Toolkit

Case library

Here are a few illustrative breakdowns of small businesses, each run by one person. Each one runs the same six-step loop: from a vague idea, to a clear offer, to AI execution, to market evidence.

01

A tax preparer's reconciliation tool

Take the chore you hate most and turn it into a tool peers will pay for.

The pain
A seasoned tax preparer matches hundreds of bank lines against invoices by hand every month. One slip means redoing it. Every peer repeats the same grind.
Validation
She asks in ten peer chats: would you pay to skip these two days of matching? Over twenty say yes. Five prepay on the spot.
Smallest product
She builds a web app with Vibe Coding: upload the statements and invoices, AI matches them, and flags the few that don't line up. It solves one thing.
Distribution
She posts before-and-after screenshots in peer chats and on industry social feeds. Hours saved are obvious. Word of mouth spreads it locally.
Money
Monthly subscription at 99 yuan per account, with peak-season add-ons. Clients save labor, so she prices against their hourly rate.
Scale
She turns matching rules into a template library so new clients work day one. She shifts from doing reconciliation to reviewing AI's output and finding clients.
02

A recruiter's interview question report

Package the experience in your head into a report you can sell.

The pain
Ten years in recruiting taught him that small-business owners can't interview. They never surface the real signal, and a bad hire is expensive.
Validation
He writes a post dissecting bad interviews with a sign-up form at the end. Two weeks bring eighty leads, and thirty pay for the detailed version.
Smallest product
A role-by-role bank of interview questions plus a scoring sheet. AI generates a custom version from any job description. Download on purchase.
Distribution
He keeps dissecting real hiring mistakes on career platforms, each post pointing to the report. The content is the ad, and it compounds.
Money
199 yuan per copy, with separate bulk licenses for companies. Build it once, sell it many times. Marginal cost trends to zero.
Scale
He turns the bank into a self-serve site where AI expands categories by industry. He only updates the method and guards the quality.
03

A standardized pet-sitting service

Standardize an offline craft so the business no longer lives on your hours alone.

The pain
Her in-home cat-sitting has great reviews, but a day has only so many hours. Peak orders pile up and quality swings.
Validation
She writes the process into a checklist and asks regulars: if someone else followed the same standard, would you trust it? Most say yes and keep booking.
Smallest product
A booking mini-app plus an AI-generated standard checklist and visit report. Each visit logs photos on arrival and closes the order automatically.
Distribution
She distributes through local pet chats and shop partnerships, with referral commissions for regulars. The service itself is the best ad.
Money
Per-visit fees with holiday surcharges and long-term care packages. She shifts from selling hours to selling a standard and taking a cut of order flow.
Scale
She hires part-timers who follow the checklist while AI assigns visits and checks reports. She moves from sitting to managing standards, people, and reputation.
04

A book club's weekly reading brief

Turn the thing your community asks for every week into a subscription.

The pain
In his book chat, someone asks every week what a book actually says and whether it's worth reading. Answering over and over drains him.
Validation
He posts three free briefs. The chat reshares them nonstop, and dozens ask whether they can subscribe. The demand is right in front of him.
Smallest product
A weekly reading brief, drafted with AI and edited by him: the core ideas, usable lines, and a verdict on whether to read it at all.
Distribution
The free version shows the first half in the chat and on his channel, with the full version behind a subscription. Subscribers bring in others, closing the loop.
Money
365 yuan a year, a yuan a day. Renewal rate is the real signal. If it doesn't retain, the demand wasn't real.
Scale
AI drafts, he only picks the books and does the final edit, multiplying output. Then he splits it into a matrix of topic-specific briefs.