Three episodes for the team lead or middle manager who's been told to 'roll out AI' and isn't sure what that means in practice. Why most AI rollouts fail at the people layer, not the technology layer. Why the people pushing back hardest are usually the ones with the most pattern recognition for failed rollouts in their career, and why the right move isn't to override them but to address what they're actually pattern-matching on. Conversational format, ~50 minutes total, written for someone who has to deliver the rollout next quarter and wants the playbook before they walk into the team meeting.

The playbook for the change manager — not the AI strategist — running an AI rollout inside an actual organisation with actual humans. Covers the diagnostic phase (which workflows are real candidates and which are pet projects), the communication structure that makes change feel deliberate rather than imposed, the early-win design that builds momentum without inflating expectations, and the accountability rhythm that keeps the project moving when the initial energy fades. Built from the failure modes of real rollouts — not the success stories everyone's already heard. Aimed at the operations leader, COO, or transformation lead who's owning the rollout and needs the structure to do it well.

A pre-flight check for every AI-rollout communication before it goes out. Covers the five framing failures that derail adoption: leading with the technology instead of the work it changes, vague timelines that read as evasion, missing answers to the questions the team will actually ask (job security, learning curve, override authority), executive language for an audience that needs operational language, and the absence of a feedback loop. Run it on the all-hands deck, the email, and the manager FAQ. Most rollouts have one of these wrong; correcting it before the message ships reduces the resistance you'd otherwise spend the next quarter clearing.

A four-phase rollout structure for AI adoption that prioritises sequence over speed. Phase 1: foundation (technical readiness, governance, the boring infrastructure work). Phase 2: pilot (one team, one workflow, one quarter — not five at once). Phase 3: expand (only after phase 2 has clean signal). Phase 4: institutionalise (so the rollout survives the original sponsor leaving). The structure prevents the most common failure mode: simultaneous pilots in five departments that all run out of energy at month three. Each phase has explicit gates — measurable criteria for moving forward, not vibes.

Twenty-one one-on-one and team conversations to run through during an AI rollout. Each is short, specific, and addresses a real concern: the engineer worried about pace, the senior who's seen three failed transformation projects, the team member who already uses AI on their own and resents it being institutionalised, the skeptic asking the question everyone else is thinking. Each entry has a suggested opening, the underlying concern to actually address, and what an honest answer looks like. Not scripts — frameworks. Used in the right order across the first ninety days, they turn the rollout from imposed to negotiated.

Seven patterns that mean an AI initiative is dying — usually months before anyone says so out loud. Performative compliance ('we tried it, didn't work for our use case'), shadow workflows (the team uses AI but doesn't admit it), the lone champion model (one person carries the project, then they leave), executive sponsorship that's loud but doesn't reroute resources, the absence of a use case the team chose themselves, success metrics defined by the technology team rather than the affected team, and the missing budget for ongoing iteration. Each gets a specific intervention; spotted early, all seven are recoverable.

Seven email lessons for the leader or middle manager actively running into AI resistance and looking for moves that work. Not 'overcome objections' in the sales sense — the resistance is usually pointing at something real, and the structured response is to engage with it rather than route around it. By session four the recipient has identified the specific resistance pattern they're facing (it's almost always one of three) and has a sequence of conversations queued up. Sessions five through seven build the longer-term cultural moves that make the resistance not recur on the next initiative.

The prompts a transformation lead actually uses across an AI rollout: the readiness assessment (synthesised from team interviews into a clear go/no-go), the rollout communication drafter (with the framing checks built in), the resistance pattern analyser (paste the pushback you're getting; get back the structural read of what's underneath it), the early-wins identifier (which workflows produce visible results in the first 30 days). Designed to be used inside Claude or ChatGPT — paste the prompt with your context, work through the structured output, the tool generates the artifact you'd otherwise build by hand.