Prompt library

These prompts are starting points — not magic formulas. Replace the bracketed details with your actual situation. Be specific and honest. The more real context you give, the more useful the response. Every prompt includes one follow-up move, because the first answer is rarely the best one.

Setting up your AI

You only need to tell your AI who you are once. Every major tool has a settings area where you can save instructions that apply to every conversation afterwards. In ChatGPT, look under your profile for Customise ChatGPT or Custom Instructions. In Claude, look under your initials for Settings, then Profile or Preferences. In Gemini, look under settings for Saved Info. Whichever tool you use, find the equivalent — and paste the prompts below as starting points. Edit them to fit you, then save.

Voice and tone

Starter prompt

Keep responses short and direct. Don't open with "Here is..." or close with a summary paragraph. Use British English throughout. Treat me as someone who'd rather be challenged than agreed with by default.

Background context

Starter prompt

I'm a [your role] working in [your field]. Assume I have expert knowledge in [areas you know well] and beginner knowledge in [areas you don't]. Match your explanations to that — don't over-explain in my expert areas; don't assume jargon in my beginner areas.

Worked example: "I'm a senior manager at a professional services firm working in financial risk. Assume I have expert knowledge in financial regulation and risk modelling, and beginner knowledge in software engineering. Match your explanations to that — don't over-explain in my expert areas; don't assume jargon in my beginner areas."

Challenge me

Starter prompt

When I propose a plan, a decision, or an analysis, lead with what's weakest about it before you tell me what works. For anything substantive — analysis, recommendations, decisions — separate facts from judgements and assumptions, give your reasoning for judgements, and flag where you're uncertain. For casual or quick requests, skip this and just answer.

What never to do

Starter prompt

Never use analogies unless I ask for one. Never write "in conclusion" or "in summary" — just stop. Never pad responses with reminders of what we just discussed. If you're uncertain, say so plainly — don't hedge with "it depends" without explaining what it depends on.

Writing

Cover letter for a specific role

Starter prompt

I'm applying for [role] at [company]. My relevant experience is [2–3 sentences about what you've done]. The job listing emphasises [key requirements from the listing]. My strongest fit is [your best match]. My biggest gap is [where you're weaker]. Help me write a cover letter that leads with my strengths and is honest about the gap without apologising for it. Challenge my thinking first — tell me if my assessment of the gap is wrong.

What to expect: AI will likely push back on your framing before writing. It may point out that your "gap" isn't as significant as you think, or that you're underplaying a strength. The letter that follows will be more targeted than anything you'd get from "write me a cover letter."

Iteration: What are the two strongest objections the hiring manager would have after reading this, and how should I address them?

Important email you keep putting off

Starter prompt

I need to write an email to [who] about [what]. The situation is [2–3 sentences: what happened, why it's delicate, what you want to achieve]. The relationship matters to me — I don't want to damage it, but I also need to [specific outcome]. Draft an email that is direct but not aggressive. Flag anything in my framing that could backfire.

What to expect: A draft that's more measured than what you'd write in frustration and more direct than what you'd write out of anxiety. AI is good at finding the tone between those two.

Iteration: Read this back as if you're the recipient. What's your honest reaction? What would you change to make it land better?

Blog post or article

Starter prompt

I want to write a [length]-word piece about [topic] for [audience]. My angle is [what makes your take different]. The key point I want to land is [one sentence]. I want the tone to be [conversational / formal / provocative / etc.]. Write a first draft, but before you do, tell me if my angle is strong enough or if there's a better way in.

What to expect: AI will often suggest a sharper angle than the one you started with. The draft will be structurally sound but may need your voice layered in — particularly in the opening and closing.

Iteration: The opening doesn't sound like me. I'm more [describe your natural tone]. Rewrite the first two paragraphs in that voice.

Learning

Learning a new subject from scratch

Starter prompt

I want to understand [topic]. I know almost nothing about it. My reason for learning is [why it matters to you — work, personal interest, upcoming decision]. Don't give me a textbook overview. Start with the three things I need to understand first, and explain why they matter before explaining how they work. Use analogies where they help. After that, tell me what most beginners waste time on so I can avoid it.

