Short answer: This is a library of 47 copy-paste prompts for Claude Fable 5, written for established business owners who are not technical. The prompts make Fable check your live systems first (QuickBooks, Stripe, email, calendar, CRM), ground every answer in real evidence, and hold every client-facing draft for your approval. They find the money first: unpaid invoices, dropped leads, stale proposals, and underpriced work. The final section turns the winners into agents that run on a schedule without you.
Most AI prompt libraries are written for people with codebases, product teams, staging environments, and internal ops departments.
The best known of them, Every’s Claude Fable 5 prompt library, is excellent for that technical crowd. This one is for everyone else.
That is not most established business owners.
You have something very different.
You have a client roster. An inbox full of unfinished conversations. Old proposals. Invoices. QuickBooks or Stripe. A calendar. A messy shared drive. A few long-term clients who quietly carry too much of the business. And probably years of valuable context trapped across emails, contracts, spreadsheets, meeting notes, and half-finished documents.
That is where Fable 5 becomes useful. When connected properly, it can help you find money, risk, missed follow-up, hidden demand, underpriced work, operational drag, and client signals you are too busy to notice.
This library is built for owners who are not technical and do not want to become technical.
Every prompt below is written to be pasted directly into Fable. You do not need to understand how the model works. You need to give it a clear job, force it to use real evidence, and stop it from taking action without your approval.
Before you start, understand these rules.
The 4 rules that make this library work
1. Live systems come first
When Fable can access live systems, those should beat stale spreadsheets.
That means accounting tools, payment processors, email, calendar, CRM, file storage, and any other connected apps should be checked first.
If Fable cannot access a system, the prompt tells it to say exactly what report, export, file, or folder it needs. No vague "please upload your data" nonsense.
2. Evidence beats guesses
Every useful business answer should come from something real:
- an invoice
- a payment record
- a proposal
- an email thread
- a contract
- a calendar record
- a client note
- a meeting transcript
- a website page
- a past project folder
If Fable has to guess, it should mark the guess clearly.
3. Drafts stay drafts
Any prompt that creates an email, message, client note, sales follow-up, proposal language, website copy, or public-facing content ends with one rule:
HOLD until approved.
Nothing gets sent, published, deleted, moved, or changed without you approving that specific action.
4. The first run will expose a mess
That is the point.
You may find unpaid invoices, open proposals nobody followed up on, duplicate subscriptions, awkward client threads, outdated pricing, or four versions of the same contract.
Do not treat that as failure.
That mess was already costing you money. Now you can see it.
Start with Prompt 1. It takes a few minutes and makes every other prompt better.
Part 1: The First Hour
1. Show me what you can actually see
Run this first. Do not assume Fable can access everything. Make it prove what it can reach.
Take inventory of what you can actually access for my business. Create two lists. First: live connections. Check which apps and tools you can access right now through connectors, integrations, plugins, or API keys. Look specifically for email, calendar, accounting, invoices, payments, CRM, file storage, project management tools, website data, and anything else connected to this workspace. For each live connection, run only a harmless read-only check. Tell me what worked, what did not work, and what you were able to confirm. Do not describe what should be possible. Describe only what actually worked. Second: local or uploaded business files. Look for business documents, spreadsheets, exports, proposals, contracts, invoices, client folders, marketing assets, SOPs, and reports. Do not open anything that appears highly sensitive without flagging it first. Then give me a plain-English map: 1. What you can access live 2. What you can only access as files 3. What you cannot access at all 4. What I would need to connect or upload to make the next prompts work better 5. Three useful things you could do for me this week using only what is available right now Use plain business language. No technical jargon.
2. Build my business memory file
This is the highest-leverage setup prompt. It creates a reference file so Fable does not keep asking the same basic questions.
Build a business memory file for me. Do not interview me first. Use existing evidence. Read what you can access: my website, email history, calendar, accounting records, CRM, proposals, invoices, contracts, client folders, service pages, and business documents. Create a one-page business memory file that covers: 1. What this business sells 2. Who buys it 3. The main offers or services 4. Typical pricing or project size 5. Key people, roles, clients, vendors, or partners 6. Common client problems 7. How we usually talk to clients based on real sent emails 8. Any standing rules I should keep in place Mark anything uncertain as [ASSUMPTION]. Do not pretend guesses are facts. Show me the draft first and let me correct it. After I approve it, save it as BUSINESS_MEMORY.md. If this workspace already uses CLAUDE.md or another agent instruction file, update that file too.
3. Find the work worth handing off
Before you outsource to an AI agent, make it identify the work that is actually repetitive, valuable, and low-risk enough to hand off.
Read my business memory file. Then review evidence of how work actually happens in my business: sent emails, recurring documents, spreadsheets, client folders, calendar patterns, project management tasks, reports, and repeated messages. Identify the top ten tasks I should hand off to you. Rank them by likely monthly time saved and business value. For each task, tell me: 1. What the task is 2. How often it appears to happen 3. What evidence you used 4. What you would need from me once 5. What output I would get back 6. Whether it is low-risk, medium-risk, or high-risk 7. Whether it should require my approval before anything external happens Then write the full ready-to-paste prompt for the top three tasks so I can start using them today.
