Before you start
Most AI efforts stop at “licences bought, people trained”. This workbook is for the part after that: setting a goal, finding the right work to change, deciding how much AI each task gets, turning the change into value, and checking it against the goal. Then doing it again.
AMPLE in one line
- Audit: set the goal, then see how the work really happens
- Model: decide how much AI each task gets
- Prime: work out what you need, then set it up
- Leverage: turn the change into value
- Elevate: measure against your goals, then go round again
Rules of thumb
- Start from the outcome, not the tool. Set the goal first: what do I want more of?
- Work at task level, not job or team level. One task is fine as a start.
- Start small: one goal, one workflow (yours or your team’s), one to three tasks, one thing to add.
- Freed time is the point. Point it at the goal, or it turns into more of the same work.
- Measure against the goal, not hours saved.
Choose your path
| Path | |||
|---|---|---|---|
| Who | Anyone who uses AI at work | A team lead and their team | A sponsor, an AI owner, and one pilot team first |
| Time | About 2 hours, then a 4-week check | 4 short sessions over 4 weeks, then a check | One full round with a pilot team (about 6 weeks), then expand |
| Start with | Your own week | One workflow the team shares | One team, one workflow, before anything company-wide |
| Use these pages | All | All | All, plus the organisation notes |
Organisation notesRun one complete round with a pilot team before rolling out. Name a sponsor who protects the team’s time, and one person who owns the data rules and tool access. Expand only what passed the four-week check.
Audit: set the goal, then see how the work really happens
GoalOne to three goals, and a task list showing where effort goes.
Goal
(the thing you make or do),
times a
It involves
One task is fine. Most tasks turn out to be a few steps. List them below.
Break it into the steps you actually do, e.g. collect the numbers, paste them into the template, write the summary, chase approval.
| # | Task | Who | How often | How long | AI used today?which tool | Low-value?copying, chasing, reformatting, re-checking |
|---|
Team / organisationFill this with the people who do the work, not about them. They know which tasks are wasted.
Model: decide how much AI each task gets
GoalA short list of changes, each with an answer, an outcome, a starting number and an owner.
Task finder questions. Answer for your own work first, then with your team if you’re running it as one.
- What do I do every week that I put off?
- What do I copy, retype or reformat from one place to another?
- What do I write from scratch that follows the same shape every time?
- What do I check by hand that a simple rule could check?
- What do I wait on, when the information already exists somewhere?
- What would I do more of if I had the time?→ keep it human
- What do I skip, or wait on someone else for, because I don’t have the skill?→ add
- Where would a working prototype, a better visual or a clearer story change the outcome?→ add
Human = you do it, AI stays out · Augmented = you and AI together, you in control · Automated = AI does it, you check
| # | Task | How often, how long | What’s it for? | Stop / Human / Augmented / Automated | Why, and what’s the risk? | If this was taken care of, I could focus on… |
|---|
Before handing a task to AI, ask: Why should AI do it? Is the risk worth it? Should a person stay in control? How much you do yourself, with AI or hand off changes with every task.
| Something I can’t do yet | Where it would change the result | First small version I’ll make |
|---|
Pick one to three tasks (frequent, time-consuming, low risk if AI gets it wrong) and one thing to add.
Write each outcome in one sentence, with today’s number.
1By goes from (today) to .
2By goes from (today) to .
3By goes from (today) to .
e.g. By the end of the month, Friday updates go from 2.5 hours (today) to 1 hour.
TeamThe team lead decides, and the team fills in the task finder together.
Prime: work out what you need, then set it up
GoalTools, connections, context, rules and checks for the chosen tasks, nothing more.
Answer for the tasks you chose. Tick each one once it’s set up, nothing more.
| Task | What must be true before it’s used | Who checks (can be you) |
|---|
Organisation notesThe data rules come from whoever owns data and security. Keep them to one page people will actually read.
Leverage: turn the change into value
GoalFreed time and new skills going into the goal every week, and a record of what it added.
Practise on real work, not features
Spend as long on where to use it as on how.
| Prompt | Week 1 | Week 2 | Week 3 | Week 4 |
|---|---|---|---|---|
| 1What I made or used AI for | ||||
| 2Where the freed time actually went | ||||
| 3What value it added | ||||
| 4What went wrong or I had to fix | ||||
| 5One thing I’ll try next week |
TeamThe team lead runs the weekly check-in, and everyone finishes something that runs.
Elevate: measure against your goals, then go round again
GoalEvidence of what changed against the goal, and the list for the next round.
| Goalfrom Audit | What I’d see if it worked | What happened | Grow / fix / drop |
|---|
| Opportunity | Measure | Today’s numberfrom Model | Number now | Where the freed time actually wentvs plan | Grow / fix / drop |
|---|
- Good measures
- the goals from Audit first · cycle time · rework · still used a month later · where the freed time went · wins or revenue where honestly linked.
- Weakest measure
- hours saved. Saved time only counts once it turns into something.
Next round
TeamThe team lead runs the four-week check, with whoever owns the numbers.
Your one-page AMPLE plan
- AAudit
- MModel
- PPrime
- LLeverage
- EElevate