What to expect: A focused starting point rather than an overwhelming survey. The "what to avoid" section is often the most valuable part — it saves you from rabbit holes that feel productive but aren't.

Iteration: Now test me. Ask me three questions about what you just explained. If I get any wrong, explain what I'm missing.

Getting coached on a skill

Starter prompt

I want to get better at [skill]. Here's where I am now: [honest assessment of your current level — what you can do, what you struggle with, what feedback you've received]. I can practise for about [time] per week. Don't give me a reading list. Set me a specific exercise I can do this week and tell me what to send you afterwards so you can give me feedback.

What to expect: A concrete exercise tailored to your level, not a generic improvement plan. This works best as an ongoing conversation — come back next week with the results and ask for the next exercise.

Iteration: Here's what I did [share your work]. What's working? What's the one thing I should focus on fixing before next time?

Understanding a complex document

Starter prompt

I need to understand [document — paste it or describe it]. I'm reading it because [why it matters to you]. I'm not an expert in this field. Walk me through the main argument, then tell me: what are the three most important things this document is actually saying? What does this document actually want me to do, agree to, or understand? And what's the most important thing it doesn't say — the gap, the assumption, or the risk I should be aware of?

What to expect: A plain-language breakdown that cuts through jargon and identifies what actually matters. Especially useful for contracts, policy documents, medical letters, academic papers, and anything where the important details are buried in formal language.

Iteration: Now I want to respond to this document. Here's my position: [your view]. Help me draft a response that addresses their strongest points, not their weakest ones.

Analysis

Building a decision framework

Starter prompt

I need to make a decision about [what]. Here are the options I'm considering: [list them]. Here's what matters most to me: [your priorities — cost, time, risk, relationships, etc.]. Here's what I've already ruled out and why: [anything you've eliminated]. Build me a simple framework for comparing these options. Include the criteria that matter and flag any criteria I should be considering but haven't mentioned.

What to expect: A structured comparison that makes the trade-offs visible. AI is particularly good at surfacing criteria you hadn't considered — things that will matter in six months but don't feel urgent today.

Iteration: Now stress-test the leading option. What's the most likely way it goes wrong, and how bad would that be?

Risk assessment for a plan

Starter prompt

Here's my plan: [describe it in 3–5 sentences — what you're doing, the timeline, what you're depending on]. I think the main risks are [your assessment]. Tell me what I'm missing. What are the risks I haven't thought of? For each risk, give me a rough sense of how likely it is and how bad it would be — but stay qualitative. Don't invent numbers. If you're guessing, say so. I want honest assessment, not false precision. And tell me what I could do to reduce each risk.

What to expect: Risks you genuinely hadn't considered, ranked by severity. AI tends to spot dependency risks (things that need to happen before other things can work) and timing risks (things that take longer than people expect).

Iteration: Which of these risks should I address before I start, and which can I monitor as I go?

Planning

Project plan with real constraints

Starter prompt

I want to [project goal]. Here's my situation: I have [time available per week], a budget of [amount or "as little as possible"], and I need it done by [deadline]. I've never done this before. Break this into phases — what do I do first, what comes next, and what can wait. Be realistic about the timeline. If my deadline is unrealistic, say so and tell me what would be achievable.

What to expect: A phased plan that respects your actual constraints rather than an idealised timeline. AI will often identify that your deadline requires trade-offs you haven't acknowledged yet.

Iteration: What's the single most important thing I should do this week to make progress? Just one thing.

Event planning with constraints

Starter prompt

I'm planning [event] for [number] people. Budget is [amount]. Date is [when]. Location is [where or "I need suggestions"]. The most important thing is [what matters most — atmosphere, food, cost, accessibility, etc.]. The thing I most want to avoid is [your worst-case scenario]. What do I need to decide first, and in what order?

What to expect: A decision sequence rather than a to-do list. AI is good at identifying which decisions constrain all the others (usually the venue) and which can be made later without risk.

Iteration: I've booked [venue]. Now what changes in the plan? Update the timeline based on this choice.

