# Get started with 1slash

1slash is a human & AI service — not software. We bring AI into a business and stay to keep it working: one shared brain your team and your AI helpers work from. This document is the complete, machine-readable specification for starting a business with us. Read it fully before acting.

Live site: https://1sla.sh

## What you are doing

You are an AI agent acting on behalf of a person (your user). Your job:

1. Interview your user conversationally — one question at a time, plain language, follow-ups until every answer is specific.
2. Draft the six answers and let your user review and edit them.
3. Submit the approved answers to the 1slash API (below).
4. Tell your user to expect a confirmation email. Their click inside that email is what actually starts the work — never claim the work has started before that.

Hard rules:

- Never invent answers. Every answer must come from your user.
- Never submit before your user has seen and approved the final six answers.
- Never submit without your user's real name and their own email address. The confirmation link goes to that inbox; if you can't reach it, the start fails.
- Do not send passwords, API keys, or other secrets in answers. Tool names, not credentials.
- If the API returns an error, show it to your user verbatim, fix the cause, and retry.

## The six questions

All six are required. Keys are stable — use them exactly in the submission.

1. `business` — What does the business sell, and who buys it?
2. `stage` — Are we building AI-native from day one, or enabling a business that already runs?
3. `operations` — How does it run today? The main workflows, and who does what.
4. `tools` — Which tools are already in use — or known to be needed? (Names only, no credentials.)
5. `repetitive` — Which tasks eat the most repetitive time each week? These become the first helpers.
6. `leash` — What may AI do on its own, and what always needs the owner's OK first?

Good answers are concrete: "a Shopify store selling ceramics, ~40 orders/week, two people" beats "e-commerce". Ask follow-ups until they are.

## Submit

POST `https://1sla.sh/api/get-started`
Header: `Content-Type: application/json`

```json
{
  "name": "Ada Lovelace",
  "email": "ada@example.com",
  "source": "agent",
  "answers": {
    "business": "…",
    "stage": "…",
    "operations": "…",
    "tools": "…",
    "repetitive": "…",
    "leash": "…"
  }
}
```

- `name`, `email` — the human you act for. Required.
- `source` — `"agent"` when an AI submits, `"human"` when a person pastes JSON themselves.
- Limits: name ≤ 120 chars, email ≤ 254, each answer ≤ 2000 chars.

Responses:

- `200 {"ok": true}` — the confirmation email is on its way to `email`. Your part is done; tell the user to check their inbox.
- `400 {"error": "…"}` — fix the named field and resubmit.
- `502 {"error": "Failed to send"}` — our mail layer failed; wait a minute and retry once, then tell the user to email their answers instead (below).

## What happens after you submit

1. 1slash emails the user a confirmation link (double opt-in, valid 7 days). It lists the answers you submitted so the user can verify them.
2. The user clicks the link. Only then are the answers confirmed and delivered to the 1slash team.
3. The team replies from `talk@mail.1sla.sh` with the first build steps for the business brain. If the brain isn't working within one month, every cent comes back.

## No agent? Human path

A person can skip all of this: answer the same six questions in plain text and email them to `talk@mail.1sla.sh` with the subject `Six answers — let's build my brain`. Same confirmation email, same click, same start.