader run a slightly more structured process and get away with a little more. Maybe that buys a plateau to grow from. It does not replace leadership.
Where EOS is lacking is the how when it comes to leadership. How to delegate. How to make people accountable. How to run 1:1s and give feedback.
AI does not supply any of that either. Our leadership team development guide does.
There is a Nordic angle here that I find useful. EOS is good at surfacing issues, and Nordic leadership is good at separating the problem from the human. AI makes the surfacing faster and louder. The separating is still a leadership skill, and it gets more important, not less, when issues arrive at machine speed.
Size changes the starting point, not the rule. EOS works from about 2 to 250 people in my experience. In a ten person company, AI mostly gives the founder back the Integrator hours they cannot yet pay for. In a 150 person company, it is a process change across departments, and the C suite has to carry it together.
And if you are still deciding whether to run EOS at all, start with our independent EOS review and the comparison of business operating systems.
The whole AI for EOS companies guide, page by page
- The Level 10 Meeting: the agenda, a free template, the five ways it dies, and AI scoring.
- The weekly KPI scorecard: 5 to 15 numbers, one owner each, and the AI pull that works.
- The EOS Integrator: the role, the job description, the cost and the authority lesson.
- Visionary vs Integrator: the two seats, a self test, and the third seat when you can afford one.
- Quarterly planning: the one day agenda and why one to three Rocks beat ten.
- The AI chief of staff: how coordination work moves to AI while judgment stays with people.
For the leadership side of AI adoption, the AI Leadership Mastermind is our peer program for CEOs, and my book AI Leadership Mastermind covers what not to do with AI projects.
So, one question. Which of the six components would you hand to AI first, and who on your leadership team would own it? Reply and tell me. I read every answer.
Common questions about AI for EOS companies
Where does AI fit in the Entrepreneurial Operating System?
AI fits in the preparation and tracking work around the six EOS components. Typical uses are drafting help for the vision, scorecard pulls and exception flags for data, capture and triage of issues, documentation and retrieval of processes, and Rock status collection, meeting pre-reads and to-do capture for traction. The people component, meaning seats and fit decisions, stays human.
Who should own AI in a company running EOS?
EOS Worldwide has published the view that Integrators should lead AI adoption. Leaders ADAPT holds that the CEO should own AI, in the same way the CEO owns and supports running EOS, with the Integrator fully bought in. Implementation is then delegated across the whole leadership team, because a single uncommitted leader can stall adoption.
Can AI implement EOS for you?
No. AI can draft documents, pull numbers, prepare meetings and track commitments, but EOS depends on leaders making decisions, holding each other accountable and running the weekly and quarterly rhythm themselves. In AI for EOS companies, the realistic goal is less preparation work and more meeting time spent on decisions, not a system that runs without the leadership team.
Does EOS need to change because of AI?
EOS was designed before current AI tools existed, so it says little about them. The core structure of vision, people, data, issues, process and traction still applies. What changes is the work around it: much of the preparation, data gathering and documentation can be automated, and leadership teams need explicit rules for what AI may and may not decide.
What should AI never do in an EOS company?
AI should not make people decisions, solve issues on the leadership team's behalf, decide consequences when commitments are missed, set Rocks or other strategic commitments, conduct performance conversations, or make promises to customers, boards or partners. It can prepare material for each of these. The decision and the accountability remain with named leaders.
What results can AI deliver for a company running EOS?
Results depend on where AI is applied and how well it is validated. Leaders ADAPT reports that client teams automating email, calendar and reporting work have recovered roughly $200,000 to $300,000 in staff hours per quarter, with about 95 percent of that work automated and decisions kept with managers (rounded, September 2026).
Your next question
- The Level 10 Meeting: the best first place for AI for EOS companies.
- The weekly KPI scorecard: the second best, and the easiest to validate.
- Fractional AI executive: when you want it installed with the rules and training in place.
What it is: a fractional AI executive who installs AI inside the operating system you already run, one Rock at a time, with validation rules, a month of training and every decision kept with your managers. Who it is for: CEOs of companies running EOS who want AI for EOS companies done properly, not self-implemented and cleaned up later. Talk to a fractional AI executive, or take the AI readiness assessment first.
