Start Here (English Edition)

English edition. S1–S6 are available in English; S7–S9 and the milestones are published in Japanese.

SOVREN Framework is a free, open-source handbook for running a small business with AI. You don’t just read it — you build something on your own computer from the first session.

Why it exists is in the Manifesto; the order of the nine skills is in the framework. This page gets you started.

On this page

• Understand it in 30 minutes
• Who it’s for
• How the copy-paste paths work
• Try it in two hours
• Build your business on it
• When you get stuck

Understand it in 30 minutes

There is one idea to understand:

Before AI defines who you are, hold your own Ontology yourself.

“Ontology” isn’t academic jargon here. It’s your own external map — how you decide, whom you work with, what you’ve done — kept in plain files on your own computer, in a form AI can read. With it, AI works from your context. Without it, you get the same generic answers as everyone else.

Who it’s for

Mainly small business owners who:

• already sell a product or service
• have been paid by a customer at least once
• have a computer and a folder they control
• read what AI produces and make the final call themselves
• don’t hand customer data, contracts, or credentials to AI without thinking

If you haven’t had your first sale yet, S1 still works. From S2 on, real customers and transactions make everything sharper.

You need	Notes
A computer	where you can manage your own folders and files
An AI agent app	one that can read and write files in a folder (S1, Step 1); no specific product required
Time	two hours to start; then one or two hours a week
Real data	your notes, customers, products, sales, actions — you don’t need it all on day one

How the copy-paste paths work

Every skill page, S1 to S6, starts with a copy-paste path: the prompts to paste into your AI agent, in the order you use them. You can start building before reading the whole page.

1. Paste one at a time, in order. Paste several at once and the AI skips steps.
2. Replace everything in 【brackets】 with your own words before sending. Left blank, you get generic answers.
3. Read everything the AI produces. Send back anything unclear or not in your words: “fix this.” Not adding files you haven’t read is the one habit that makes an Ontology grow.
4. When something doesn’t work, read the page. The prompts carry only the essentials; the reasons are in the text.

The prompts work as written, but adapt them to your business. Save the ones you reuse in your own folder.

As of October 2026: this framework assumes an AI agent that reads and writes files (such as Claude Code, Codex, or Antigravity). Most can work inside a chosen folder, ask before editing, read a rules file (AGENTS.md and similar) automatically, connect to outside services (connectors), and store procedures you can call by name (skills). Product names and screens change, so the pages describe features rather than brands.

Try it in two hours

1. Create one folder for your business and open only that folder in your AI agent
2. Paste prompts ① and ② from S1’s copy-paste path — the folders and the rules file
3. Talk through one decision you’ve been unsure about, and turn it into a concept file with ③ and ④
4. Log today with ⑤

myproject/
  ├── AGENTS.md       rules the AI reads every time
  ├── inbox/          voice memos and rough notes
  ├── ontology/       people, organizations, concepts, relations
  └── actionlog/      what you did and what happened

Build your business on it

Each skill goes: read → build the smallest version → try it once with real customers, work, or numbers → write the result back → next skill. Progress is measured by evidence of use, not pages read.

Skill	What you build
S1: Build your own information system	an AI-readable workspace, your decision criteria in words, a self-filling record
S2: Make information usable	outside knowledge as structured files, a market chosen with data, a three-year plan
S3: Take control of your voice	content from your real work, landing pages, a post → page → inquiry pipeline
S4: Multiply your revenue streams	a center business, revenue stages in equations, new high-margin offers, a monthly test cycle
S5: Build your money data	a monthly close for your business and yourself, runway, a tax calendar, alerts
S6: Manage your team with AI	where people data lives, team structure, instructions, acceptance, and your own dashboard
S7–S9 (Japanese)	an AI-centered organization, capital planning, managing many assets at once

When you get stuck

Symptom	Go back to
The AI only gives generic answers	S1 — add more of your own data
Your market or customer is vague	S2 — organize your existing deals
Posting doesn’t lead to inquiries	S3 — check the pipeline and measurement
You keep collecting new business ideas	S4 — narrow with revenue stages and verdict criteria
Money decisions run on gut feel	S5 — close the month
Your own workload never shrinks	S6 — split roles and hand work out

→ Start now: S1: Build your own information system