4. Set my red lines
Hard limits stay hard. Working rules start strict and loosen as trust builds. Do not skip the second part: an agent that needs approval for everything forever is an agent you will stop using.
Add these standing rules to my business memory file and follow them in every future session. Permanent red lines. These never change, no matter what: 1. Never commit me to a price, discount, refund, payment plan, scope change, or contract term. Money and legal promises are mine alone. 2. Never present a guess as a fact. Mark anything uncertain as [ASSUMPTION]. 3. When something could cost real money, create legal exposure, or damage a client relationship, stop and show me before acting. Working rules. These are my starting defaults. They are meant to loosen over time, but only when I explicitly say so: 4. Client-facing and external messages: draft and HOLD until I approve each one. 5. Publishing anything publicly: approval first. 6. Deleting, moving, renaming, or overwriting files: show me the plan first. How loosening works: - Only I can loosen a working rule, and only in specific carve-outs, never wholesale. Example: "You can file invoices into the client folders without asking" or "Send routine payment receipts automatically." - When I grant a carve-out, record it in the memory file with the date and its exact limits. Everything outside the carve-out stays on approval. - If you have earned a strong track record on some category of task, you may propose a carve-out and show me the evidence. I decide. One more rule so this does not slow us down: never use these rules as a reason to stall. If a task can proceed safely in draft form, proceed and bring me the drafts. Ask only when a red line is actually in play. Confirm these rules back to me in plain English. Then save them to the business memory file.
Part 2: Money
5. Who owes me money?
This is usually the fastest money prompt in the whole library.
Build me a true accounts receivable picture. Use this source order: 1. Live accounting, invoicing, or payment systems first, such as QuickBooks, Xero, Stripe, Square, FreshBooks, Wave, or similar 2. Invoice exports, payment exports, spreadsheets, or invoice PDFs second 3. If neither is available, stop and tell me exactly which report to export and where to find it Do not call anything overdue until you cross-check whether payment was actually received. Show me: 1. Total outstanding amount 2. Every unpaid invoice by client 3. Aging buckets: current, 1-30 days, 31-60 days, 61-90 days, and 90+ days 4. The three clients holding the most outstanding money 5. Partial payments 6. Credits that appear unapplied 7. Invoices created but not sent 8. Anything that looks odd or inconsistent Then draft a collection message for each genuinely overdue invoice. Use the relationship history from my email before drafting: - Good client under 30 days late: friendly nudge - 31-60 days late: clear follow-up - 61-90 days late: firmer message - 90+ days late: flag as call-first and give me three talking points instead of an email Mark every draft HOLD. Do not send anything. End with one sentence: the single invoicing habit that would make this list shorter next month.
6. Where is my money actually going?
Analyze my business expenses. Use live accounting or payment systems first. If you cannot access those, use exports, spreadsheets, bank/card reports, receipts, or expense files. If you have none of those, tell me exactly what report to download and from where. Group expenses into plain business-owner categories: - Payroll - Contractors - Software - Rent or office - Marketing - Advertising - Travel - Professional services - Taxes and fees - Equipment - Everything else Then show me: 1. Total expenses for the period reviewed 2. The fastest-growing categories 3. The five biggest individual expenses 4. Recurring charges that look unusually high 5. Duplicate or suspicious charges 6. Expenses that do not clearly connect to revenue, delivery, retention, or operations Keep it to one page. Use plain language, not accounting jargon.
7. Catch the subscription leak
Find every recurring subscription or recurring charge in my business. Use accounting, bank/card, or payment data first. If unavailable, use invoices, receipts, email confirmations, or exports. For each recurring charge, list: 1. Vendor 2. Monthly or annual cost 3. Billing frequency 4. What the tool or service appears to be for 5. Last evidence of use, if available 6. Whether it looks essential, questionable, or likely unused Look for usage evidence in email alerts, login notices, invoices, files, calendar events, team references, or project records. Flag anything with no evidence of meaningful use in the last 90 days. At the top, show the total estimated annual savings if I cancel the likely-unused items. Do not cancel anything. Give me a cancel-candidate list only.
8. Price check
Review my pricing. Use proposals, invoices, contracts, website pages, sales emails, and service documents from the last two years. Tell me: 1. When I last raised prices 2. My average project/order value by year 3. Which services are being quoted at the same price as one or two years ago 4. Which clients or offer types are still on old pricing 5. Where my scope has grown but the price has not Then review three public competitors or comparable providers. Use only public information you can actually find. Tell me what they charge, how they package similar work, and where I appear underpriced or overcomplicated. Do not tell me to "charge what I'm worth." Give me specific numbers and the reasoning behind them.
9. Which clients are actually profitable?
Revenue rank and profit rank are not the same thing.
