Skip to content
Practical AI

AI Rollout Support

Rolling out Claude, Copilot or Codex? We design the pilot, train champions, prepare the guidance and track real use, so the licences turn into better work instead of an unused icon.

Week by week, teams move from pilot to live on the eagle’s rollout board. A stalled team gets a champion and a guide, catches up, and the assistant goes company-wide.

Why rollouts stall.

Use peaks in week one, then quietly fades. It’s rarely the tool. Pick a reason to see what fixes it.

Illustrative sketch: licences assigned stay flat, while the number of people using the tool for real work peaks in week one and then fades to a low level. The gap between the two is paid for but not used.Licences assignedPaid for, not usedWeek oneUsed for real workLaunchTwo months on
Illustrative sketch, not measured data.

No clear use

“I’ve got the button. I’m just not sure what I’d use it for.”

What fixes itWe name two or three real tasks for each team before launch.

A rollout, week by week.

Five stages, each with a checkpoint to pass before the next.

Before launch

Get ready

Settle what the tool can see, what people may put in and who goes first.

Checkpoint to move on

The worst oversharing is fixed, the policy is published and every pilot team has named tasks.

Typical weeks for a pilot of a few teams. We set the pace with you.

Claude, Copilot or Codex?

They overlap, but each has a natural home. Pick one.

Long documents, analysis and files such as spreadsheets.

  • Review a contract
  • Summarise an export

Before switching onAgree which files may go in and which connectors are on.

Using more than one? Our AI Tool Selection & Secure Use work tests the mix. Copilot here means Microsoft 365 Copilot; we check current versions when we plan.

What we measure.

Whether the work changed, not whether licences were handed out. Pick a measure to see it on the dashboard.

  • Licences assignedShows what you paid for, not what changed.
Finance pilot, week 5Template, illustrative

Tasks done with the tool: Which named tasks, in which teams? Not how many people logged in.

TaskDone with the toolQuality sampleOpen blockers
Month-end commentaryRoutinelyPassedNone
Supplier repliesSome of the teamPassed after a fixNo shared mailbox access
Board pack draftNot yetNot sampledUnclear what can go in

ConfidenceMost know what can go in. Fewer know how to check a figure.

RecommendationGo for finance. Hold the board pack until the policy question is answered.

What you receive.

Written for your tools and teams, and yours to keep running after we step back.

  1. Before launchA rollout plan

    Stages, teams, dates and checkpoints.

  2. Week 1A champion pack

    Likely questions, answers and exercises.

  3. Week 1A dashboard template

    Ready to fill in for each wave.

  4. From week 3Role guides

    What to use it for, and how to check.

  5. Week 5A go/no-go recommendation

    Evidence and reasons, even if it’s no.

Why have a security team do this?

“We can’t roll this out until security signs it off.”

We are the security sign-off.

How we do it

The AI Readiness & Data Exposure Review and AI Use Policy & Governance happen before the pilot, so the questions your security team would ask already have written answers.

  • Data questions firstWhat could leak is settled before switch-on.
  • Access set up properlyWe configure and test it ourselves.
  • Verified with your peopleSessions in plain English, not slide decks.
Planning a rollout?

Tell us the tool, the number of people and the start date.

Discuss AI Rollout Support