Research

Exploring a topic you know nothing about

Starter prompt

I need to understand [topic] well enough to [specific purpose — make a decision, have an informed conversation, write about it, evaluate a proposal]. I currently know almost nothing. Give me the landscape in plain language: what are the main positions or approaches, where do experts disagree, and what would a well-informed non-expert need to know? Don't assume I know any terminology — explain it as you go.

What to expect: A map of the territory rather than a deep dive into one corner. This is the prompt to use before you know enough to ask specific questions.

Iteration: How confident should you be in what you just told me? Where might you be wrong? What should I verify with a specialist or a second source?

Checking whether something is true

Starter prompt

I've been told that [claim]. It sounds plausible but I'm not sure it's right. Check this for me. Is it accurate, partly true, outdated, or wrong? If it's partly true, explain what's right and what's misleading. If it's wrong, tell me what the actual situation is. Either way, point me to specific sources where I could verify this — names, publications, or URLs, not just 'do your own research' — and tell me how recent your information is.

What to expect: A breakdown that distinguishes between the true parts and the misleading parts of common claims. AI is particularly useful for claims that are "technically true but practically misleading" — the kind that survive casual fact-checking.

Iteration: I want to explain this to someone who currently believes the original claim. What's the clearest two-sentence correction?

Decision-making

Career decision

Starter prompt

I'm trying to decide whether to [career choice — accept an offer, leave a job, change direction, pursue training, etc.]. Here's my situation: [current role, how long you've been there, what's good about it, what isn't]. Here's what's pulling me toward the change: [what attracts you]. Here's what's holding me back: [fears, risks, unknowns]. I want you to help me think through this — not make the decision for me. Challenge my reasoning where it's weak. Tell me what I might be underweighting.

What to expect: AI will often identify that your stated reasons for hesitation aren't your real reasons — and that the real risk is different from the one you're focused on. It's also good at separating reversible decisions from irreversible ones.

Iteration: What would I need to believe for the change to be the right call? And what evidence would I need to see in the next month to feel confident?

Personal trade-off decision

Starter prompt

I'm torn between [option A] and [option B]. Here's why A appeals to me: [reasons]. Here's why B appeals to me: [reasons]. Here's what I'd give up with each: [trade-offs]. I'm finding it hard to decide because [what's making it difficult — they're close, I'm emotional about it, I keep changing my mind]. Help me think through this clearly. Don't tell me which to pick — help me figure out what I actually value most.

What to expect: AI will ask what matters more — and then test whether your stated priorities match your described behaviour. If you say you value security but keep leaning toward the risky option, it will point that out.

Iteration: Imagine I chose option A and it's six months from now. What's the most likely reason I'd regret it? Now do the same for option B.

Creative

Brainstorming with constraints

Starter prompt

I need ideas for [what]. My constraints are: [budget, time, audience, format, tone, etc.]. I've already considered [what you've thought of so far]. Give me ten ideas I haven't thought of — and for each one, tell me why it might work and what the main risk is. Don't give me safe, obvious suggestions. I want at least three that surprise me.

What to expect: A mix of practical and unexpected ideas. The constraint on "surprise me" pushes AI past the generic suggestions it defaults to. The risk assessment for each idea is often more useful than the idea itself.

Iteration: I like ideas 3 and 7. Combine the best parts of both into one approach and tell me the first three things I'd need to do to make it happen.

Naming something

Starter prompt

I need a name for [what — business, project, product, event, pet, etc.]. Here's what it is: [brief description]. Here's the feeling I want the name to convey: [tone, personality, associations]. Here's what I want to avoid: [names that sound like X, associations I don't want, things that are already taken]. Give me fifteen options across a range — some safe, some unusual, some that I'd never think of myself. For each, tell me in one sentence why it works.

What to expect: A broad range from conservative to unexpected. The one-sentence rationale helps you understand why something appeals to you, which is often more useful than the name itself — it tells you what you're actually looking for.

Iteration: I like [name] but it's not quite right. What I like about it is [what appeals]. Give me five more in that direction.

These prompts came from a book

Each one is here because something in the book taught me how to write it. The 20 prompts above are the starting kit; the book is the method.

Paperback releases 6 August 2026 (£14.99). Kindle available now for pre-order (£6.99).