15 Fable prompts for people who don't code
Copy-paste stance prompts that run on your statements, calendar, inbox, and landing page. Here's how to get owner answers from Fable 5.
Jul 6, 2026
6 min read
Fable 5 left the Claude subscription plans on July 7. After that it’s metered: $10 per million input tokens, $50 per million output.
So every video and thread that week was some version of “run these coding jobs before the deadline.” Big refactors. Test suites. Migrations.
Great advice, for the people it’s for. But most people don’t have a codebase. They have a business, a calendar, an inbox, and a family schedule that eats 30 minutes a day nobody can account for.
The coding crowd figured out something first: Fable’s edge was never typing speed. You’re renting judgment. And judgment works on anything you can paste into a chat.
I spent the last year as the non-technical one learning this the hard way. My prompts used to be polite little work orders. “Summarize this.” “Review that.” The output was fine. Fine is the problem. An assistant with no stake gives you assistant answers.
The fix isn’t a better-worded task. It’s a stance. Give the model an identity with something to lose, a goal instead of a checklist, and permission to be blunt. That’s also exactly how Anthropic says to prompt Fable: brief goals over step-by-step instructions, because over-specifying degrades it.
So here are 15 stance prompts for the rest of us. Every one is copy-paste. Every one runs on things you already have: your statements, your calendar export, your notes, your landing page.
The business ones
1. The eulogy
Prompt
Write the eulogy for my business as if it died in 2028. What killed it? Be specific enough that I can prevent it, and tell me the earliest warning sign I’d see this quarter.
Morbid? Yes. But post-mortems written in advance are the cheapest consulting on earth.
2. The acquirer
Prompt
You’re a competitor who just acquired my company and got full access to everything. What do you fix first? What did we build that you’d never have wasted time on? Rank by what changes the P&L, not what offends your taste.
Nobody inside a company can answer “what would a rival delete.” The model has no feelings about your sunk costs.
3. The churn confession
Prompt
Here’s who canceled last quarter and the reason they gave on the way out. What’s the real reason none of them are saying out loud? Tell me the one fix that would have kept the most of them, and what it would have cost to ship.
Compare its answer to what you assumed was wrong. The gap between the two is usually the whole insight.
4. The refund letter
Prompt
Read my landing page. Write the refund-request email my most disappointed customer would send after 30 days. Then quote the exact sentence of my marketing that over-promised.
This one finds the gap between what you sell and what you ship, fast.
5. The pricing bully
Prompt
You’re a pricing consultant paid only on results. Here’s what I charge: [your prices]. Argue me into the price I’m afraid to charge. Give me the three objections I’ll hear and the reply to each.
6. The mystery shopper
Prompt
Go through my signup flow cold, as an impatient stranger with a full inbox. Narrate your inner monologue. Tell me the exact moment you almost left, and what would have kept you.
The money ones
7. The CFO who hates me
Prompt
Here are my card statements. Rank every subscription by cost-per-actual-use, write the cancellation list, and tell me which one I’m keeping purely out of identity.
8. The calendar audit
Prompt
Here’s my last month of calendar. You’re a ruthless COO who bills my time at $500 an hour. Which meetings were theft? Which standing meeting dies today? Show your math.
The personal ones
9. The pattern reader
Prompt
Here are my notes and journal entries from the last few months. What pattern am I clearly avoiding? Tell me the thing I already know but haven’t said out loud. One paragraph. No softening.
Fair warning: this one is not fine. Run it anyway.
10. The inbox archaeologist
Prompt
Read my sent email from the last 90 days. What do I keep promising and not delivering? Who am I slowly ghosting? What does my reply speed say about my actual priorities, versus my stated ones?
11. The household COO
Prompt
Here’s our family’s typical week. Find the recurring 30 minutes we lose every day, figure out what it’s actually costing us, and design it out of the schedule.
We run a version of this at home with an agent named Juno. The 30 minutes exists. It’s usually somewhere between school pickup logistics and “what’s for dinner.”
12. The negotiation postmortem
Prompt
Read the email or text threads from my last three negotiations — a raise, a deal, a price, anything with back-and-forth. Where did I cave first each time? What’s the tell right before I fold, and what should I say instead next time?
Roleplaying the next negotiation is a parlor trick. Grading the last three from the actual transcript is how the pattern breaks.
The meta ones
13. The last-mile detector
Prompt
Here’s my projects folder. What’s 80% done that one honest afternoon would ship? Rank everything by value-per-remaining-hour, and tell me which ‘almost done’ project I should officially kill instead.
I have 350+ pieces of content drafted and about 10 published, so this prompt was built from personal shame. It works.
14. The agent audit
Prompt
Read my CLAUDE.md and agent instructions. Which instruction is actively making my agents dumber? What am I over-specifying that you’d handle better on your own? Rewrite the file, shorter.
Instructions written for older models actively drag Fable down. Shorter is smarter now.
15. The honest investor update
Prompt
Draft the investor update I’d write if I were being fully honest. Include the number I’m hiding, and one sentence on why I’ve been hiding it.
You don’t have to send it. You do have to read it.
How to run these
Give it real material. Stances need stakes, and stakes need data. Paste the statements, attach the export, point it at the folder. Vague inputs get you horoscopes.
Set effort to high. Anthropic recommends high as the default and xhigh for the heavy stuff. These audits are the heavy stuff.
Don’t ask it to explain its reasoning. That can trigger a refusal, and in some setups Fable quietly hands your task to Opus 4.8 and burns your usage doing it. Ask for conclusions and evidence instead.
Mind the meter. These prompts chew through tokens reading your material. That’s exactly what gets expensive at metered rates, so batch them and bring your best inputs.
And if you do have a codebase: I wrote 10 more in the same style that treat it like a company. The day-one CTO, the activist investor, the departing cofounder.
The models got smart enough to tell you the truth. Most people are still asking them to summarize things.
I write about running a business (and a household) with AI agents, the wins and the faceplants. The newsletter is here.
Written by
Cathryn Lavery
Cathryn went from designing buildings to architecting products. She founded BestSelf, bought it back from private equity in 2024, and rebuilt it AI-native. She's currently building something new in AI. Little Might is where she doesn't have to keep it all in her head.
Related reading
-
Jul 6, 2026
10 prompts that make Fable act like an owner, not an assistant
-
Jun 18, 2026
How to build an AI house manager for your family
-
May 27, 2026
How I Plan Projects With AI: The Beads Workflow
-
May 27, 2026
Six Non-Coding Things You Can Do With Claude Code
-
May 5, 2026
How to Give AI Agents Real Access Without Risking Your Data