Rank my clients by profit per hour of attention, not just revenue. Use invoices, payment records, proposals, email volume, meeting history, revision rounds, support requests, project delays, and client notes. For each client, estimate: 1. Total revenue 2. Number of emails or threads 3. Number of meetings 4. Revision rounds or support requests 5. Timeline friction 6. Scope creep signals 7. Estimated attention required Then rank: - Best five clients by profit per hour of attention - Worst five clients by profit per hour of attention - The client who looks good on revenue but appears costly once effort is counted - The client who is underrated because they are low-drama and profitable Show the evidence used. Mark estimates clearly.
10. How exposed is my business?
Analyze my client concentration risk. Use invoices, payment records, and revenue data from the last 12 months. Tell me what percentage of revenue came from: 1. My largest client 2. My top three clients 3. My top five clients Then review the relationship evidence for those clients: contracts, renewal dates, payment history, recent email tone, meeting history, and any unresolved issues. Tell me plainly: 1. What happens if the largest client leaves next month 2. Whether the relationship looks stable, uncertain, or at risk 3. The fastest two realistic moves to reduce this exposure 4. Which past or current clients are most likely to help fill the gap Use numbers. Do not reassure me unless the evidence supports it.
11. The next 90 days of cash
Build a simple 90-day cash picture. Use live accounting, payments, invoicing, payroll, and expense tools first. Use files and exports second. Include: 1. Expected money in, by week 2. Expected money out, by week 3. Invoices due and their real due dates 4. Recurring charges 5. Payroll or contractor payments, if available 6. Known upcoming commitments from email, contracts, or calendar 7. Any payments that look uncertain Lay it out week by week. Flag any week that looks tight. If a tight week exists, list the three fastest levers in order: 1. Which invoice to chase 2. Which expense to pause or delay 3. Which payment or commitment to renegotiate Use a simple table. No accounting jargon.
12. Tighten my payment terms
Analyze my payment terms and actual payment behavior. Use invoices, contracts, proposals, accounting records, payment records, and email threads. Tell me: 1. Current payment terms I use 2. Average days-to-pay by client 3. Clients who consistently pay on time 4. Clients who consistently pay late 5. Services or project types where deposits should be required 6. Where my current terms create avoidable cash strain 7. Whether late fees are worth using for my client mix Then draft: 1. A cleaner payment terms paragraph for future proposals 2. A deposit requirement paragraph where appropriate 3. A note for one or two chronic late payers, matched to the relationship Mark all notes HOLD. Do not send anything.
Part 3: Offers and Pricing
13. Tier my past clients and design their next offer
Your past clients are usually the cheapest revenue you will ever generate.
Review my past and completed clients using invoices, payment records, email history, client notes, meeting records, project files, results, complaints, testimonials, and repeat work. Rank clients into three tiers: Tier 1: paid well, paid on time, got strong results, low drama, good relationship, likely to buy again or refer Tier 2: solid clients with some friction, smaller budgets, slower timelines, or less obvious expansion potential Tier 3: small, slow, difficult, low-margin, or high-attention clients For Tier 1 and Tier 2, design two high-ticket offers grounded in what these clients actually asked for, complained about, wanted next, or paid for before. For Tier 3, design one lower-touch offer between $500 and $1,000 that helps them without consuming too much of my time. For each offer, include: 1. Offer name 2. Promise 3. Price 4. Who it is for 5. Why this tier would say yes 6. Evidence from real emails, invoices, notes, or projects 7. What not to include so the scope stays controlled Do not invent demand. If the evidence is weak, say so.
14. Find the offer hiding in my inbox
Read my inbound emails from the last year. Find every time someone asked for something I do not officially sell, including: - "Do you also do X?" - "Can you help with Y?" - "Would you be able to look at this?" - One-off favors - Small requests I handled manually - Repeated advice I gave for free - Problems people asked about before or after buying Group the requests by theme. If three or more people asked for the same thing, treat it as possible offer demand. For each candidate, show: 1. What people asked for 2. The real quotes proving demand 3. Who asked 4. Whether they were prospects, clients, or past clients 5. Suggested offer format 6. Suggested price based on comparable work I have billed before 7. Whether it should be high-ticket, low-ticket, retainer, workshop, audit, or add-on Do not create an offer unless the demand is visible in the evidence.
15. Productize the custom work I keep repeating
Review my last ten proposals and project files. Find the work I keep scoping from scratch even though the core is basically the same. Identify: 1. Steps that appear in almost every project 2. Deliverables that barely change 3. Parts that genuinely vary 4. Common client requests 5. Common delays 6. Revisions or scope creep patterns 7. What I actually charged 8. How long delivery really took Then package the repeatable core as a fixed-scope, fixed-price offer. Include: 1. Offer name 2. Who it is for 3. What is included 4. What is explicitly not included 5. Required client inputs 6. Timeline 7. Price based on historic billing 8. One-page sales description The goal is simple: next time this request comes in, I send one page instead of writing a custom proposal from scratch.
