AI Guides › Playbooks
By Nigel Guy · 7 min read
Most people audit their working life by listing what they have been busy with. A long list of jobs, projects and favours feels like progress, so the audit ends with a warm feeling and no decisions. Busy is a record of time spent. It tells you nothing about which of your skills other people would pay for, or which hours are quietly earning you nothing.
The rule: audit what you already hold, not what you might become, and make every answer end in a dated action or a named thing you will stop.
The Four-Lens Experience Audit points an AI model at your own history and forces four different readings of it. Each lens has a different question and a different output. If a lens produces no action, that lens is not finished.
| Lens | The question it asks | What you walk away with |
|---|---|---|
| 1. Skill to offer | Which single skill could become something sellable soon? | One small offer, with a format and a rough scope |
| 2. Knowledge to shortcut | What took you years to learn that someone else would pay to skip? | One topic, and the mistakes it would save a buyer from |
| 3. Repeat to system | What do you do by hand again and again? | One thing to turn into a template, checklist or document |
| 4. The honest no | Which commitment returns the least for the time it takes? | One thing to stop or shrink, and what stopping costs |
Lenses 1 to 3 find things to build. Lens 4 finds the hours to build them in. Skipping it is the usual way an audit becomes another list of things you will not get round to.
Before you open any AI tool, spend fifteen minutes writing plain notes: jobs and what you actually did in them, projects you finished, things colleagues, friends or customers keep asking you for, and tasks you do every week. Do not polish. Messy notes with specifics ("I rebuilt the supplier onboarding pack three times") beat a tidy CV summary, because the model can only find leverage in details you give it.
Your history is personal data and may contain other people's. Before pasting, remove client names, employer-confidential figures and anything you are bound by contract not to share. Anthropic's privacy help says consumer chats can be used to improve Claude if you allow it, controlled by a setting called Model Improvement in Privacy Settings, and that Incognito chats are not used even when it is on. Check your own setting rather than assuming. At time of writing, Claude's free plan is enough to run this audit; Pro is listed at $20 a month (about £15 at time of writing, check the £ price at checkout), and adds Projects, which are handy if you want to keep the audit and its follow-ups together.
Fill in your notes and your constraints. The prompt asks for questions before answers, which stops the model inventing a career for you.
You are a candid career-and-business analyst. Your job is to find leverage in the experience I already have, not to encourage me.
Context
- My notes on what I have done and know: [PASTE_YOUR_NOTES]
- Hours I can realistically spend on new work each week: [HOURS_PER_WEEK]
- My constraints (employer rules, money, location, anything off-limits): [CONSTRAINTS]
- Who I would most like to help, if I have an idea: [TARGET_PERSON_OR_NONE]
What a good result looks like
Four short sections, one per lens below. Every section ends with one concrete action I could start within seven days. Evidence from my notes is quoted or clearly referenced; anything you are assuming is labelled "assumption".
Lenses
1. Skill to offer: pick the one skill in my notes that could become a small paid offer (template, guide, short workshop or service). Say what the offer is, who it is for, and why this skill over the others.
2. Knowledge to shortcut: identify what I learned slowly that a newcomer would value skipping. List the three costliest mistakes a newcomer would make.
3. Repeat to system: find work I do by hand repeatedly. Propose the form it should take (checklist, template, document or small automation) and what I would put in it first.
4. The honest no: from the commitments in my notes, name the one with the lowest return for its time. State what stopping it would cost me, including awkward conversations.
Rules
- If my notes are too thin for any lens, ask me up to five specific questions first and wait for answers. Do not guess.
- Do not flatter me. If a lens has no good answer, say so.
- Do not invent market sizes, prices or demand figures. Where price or demand matters, say what I should check and where.
Before answering, check that each section cites my notes, ends in an action, and contains no unlabelled assumption.
Fill in: your notes, weekly hours, constraints, and an optional target reader.
A first answer is a draft. Send this second prompt to test the strongest-looking idea before you spend a weekend on it.
Take the lens-1 offer from your audit. Act as a sceptical buyer in [TARGET_BUYER]'s position.
1. Give the three most likely reasons I would not buy this.
2. Name the cheapest way I could test real interest in [TIMEFRAME] without building anything first.
3. Say what result would count as a pass and what would count as a fail, as numbers I choose in advance. Ask me for those numbers rather than proposing them.
Keep it under 250 words. Flag any claim that depends on facts you cannot know about my market.
Fill in: the buyer you have in mind and how long you will give the test.
Copy the four rows below into a note and complete them in your own words, not the model's. Writing the action yourself is the check that you believe it.
| Lens | Chosen item | Action | Date | Review date |
|---|---|---|---|---|
| Skill to offer | ||||
| Knowledge to shortcut | ||||
| Repeat to system | ||||
| The honest no |
Put the review date four weeks out. At review, mark each row done, changed or dropped. Dropped is a legitimate result.
Imagine an operations coordinator at a small logistics firm. Her notes mention rebuilding the new-starter pack three times, answering the same rota questions from friends in other firms, and a fortnightly committee she attends but rarely contributes to.
None of those outputs is evidence of demand. They are candidates for a small test, which is exactly what Step 4 is for.