AI in Your Operating System: The One Page Checklist
Every operating system on the market was designed before AI. EOS included. AI is now a massive culture change, and a leadership team that runs a system needs to know where AI goes in it, and where it never goes.
This is the one page audit I use with leadership teams. Four questions: which meetings, which numbers, which processes, and which people decisions stay human. Each line maps to one of the six components of an operating system: Vision, People, Data, Issues, Process and Traction. Tick what is true today. The tally tells you whether to write the rules first or start one AI priority this quarter.
Clients have saved $200,000 to $300,000 in hours per quarter from email, calendar and reporting automation, with about 95 percent of that work automated and the decisions kept with the managers. Those are my client figures, rounded, September 2026. None of them came from AI making a decision.
The AI in your operating system checklist
Which meetings, numbers and processes can AI take on in our system, and which people decisions stay human?
What you get
- 27 lines across meetings, numbers, processes and people decisions
- Each line mapped to Vision, People, Data, Issues, Process or Traction
- A live tally and what it means for your next 90 days
- A print version and a CSV of your answers
Open the full template
Enter your first name and work email and the template opens on this page at once. What you type into the template stays in your browser and is never sent.
Done. The template is open below.
The template is open. You gave us your email on an earlier Leaders ADAPT tool, so there is nothing to fill in.
Tick every line that is true in your company today.
Which meetings
| True | Line | Component | Where AI sits |
|---|---|---|---|
| The weekly leadership meeting has a fixed agenda and runs whether or not everyone is there. | Traction | Human rule | |
| Status on every quarterly priority is collected before the meeting, not read out during it. | Traction | AI can collect | |
| A pre-read with the numbers, priorities and top issues goes out 24 hours ahead. | Traction | AI can draft | |
| To-dos are captured with one owner and one date before people leave the room. | Traction | AI can capture | |
| The team scores each meeting, and would accept a second score from AI against its own meeting rules. | Traction | AI can score | |
| Issues are captured in one list between meetings, with duplicates merged before the meeting. | Issues | AI can triage | |
| The solve stays in the room: the team decides, AI only frames the issue. | Issues | Stays human | |
| Planning day notes become a draft plan within a day, and the choices in it are made by the leadership team. | Vision | AI can draft |
Which numbers
| True | Line | Component | Where AI sits |
|---|---|---|---|
| Every scorecard number has a written definition of what counts. | Data | Human rule | |
| Every number has one owner, by name. | Data | Human rule | |
| Revenue, cost split into fixed and variable, and cash position are on the scorecard. | Data | Human rule | |
| Numbers are pulled from the source systems at a fixed time each week, not typed in. | Data | AI can pull | |
| Every miss against goal is flagged automatically, with one line on what changed in the data. | Data | AI can flag | |
| For the first month, every automated number is checked against a manual count. | Data | Validation | |
| The explanation of why a number moved comes from its owner, not from a tool. | Data | Stays human |
Which processes
| True | Line | Component | Where AI sits |
|---|---|---|---|
| Core processes are written as short checklists with one owner each. | Process | Human rule | |
| New walkthrough recordings are turned into draft checklists for the owner to approve. | Process | AI can draft | |
| People can ask how we do something and get the answer from the manual, not from the founder. | Process | AI can retrieve | |
| Repeated AI work is turned into reusable skills with validation checks, not long prompts that run unchecked. | Process | Validation | |
| There is a written list of what agents may do alone and what they must escalate to a named person. | Process | Human rule | |
| Anything that leaves the building gets a human sign off first. | Process | Stays human |
Which people decisions stay human
| True | Line | Component | Where AI sits |
|---|---|---|---|
| Hiring, exits and seat changes are decided by people only. | People | Stays human | |
| Performance conversations are held by a person, never delivered by a tool. | People | Stays human | |
| Consequences for missed commitments are decided by the leadership team. | People | Stays human | |
| Promises to customers, lenders and the board are made by a named person. | People | Stays human | |
| The CEO owns AI adoption and the Integrator is fully bought in. | People | Human rule | |
| The whole leadership team is on board, and nobody is quietly sandbagging it. | People | Human rule |
The six components, and where AI sits in each
| Component | Where AI helps | What stays human |
|---|---|---|
| Vision | Drafting help on the plan | The choices in it |
| People | Never. At most it books the meeting | Every seat, every review, every exit |
| Data | Scorecard pulls and exception flags | Deciding what a red number means |
| Issues | Capture and triage | The solve |
| Process | Documentation and retrieval | Deciding the process, or excusing someone from it |
| Traction | Priority status, pre-reads, to-do capture, meeting scoring | Setting the priorities and holding people to them |
How to read your tally
- 0 to 10: rules first. Write the definitions, the owners and the validation checks before anything is automated.
- 11 to 19: ready for one AI priority. Pick the scorecard pull or the weekly pre-read, one owner, 90 days, reviewed every week.
- 20 to 27: ready to extend. Keep the checks and add the next workflow only when the first one is trusted.
The component names are the six parts of an operating system as EOS names them. The lines are our own.
Independence
Leaders ADAPT is independent and is not affiliated with, certified by or endorsed by EOS Worldwide or any other framework owner. The six component names are used only to describe the system. This checklist is our own template, and no proprietary worksheet is reproduced.
Where this fits
This checklist belongs to the guide on AI for companies running EOS, which maps AI onto every part of the system.
- AI for companies running EOSThe hub. Where AI fits in each component, and what stays human.
- The Level 10 MeetingThe best first place for AI: pre-reads, capture and scoring the meeting.
- The weekly KPI scorecardThe scorecard pull that works, and how to validate it.
- What not to delegate to AIThe full list of decisions that stay with people.
Common questions
What does the AI in your operating system checklist cover?
Four areas: which meetings AI can prepare, which numbers it can pull and flag, which processes it can document and retrieve, and which people decisions stay human. Every line maps to one of the six components of an operating system, Vision, People, Data, Issues, Process and Traction, so you can see where AI belongs and where it does not.
Where should a company running EOS start with AI?
In the Data or Traction rows. The scorecard pull and the weekly meeting pre-read are the safest first projects, because the output is easy to check against what the team already knows. Treat it as one AI priority for the quarter: one workflow, one owner, a finish line at day 90, reviewed every week.
Who should own AI in a company running an operating system?
The CEO, the same way the CEO has to own the operating system, with the Integrator fully bought in. The CEO then delegates the implementation to the whole leadership team, not only the Integrator. AI is a culture change, and one person quietly sandbagging it is enough to stall it.
What should AI never do in an operating system?
Make people decisions, solve issues, decide consequences for missed commitments, set strategic commitments, hold performance conversations or make external promises. AI can prepare every one of those moments. The decision itself stays with a person who can be held accountable for it.
Put the first AI priority in with someone who has done it
What it is: A fractional AI executive from Leaders ADAPT who installs the first AI workflow, the validation rules and the training inside the operating system you already run, with every decision kept with your managers. Who it is for: CEOs of companies running EOS or a similar system who want AI in the weekly rhythm without losing control of it.