16. Build my offer ladder
List everything I currently sell using my website, proposals, invoices, contracts, and sales emails. Create an offer ladder from lowest price to highest price. For each offer, include: 1. Name 2. Buyer 3. Price 4. Promise 5. Delivery model 6. Whether it appears active, outdated, confusing, or overlapping Then identify gaps: 1. Where a buyer has no easy starting point 2. Where a happy client has nowhere to go next 3. Where two offers compete with each other 4. Where an offer attracts the wrong buyer 5. Where I am missing a high-ticket or recurring option 6. Where I have too many low-value offers Propose the smallest set of changes needed to make the ladder work. Every recommended offer must be grounded in real evidence from emails, invoices, proposals, or client demand.
17. Turn project clients into recurring revenue
Find every client who completed a project and then came back within six months with questions, small requests, follow-up work, troubleshooting, extra advice, or another project. Use invoices, emails, support threads, meeting notes, and project records. That pattern may indicate retainer demand. For each client, show: 1. Original project 2. Follow-up requests 3. Whether the follow-up was paid or unpaid 4. Time/attention consumed 5. What recurring need appears to exist Then design a retainer offer: 1. What is included monthly 2. What is excluded 3. Monthly price 4. Minimum term 5. Best time to offer it 6. Which clients should receive the offer first Draft the pitch email for the three past clients most likely to say yes. Use relationship-specific details from the record. Mark all drafts HOLD.
18. The mid-engagement upsell
Review my current active clients. For each client, read recent emails, notes, meeting records, proposals, and project files. Find adjacent needs they have mentioned in passing. Look for: - Problems they said they would handle later - Requests outside current scope - Repeated questions - Bottlenecks slowing the current project - Things they need after this engagement ends - Risks they do not appear to have solved yet For each client, show: 1. Client name 2. Current engagement 3. Adjacent need 4. Exact signal or quote 5. Natural expansion offer 6. Best moment to raise it 7. One or two sentences I can say on the next call Do not write pitch emails. Give me conversation openers only.
19. Kill or reprice the offer costing me attention
Compare every service or offer I sell using real business records. Score each offer on: 1. Revenue 2. Frequency of purchase 3. Delivery time 4. Email volume 5. Meeting load 6. Revision rounds 7. Scope creep 8. Client satisfaction signals 9. Profit per hour of attention 10. Strategic value Find the offer that performs worst. Make the case for whether I should: - Kill it - Reprice it - Narrow the scope - Turn it into a template/product - Move it to a lower-touch delivery model Show the numbers and evidence. Then tell me what to do with existing clients on that offer. If a transition note is needed, draft it and mark it HOLD.
20. Pre-sell before I build
I am considering a new offer: [describe the offer]. Before I build anything, help me test whether people actually want it. First, write a plain-language one-paragraph description of the offer. No hype. Then search my email history, client notes, invoices, proposals, and past conversations to identify the ten warmest people for whom this offer appears relevant. For each person, show: 1. Why they are a fit 2. What evidence supports that 3. Whether they are a client, past client, prospect, or referral partner 4. What specific pain or goal this offer connects to Then draft a short personal note to each person asking one honest question: would this be useful, and what would make it a yes? Make the tone personal, not mass marketing. Mark all drafts HOLD. End with a simple test rule: what level of positive response would justify building this, and what response would tell me not to build it.
21. Write the offer one-pager
Create a one-page sales document for this offer: [offer name]. Use evidence from my business records. Include: 1. The problem in the buyer's own words, pulled from real emails or notes 2. Who this is for 3. Who this is not for 4. What they get 5. What it costs 6. How long it takes 7. What happens after they say yes 8. Proof from real past results 9. Common objections and plain answers 10. Clear next step Write it in my voice based on sent emails, not generic marketing voice. Flag every claim you could not verify in my records. Do not invent proof, testimonials, numbers, outcomes, or guarantees.
22. Raise the price on the next proposal
Review my proposals, invoices, and contracts from the last two years. Find the service most overdue for a price increase. Show me: 1. Current price 2. Oldest price found 3. Average price over time 4. Scope changes over time 5. Delivery effort 6. Competitor or market comparison if available 7. Recommended new price 8. Why that number makes sense Then draft: 1. The pricing language for the next proposal 2. A no-apology explanation of the new price 3. A recommendation for current clients on old pricing 4. Whether to grandfather anyone, and for how long 5. A two-sentence answer for when a current client asks why the price changed Keep the language clear, direct, and calm. No long justification paragraph.
Part 4: Clients
23. Which clients are drifting?
Cooling clients rarely announce it. They just start replying slower.
Review my active client relationships using recent emails, meeting notes, calendar records, project management updates, support requests, and client notes. Score each active client as: - Warm - Neutral - Cooling - At risk Use real signals, including: 1. Reply speed changing 2. Shorter or colder messages 3. Missed meetings 4. Unanswered questions 5. Delayed approvals 6. Payment delays 7. Reduced enthusiasm 8. Fewer proactive updates 9. Threads where I failed to respond 10. Scope or expectation tension Show the cooling and at-risk clients first. For each one, give me: 1. Relationship score 2. Evidence 3. Likely reason 4. Best next action 5. Draft no-agenda check-in note in my voice Mark all drafts HOLD.
