# The Claude Rollout Checklist

The 10 decisions to make before deploying Claude to a team — so the rollout is a
deliberate plan instead of a pile of individual seats nobody was trained on.

## 1. Who gets a seat, and why

Don't default to "everyone." Start with the roles where AI-assisted work has the
clearest, most immediate payoff, and expand from there once adoption is real, not
theoretical.

## 2. What data can and can't touch Claude

Define this before day one, not after an incident. Classify your data (public,
internal, confidential, regulated) and decide explicitly what's in-scope for Claude
to see, especially if you're connecting it to internal systems via MCP.

## 3. Admin controls and SSO

Decide who administers the account, how provisioning and de-provisioning works when
people join or leave, and whether SSO is required before rollout (it should be, for
anything beyond a small pilot).

## 4. Usage policy

A short, plain-language policy on acceptable use — what Claude should and shouldn't
be used for, how to handle output that touches customer data, and who to ask when
someone's unsure. Long legal documents don't get read; short and specific ones do.

## 5. The first three workflows to automate

Pick three concrete, recurring workflows to target first — not "use AI more." Good
first candidates are high-frequency, well-defined tasks with a clear "done" state:
meeting recaps, first-draft reports, research synthesis.

## 6. Governance and approval boundaries

For anything Claude can take action on (via Cowork, agents, or MCP connectors),
decide what requires human review before it goes out, and what's safe to run
autonomously. Start narrower than feels necessary; widen the boundary as trust is
earned, not assumed.

## 7. Training plan

Decide who trains the team, on what cadence, and whether it's generic or built around
your team's actual tools and workflows (it should be the latter — generic training is
the single biggest reason "AI training" doesn't change Monday morning behavior).

## 8. How you'll measure adoption

Pick 2-3 numbers you'll actually track — seats activated, workflows automated, hours
saved, whatever is real for your org — before rollout starts. Without a baseline,
"is this working" becomes a vibes-based question three months in.

## 9. A pilot group before a company-wide push

Roll out to one team or department first. Use what you learn — what confused people,
what broke, what workflow actually stuck — to fix the plan before scaling it to
everyone else.

## 10. Who owns this after the rollout

Deployment isn't a one-time event. Name an internal owner (or team) responsible for
onboarding new hires, updating the usage policy, and expanding to new workflows once
the initial rollout is done. Without an owner, adoption plateaus the moment the
original project team moves on.

---

*Provided by Timbaline AI — Claude implementation, consulting & training for
enterprises. This checklist is provided as a free starting point; every real rollout
we run is custom-scoped to the specific team and workflows involved.*