The capability arrives at rollout with the exceptions already written in, the review steps where the reviewers want them and a team that asked for half of what it contains. The billing coordinator recognizes the rule about disputed invoices because it was theirs. The account executive finds the proposal check they suggested. The team uses it from the first morning, and it keeps getting better because the people closest to the work know where to push next. This is how you 10X what your team can do: carry the judgment of the people who do the work into every run.
A plan drawn up only in a leadership meeting meets the work for the first time at rollout. The exceptions surface then, a review step lands in the wrong place and the team learns what was decided about its own job. Building the plan with the people who do the work brings all of that forward to week one.
Their method becomes the standard.
Capturing how people work increases their influence over how the work develops. The method a colleague describes becomes the standard the capability follows, and the awkward cases they remember become its tests. A person who walks through a difficult case is writing rules the whole team will run.
They do not need to document everything first. A walkthrough of the last few real cases gives a better start than a blank template, and we do the writing.
Five questions to ask the people who do the work.
Ask them in a short conversation, one piece of work at a time, with a recent real case on the table.
- “Walk me through the last time you did this.”You hear the real sequence, the inputs actually used and the steps the written procedure skips.
- “What made the last awkward one awkward?”You find the exceptions and the judgment behind each one. Your SOP describes the happy path shows how to write them down, and SOP exception capture turns them into rules and tests.
- “What do you check before you hand it on, and to whom?”You see the handoffs and the review standard the next role depends on.
- “What do you wait for, retype or chase?”You locate the friction, which is where prepared work helps most.
- “If you could change one thing about how this runs, what would it be?”You collect the improvement ideas, and the suggestions log has its first entries.
A feedback process people can see.
Suggestions keep coming after the pilot starts, for as long as people can see what happens to each one.
Feedback process for an AI pilot
- Anyone who runs the pilot or receives its output adds a suggestion to the shared log, in their own words, with the case that prompted it.
- The pilot owner reads every new suggestion at the weekly check.
- Each suggestion gets one status: adopted into the pilot, queued for a later version, or declined with the reason.
- The owner tells the person who made it, and the log shows the decision to everyone.
- An adopted suggestion becomes a rule, a test case or a change to the job aid, and the log records which.
When the pilot changes how someone works, talk about it in their next one-to-one. Manager feedback preparation helps a manager arrive with specific observations and the colleague’s own suggestions in front of them.
Show people what happened to their suggestions.
The log is the part people read. Every suggestion is on it, with the role that made it and what became of it. This one comes from a pilot of client-status update preparation.
Illustrative example with invented data. An invented 60-person engineering firm, six weeks into its pilot.
| Suggestion | From | Status | What happened |
|---|---|---|---|
| Take approved dates from the change log, never from meeting notes. | Project coordinator | Adopted into the pilot | Now a rule and a test case. |
| Flag any project where the client has not replied in two weeks. | Project lead | Adopted into the pilot | Now an exception rule: the draft flags the project for the project lead. |
| Show unbilled change orders in the draft. | Billing coordinator | Queued for a later version | Needs the billing export; planned for the next version. |
| Draft each update in the client’s preferred format. | Client manager | Queued for a later version | Waits on the formats, which client managers are collecting. |
| Send the updates automatically every Friday. | Office manager | Declined, with the reason | Sending stays with the client manager, who reviews each update first. |
| Add a risk score to every project. | Principal | Declined, with the reason | No agreed definition of risk yet. Project leads will draft one. |
A declined suggestion with its reason shows it was weighed, and the principal’s idea gets the same treatment as the coordinator’s. The risk score also shows how a declined idea becomes the next piece of work: an agreed definition of project risk is now on the plan’s list, owned by the project leads.
Your platforms, your pace, your people.
Fully Powered works across ChatGPT, Claude, Copilot and Gemini, in the workspace your organization already uses. The owner sets the pace, and the team’s suggestions shape it. The organization keeps control of how the work gets done, and the people who do it have a direct say in how it changes. In the first 90 days of an AI adoption plan, these conversations happen in weeks 1 and 2, before the pilot is chosen.
Bring one piece of work and the roles of the people who do it.