24. The client I should call today
Based on everything you can access (emails, invoices, notes, meeting records, project status, payment history, and client relationship signals), tell me which one client I should personally call today. Give me only one answer. Include: 1. Client name 2. Why this is the highest-priority call 3. What is at stake 4. Three specific things to say on the call 5. One thing not to say 6. Whether the goal is retention, repair, upsell, collection, clarity, or appreciation Do not give me a list of ten. Pick the one that matters most.
25. Draft the awkward email
I need to send a difficult email and I have been avoiding it. Situation: [late payment / price increase / saying no to scope creep / ending an engagement / pushing back a deadline / correcting a misunderstanding / refund issue / boundary issue] Client or contact: [name] Before drafting, read our email history, project history, invoices, contracts, and notes so you understand the relationship and context. Then draft two versions: 1. Softer version 2. Firmer version Both should be: - Direct - Warm where appropriate - Clear - Not over-apologetic - Not corporate - Not defensive - Specific about what happens next Tell me which version you would send and why. Mark both drafts HOLD.
26. Turn a finished project into a case study
Find our best recently finished client project based on evidence in my files, emails, notes, invoices, results, testimonials, final messages, or before-and-after proof. Pick one project. Reconstruct the case study from the actual record. Include: 1. Type of client or business, without naming them unless already approved 2. Where they started 3. What problem they had 4. Why they chose us 5. What we did 6. What changed 7. Proof of outcome 8. Client language or quotes where available 9. What made the work successful 10. What a similar client should learn from it Flag any claim you cannot verify. Draft a one-page case study. Do not use the client's name unless I explicitly approve it.
27. Build the onboarding pack we never wrote
Review the email threads, documents, forms, checklists, contracts, and project files from the last three clients we onboarded. Reconstruct what actually happens during the first two weeks. Identify: 1. What we send 2. What we ask for 3. What clients often forget 4. What causes delays 5. What we repeat manually 6. What expectations are unclear 7. What would make onboarding smoother Then build an onboarding pack: 1. Welcome email 2. What-to-expect one-pager 3. Client checklist 4. Internal checklist 5. Timeline for the first two weeks 6. FAQ based on real client confusion Write the welcome email in my voice based on real sent emails. Mark the email HOLD until approved.
Part 5: Sales and Leads
28. Mine the inbox for money
Most established businesses have revenue sitting in dead threads.
Go through my inbox and sent mail from the last 90 days. Find every thread where someone showed buying intent and the conversation died. Look for: - Inquiries I never answered - Warm replies I dropped - People who asked for pricing - People who requested a proposal - "Let me think about it" threads - Referral introductions that went nowhere - Past clients asking about something new - Prospects who said "circle back" - People who opened a buying conversation but never got a clean next step Rank the opportunities by likely value and likelihood of revival. For each thread, show: 1. Contact 2. Company if available 3. Buying signal 4. Last message 5. Approximate value if known 6. Why it died 7. Recommended next step 8. Draft revival reply in my voice Mark every draft HOLD. Show me the list before anything is sent.
29. Follow up on every open proposal
Find every proposal, quote, estimate, or pricing email I sent that never received a clear yes or no. Use email, CRM, proposal software, files, and accounting records. For each open proposal, show: 1. Prospect or client 2. Amount 3. Date sent 4. Service or offer 5. Last thing either side said 6. How cold the opportunity is 7. Whether it should be followed up, closed out, or revived later Then draft a follow-up matched to the situation: - Recent proposal: light nudge - 30+ days old: clear check-in - 60+ days old: closing-the-loop message - Very old but still relevant: honest revival note Make it easy for them to say no. A fast no is better than slow silence. Mark all drafts HOLD.
30. Profile my best client so I can find more
Identify my three best clients using evidence, not memory. Use invoices, payment history, project records, email tone, referrals, repeat work, testimonials, satisfaction signals, low-drama delivery, and profit per hour of attention. For each best client, identify: 1. Industry 2. Company size or buyer type 3. Situation when they hired me 4. Trigger that made them buy 5. Pain they wanted solved 6. Words they used in their first email or buying conversation 7. What they valued most 8. Why they were easy or profitable to serve Then write the shared profile across the best three clients. Finally, list ten specific places, events, moments, platforms, referral sources, or buying triggers where I could find more people like them. Do not say "networking events." Be specific.
31. The reactivation batch
Build a list of past clients we have not worked with in over a year. Use invoices, email history, CRM, project files, and notes. Sort them into two lanes: Lane 1: real relationship, good experience, worth a personal note Lane 2: transactional client, lower relationship depth, lower-touch reactivation For each person or company, show: 1. Last project 2. Last payment date 3. Last meaningful interaction 4. Relationship strength 5. Best reason to reconnect 6. Whether there is a likely next offer Draft one reactivation note for each lane. For Lane 1, reference something real from the relationship. For Lane 2, keep it useful and low-pressure. Do not make warm contacts sound like they are receiving a mass email. Mark all drafts HOLD.
