adopt-ai-train-the-trainer git:20261011.dce569cA

Immutable. This exact content never changes and is served at /api/v1/blob/983cdf09f7c1f72b.

---
name: adopt-ai-train-the-trainer
description: Prepares internal trainers to deliver AI sessions, with delivery learning points, a teach-back plan, four practice rounds that rise in difficulty, notes per unit and a readiness conversation with no score. Use for "run adopt-ai-train-the-trainer", "AI train the trainer", "prepare internal AI trainers", "train-the-trainer for AI sessions", "teach-back plan for AI workshops", "our trainers are barely ahead of the room", "what if the tool is wrong in front of the group", "scale AI training with internal trainers", part of the Claude Guide for AI Adoption Pack by Polar Bear.
---
# AI Train-the-Trainer Kit
## When To Use
One person cannot teach everyone and the internal trainers are a few weeks ahead of the room. The session exists; the people who will deliver it have seen it once. It answers: how does each trainer practise the hard moments of an AI session before real learners see them, and how do we agree together that they are ready?

## When Not To Use
If the session itself is not designed yet, use Hands-On AI Session Plan first: this kit prepares the people who deliver, it does not design what they deliver. If you want colleagues who help peers between sessions without teaching, use AI Champions Network.

## Inputs
- The session plan and materials for each unit, with their what-if notes.
- Who the trainers are (role, delivery experience, how long they have used the tool), the first delivery date, the prepared answer on jobs and who owns it, and what changed in the tool since the materials were written.
If you have none of this, I start from the unit list and the trainers' roles and mark the output as a first draft.

## Approach
This is behaviour modelling applied to facilitators, from the meta-analysis by Taylor, Russ-Eft and Chan (2005): learning points, a model, practice, then feedback against the same points. Here the learning points are about delivery, and the hard part of delivering AI sessions is not the content: it is the tool being wrong live and the question nobody briefed you on. The failure this prevents: a walkthrough of the slides and good luck, after which a new trainer meets a wrong answer on the big screen, talks past it, and teaches the room that checking is optional.

## Workflow
1. Ask at most three questions: how many trainers and how experienced, which unit is hardest to deliver, and whether a co-facilitator is available for first deliveries.
2. Write four to six delivery learning points for AI sessions as short rules, for example "when the tool is wrong live, stop and check it in front of the room". If a point cannot be seen in a delivered segment, rewrite it.
3. Plan the model: an experienced facilitator runs one segment and the group names where each point showed. Pick a segment with a real miss in it, not the smoothest one. Then plan the teach-back: each trainer delivers a segment (length is a placeholder you set) and gets feedback from peers against the same points, as what worked and one thing to try.
4. Build four practice rounds in this order: a clean run; the tool gives a wrong answer live; a sceptic with a fair objection; the question about jobs, then a question you cannot answer. For jobs, the trainer rehearses handing the question to its owner. For the unknown: "I do not know, here is who does, here is when you will hear".
5. Write notes per unit: what trips trainers up, what to stress, where timing slips, what changed in the tool since the materials were written.
6. Write the readiness conversation, with no score: the trainer's own view, the support they want, and the first delivery agreed together (solo, with a co-facilitator, or after another round).

## Output Format
```markdown
# AI Train-the-Trainer Kit
Programme: [name] | Trainers: [number, roles] | First delivery: [date] | Kit owner: [role]
## Delivery learning points
1. [short rule] 2. [short rule] 3. [short rule] 4. [short rule]
## Model and teach-back plan
| Slot | Segment | Length | Feedback against | Feedback form |
|---|---|---|---|---|
| Model: [experienced facilitator] | [segment] | [you set] | [learning points] | The group names where each point showed |
| [slot] | [segment] | [you set] | [learning points] | What worked, one thing to try |

## Practice rounds
| Round | What is added | What to practise |
|---|---|---|
| 1 | Nothing: a clean run | [timing, handing the task to the room] |
| 2 | The tool gives a wrong answer live | [stop, check against the source in front of the room] |
| 3 | A sceptic with a fair objection | [take it seriously, test it on a real task] |
| 4 | The question about jobs, then a question you cannot answer | [hand over to [owner]; "I do not know, here is who does, here is when you will hear"] |

## Notes per unit
| Unit | What trips trainers up | What to stress | Where timing slips | What changed in the tool |
|---|---|---|---|---|
| [unit] | [note] | [note] | [note] | [change and date, or none known] |

## Readiness conversation (no score)
Your view of delivering [unit]: [their words] | Support you want: [support] | First delivery, agreed together: [solo, with a co-facilitator, or after another round]
## Decision
[Enablement lead] and each trainer agree the first-delivery arrangement by [date]. [Kit owner] updates the notes per unit after the first deliveries, by [date].
```

## Done When
- Every trainer delivers at least one segment and gets feedback against the learning points.
- The four practice rounds appear in order, and each unit has notes including what changed in the tool.
- The jobs question has a named owner the trainer hands it to, and the readiness section has no score, rating or pass mark anywhere.

## Quality Bar
- Feedback describes delivery against the learning points, never the trainer as a person. No certification, scorecard or comparison between trainers.
- A trainer saying "I do not know" with a route to the answer counts as good delivery.
- Lengths are placeholders you set, and a trainer may ask for another round or a co-facilitator without giving a reason.
- Readiness is a conversation, never a score or a ranking of trainers.

## Next
Run adopt-manager-ai-kit (Manager AI Kit) to equip the managers the learners go back to.

## About the makers
This pack is made by Polar Bear, a consultancy built by ex-McKinsey founders with a dream to make AI work for People, not instead of them. We help our clients build people systems and AI-first ways of working, and we run our own company on Claude. If your team has outgrown the self-serve version, message Pauline (linkedin.com/in/paulinebertry).