Part 6: Marketing
32. A month of content from what I already know
Create a month of content ideas from real business evidence. Read what you can access that shows my expertise: - Sent emails where I explained something to a client - Proposals - Meeting notes - Client questions - Past posts - Case studies - Sales calls or transcripts - Internal notes - Training materials - Website pages Extract 20 content ideas. Every idea must come from a real moment. For each idea, include: 1. Source 2. Real client question, mistake, objection, or situation behind it 3. One-sentence point 4. First line written out fully 5. Best format: post, article, video, email, newsletter, or FAQ 6. Why my best buyer would care Cut anything generic. If it could appear on any competitor's blog, do not include it.
33. Make my website answer real buyer questions
Read my last 100 inbound emails from prospects and clients. List the questions people actually ask before buying. Then read my website. Show me the gap between what buyers ask and what the website answers. For the top five unanswered questions, draft the answer as it should appear on the website. Use plain language and my real phrasing from sent replies where possible. For each answer, include: 1. The buyer question 2. How often it appears 3. Whether the website answers it now 4. The answer I should add 5. Where it should go on the site 6. Why it matters for conversion 7. Whether it would help AI assistants understand and recommend my business Write the answers so a human trusts them and a machine can quote them cleanly.
34. Rewrite the page that matters
Review this services page: [URL] Read it as a skeptical buyer with three competitor tabs open. Do not be polite. Tell me: 1. What is unclear in the first five seconds 2. Where I talk about myself instead of the buyer's problem 3. Which claims have no proof 4. Where the offer is confusing 5. Where the page creates friction 6. What questions a serious buyer still has 7. Why someone might leave without contacting me Then rewrite the page using proof from my actual files, emails, case studies, testimonials, proposals, results, and client language. Flag every claim you cannot verify. Do not invent numbers, testimonials, logos, outcomes, or guarantees.
35. Competitor sweep
Compare my business against these three competitors: 1. [URL] 2. [URL] 3. [URL] Review their websites and public presence, then review mine. Give me a one-page competitive positioning brief. Include: 1. What they claim that I do not 2. What I can prove that they cannot 3. Where their offers are clearer than mine 4. Where my offer is stronger but poorly explained 5. What buyer objections they answer better 6. What buyer objections I answer better 7. The positioning gap none of us clearly owns 8. The one change I should make first Ground every point in something you actually read. Use quotes where useful. No generic strategy language.
Part 7: Operations
36. Write down the process that lives in my head
Use this before vacation, hiring, or delegation.
There is a process in this business that mostly lives in my head: [name the process]. Do not ask me to explain it step by step. Instead, reconstruct the process from evidence. Review the last five times this process happened using emails, files, documents, forms, client folders, project records, calendar events, invoices, notes, and messages. Draft the process as it actually happens, including: 1. Trigger 2. First step 3. Owner 4. Tools used 5. Files involved 6. Client communication 7. Internal steps 8. Exceptions 9. Workarounds 10. Common mistakes 11. Final output 12. What confirms it is done Then show me the draft and ask me only about the steps that left no paper trail. After I answer, create a clean SOP that someone could follow if I were unreachable for two weeks.
37. The Friday report
Build this week's Friday report from what you can actually access. Use live systems first: accounting, payments, invoices, email, calendar, CRM, and project management tools. Use files or exports second. Create a one-page report with: 1. Money received this week 2. Money still owed 3. Overdue invoices 4. Open proposals 5. Client threads that went quiet 6. Commitments I made in email this week that do not appear completed 7. Meetings or calls that require follow-up 8. One relationship I should pay attention to 9. One operational issue that needs fixing 10. The single most important thing for Monday morning Show the source behind each important item. Then save the instruction so that when I ask for "the Friday report," you produce this exact format again.
38. Clean up the shared drive
Review this folder or shared drive: [folder/drive]. Do not move, delete, rename, archive, or overwrite anything yet. First, map the mess. Identify: 1. Duplicate files 2. Multiple versions of the same document 3. Old contracts 4. Files with unclear names 5. Files in the wrong place 6. Empty or unused folders 7. Sensitive files that need caution 8. Files that appear safe to archive 9. Files that should not be touched Then propose a simple folder structure a new hire could understand on day one. Show me the full move plan: - What moves - Where it moves - What gets renamed - What gets archived - What stays untouched - What requires my decision Wait for my approval before making any changes. After approval, execute the approved plan and give me a report of exactly what changed.
39. Turn a meeting into an action list
Read this meeting transcript, recording, or notes: [attach or link]. Read the whole thing, not just the summary. Extract: 1. Every commitment made 2. Who owns each commitment 3. Deadline or date mentioned 4. Implied deadline if no date was said 5. Open questions 6. Risks or ambiguities 7. Decisions made 8. Follow-up needed Then draft a follow-up email to the attendees confirming what was agreed. Write it in my voice based on my prior sent emails if available. Flag anything ambiguous that will cause problems if nobody clarifies it. Mark the email HOLD.
Part 8: Deciding and Hiring
40. Pressure-test a decision before I make it
I am about to make this decision: [describe decision]. Your job is to argue against it before supporting it. First, review the relevant evidence from my business: financials, emails, client records, proposals, contracts, operations, team workload, calendar, past decisions, and anything else connected. Then give me: 1. The three strongest reasons this decision could go wrong 2. Evidence from my own business supporting each risk 3. What I would need to believe for this decision to be right 4. What data is missing 5. The cheapest way to test the decision before fully committing 6. The cost of waiting 7. The cost of moving too fast Then tell me what you would do and why. Do not give me generic pros and cons. Use my actual situation.
41. Draft the job description from the work itself
I think I need to hire someone, but I am not clear on the role. Do not start with a job title. Review the evidence of what is overflowing in the business: - Sent emails - Repeated tasks - Late work - Calendar load - Client follow-up gaps - Reports - Admin work - Project bottlenecks - Files updated at the last minute - Work I keep doing manually - Things clients wait on Tell me what role this actually is based on the work. Then draft the job description. Include: 1. Real role title 2. Why the role exists 3. Outcomes they own 4. Responsibilities 5. What they should not own 6. Tools they need to use 7. Three real scenarios from my business they would handle in week one 8. What good performance looks like in 30, 60, and 90 days 9. What experience matters 10. What experience does not matter Tell me honestly whether any of this should be handed to you before I hire.
42. The overnight brief
Use this when the agent environment supports a longer-running task and you want it to keep going instead of stopping at the first missing piece.
I am handing you a large task: [describe task]. Work it through to the end as far as you safely can. Rules: 1. Never send anything external. 2. Never publish anything. 3. Never delete, move, rename, or overwrite files. 4. Never commit me to a price, deadline, promise, meeting, or decision. 5. If information is missing, make a reasonable assumption only when needed to continue. 6. Mark every assumption clearly as [ASSUMPTION]. 7. If something is too risky to assume, flag it and continue with the parts you can complete. 8. Keep a running list of sources used. 9. Check your own work before calling it finished. When I return, give me: 1. The finished work 2. Every assumption you made 3. Anything you could not resolve 4. Anything that requires my approval 5. The sources you used 6. How you checked the output 7. The next three actions I should take Do not ask me questions unless the task truly cannot continue without the answer.
Part 9: Build Your Agents
This is where the real leverage lives.
A prompt saves you an hour. An agent is that same prompt running on a schedule, watching your real systems, with drafts waiting for you in the morning, whether or not you remembered to ask.
Everything above this line was practice for this part.
43. Design my agent roster
Run this after you have used a few prompts from this library. Fable has seen your business by now. Let it tell you what to automate.
You have now seen my business: the connected tools, the files, the patterns in my email, where the money comes from, and where my attention goes. Based on that evidence, design my agent roster: the five recurring agents that would save me the most hours or recover the most money. Rank them by leverage. For each agent, show: 1. Agent name 2. What it watches (which inbox, tool, or folder) 3. What it produces (drafts, a report, a ranked list) 4. How often it should run 5. The specific evidence from my business that justifies it 6. What could go wrong if it misfires 7. Risk level: low, medium, or high Rules: - Ground every agent in evidence you actually saw. Do not propose an agent for something that happens twice a year. - Every agent starts draft-only: it prepares, I approve. Then build the number one agent right now: write its standing instructions, set its schedule, and run it once in front of me so I can see exactly what will be waiting for me each morning.
44. Turn any prompt into a recurring agent
The graduation prompt. When something from this library earns its keep twice, run this on it.
Take this task and turn it into an agent that runs on a schedule without me starting it: [name the prompt or describe the job]. Set it up completely: 1. Write the standing instructions, including my voice and my red lines 2. Choose the schedule based on how often the underlying thing actually changes: daily for inboxes, weekly for money, monthly for pricing 3. Decide where the output lands so I see it without hunting for it 4. Start it in draft-only mode: everything it produces waits for my approval Then define its promotion path: 1. Track its accuracy as I approve or correct its output 2. After it has been right twenty times in a row on some category of item, propose moving just that category to automatic 3. Bring me the evidence when you propose the promotion, and I decide 4. Promotions are carve-outs, not blank checks: record each one in the memory file with its exact limits, and my permanent red lines (committing me to money or legal terms) are never up for promotion Run it once right now as a test so I can see the output before it goes on the schedule.
45. The follow-up agent
The one agent almost every established business should run. Money dies in the gap between "they wrote to us" and "we replied."
Build me an agent that makes sure nothing with money attached ever slips again. Every morning it sweeps my email and open proposals for: 1. New inquiries that arrived since yesterday 2. Proposals with no reply after five business days 3. Anything I promised in writing that shows no evidence of being delivered For each item it finds, it drafts the right response in my voice from my sent mail, all marked HOLD. Then it hands me one short morning list: 1. What needs a decision 2. The drafts, ready to approve 3. Nothing else If a morning is clean, the report is one line saying so. Build it, run today's sweep right now, and show me the output before you put it on the schedule.
46. The Monday money agent
Build me an agent that runs before I start work every Monday. It produces one page: 1. Current receivables aging from my connected accounting tool 2. Payments that arrived last week 3. Chase drafts for anything genuinely overdue, cross-checked against payments received so it never chases someone who already paid 4. The one number I should be watching this week, with the reason in plain English All drafts HOLD. Build it and run the first one now, so Monday's version arrives in a format I have already approved.
47. Build me a tool, not a document
The prompt that shows a non-technical owner what Fable actually is. It does not just write about your business. It can build for it.
Look at the evidence of my business and find the place where I or my clients keep doing something manually that a simple tool would fix. Look for patterns like: - Quotes that take an hour to assemble from old proposals - An intake process that lives in six back-and-forth emails - A calculation I redo in my head every time someone calls - A status question clients keep emailing me to ask Propose the one tool worth building first. Describe it in plain terms: 1. What I would see 2. What my client would see 3. What manual work it replaces 4. What it would need to stay accurate Then build a working version I can try today. Rules: 1. It runs on my machine only. Nothing public. 2. It touches no live client data until I say so. 3. Take the time to check your own work before you show me. I care that it works, not that it was fast. When I have used it and it holds up, we will talk about where it should live permanently.
How to Use This Library
Do not run all 47 prompts.
That is not the point.
Start with the setup sequence:
- Prompt 1: Show me what you can actually see
- Prompt 2: Build my business memory file
- Prompt 4: Set my red lines
Then pick the prompt that makes you slightly uncomfortable.
That reaction is useful.
It usually points to the place where the business is leaking money, attention, margin, or trust.
If you want the fastest revenue sequence, run these in order:
- Prompt 5: Who owes me money?
- Prompt 29: Follow up on every open proposal
- Prompt 28: Mine the inbox for money
- Prompt 13: Tier my past clients and design their next offer
- Prompt 17: Turn project clients into recurring revenue
- Prompt 22: Raise the price on the next proposal
If you want the fastest operational clarity sequence, run these:
- Prompt 3: Find the work worth handing off
- Prompt 36: Write down the process that lives in my head
- Prompt 37: The Friday report
- Prompt 38: Clean up the shared drive
- Prompt 41: Draft the job description from the work itself
Then graduate. Any prompt that has been useful twice belongs in Part 9:
- Prompt 43: Design my agent roster
- Prompt 44: Turn any prompt into a recurring agent
- Prompt 45: The follow-up agent
- Prompt 46: The Monday money agent
The prompts save you hours. The agents work while you are not there.
And when Fable does something correctly, say so clearly:
"That was right. Remember how we did this."
That is how this stops being a prompt library and starts becoming business infrastructure.
Common questions
What is Claude Fable 5 and why does it matter for a small business?
Claude Fable 5 is Anthropic's most capable AI model, built for long, multi-step work. Connected through plugins and integrations, it can read your real business systems, such as email, QuickBooks, Stripe, your calendar, and your CRM, plus the files on your computer, then work through a whole job and check its own output. For a business owner that means it can build an accounts receivable picture, mine an inbox for dropped leads, or design offers from real client evidence, rather than just answering questions.
Do I need to be technical to use these prompts?
No. Every prompt is written to be pasted as-is. Each one tells Fable which systems to check first, what evidence to use, what to show you, and where to stop for your approval. If Fable cannot reach a system, the prompts make it tell you exactly which report to export and where to find it.
Is it safe to connect an AI agent to my business email and accounting?
It is safe when you set rules first. The library's setup prompts establish permanent red lines (the agent never commits you to money or legal terms, and never presents a guess as a fact) and working defaults (client-facing messages stay drafts marked HOLD until you approve each one, publishing needs approval, and file changes need a plan first). Nothing gets sent, published, or deleted without you.
Which prompt should a business owner run first?
Run the inventory prompt first: it makes Fable test what it can actually reach, live connections and files both, and report what worked. Then build the business memory file so every future session starts already knowing your business, and set your red lines. After that, most owners get the fastest payoff from the accounts receivable prompt: who owes what, aged into buckets, with chase drafts held for approval.
What is the difference between running a prompt and building an agent?
A prompt is a job you start by hand and it runs once. An agent is the same job on a schedule, watching your real systems and leaving drafts waiting for you, whether or not you remembered to ask. The last section of the library turns any prompt that has proven useful into a recurring agent, starting draft-only and earning wider permissions only when you explicitly grant them.
Do these prompts work with ChatGPT or Codex?
Mostly, yes. The prompts assume an AI agent that can reach your real systems through connectors and take multi-step action. Claude Fable 5 is the strongest fit for the long-running jobs, but Codex and ChatGPT with connectors can run versions of most of them. The rules that make the library work, live systems first, evidence over guesses, and drafts held for approval, apply the same in any tool.
Running on a CRM instead of an inbox? The same evidence-first approach powers the 12-prompt HubSpot revenue audit. And if you want to know whether AI agents can even read your business from the outside, start with agent-readiness.