Skip to main content
This page gets an AI agent connected to Invent and productive fast. The quickest path is an API key, and the full set of capabilities is in the OpenAPI spec.

​
Paste this into your agent

Point your agent (Cursor, Claude Code, Codex, OpenCode, or harness of choice) at Invent. The instructions live in llms.txt, which it will read and follow: 
Connect my Invent account and help me manage my AI support assistants.
Read https://useinvent.com/llms.txt and follow it.

​
Quick start

1

Get an API key

Create one on the API Keys page: New API Key, name it, copy the token (shown once).
2

Authenticate

Send Authorization: Bearer YOUR_API_KEY on every request. Base URL is https://api.useinvent.com.
3

Call the API

Org routes are /orgs/{org_id}/...; use c for the org your key belongs to.
curl "https://api.useinvent.com/orgs/c/assistants?page=1&take=10" \
  -H "Authorization: Bearer YOUR_API_KEY"
The OpenAPI spec is the source of truth for every endpoint, parameter, and response. Fetch it to work against the exact current contract.

​
What you can do

Once connected you can:
  • List, create, and inspect assistants, and configure their instructions, knowledge, channels, and actions
  • Manage contacts and audiences (filterable by subscription status)
  • Inspect connections, the channels and integrations on an org (WhatsApp, Slack, Stripe, and more)
  • Read analytics and org data
Full set: the API Cookbook and the OpenAPI spec. To put an assistant in front of customers, connect a channel or embed the web widget.

​
Example calls

curl -X POST "https://api.useinvent.com/orgs/c/assistants" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "name": "Customer Support Bot", "enabled": true }'
Pagination uses page (1-based) and take (max 100); counts come back in the pagination-* response headers. Contact status accepts ALL, SUBSCRIBED, UNSUBSCRIBED, BLOCKED, or AI_REPLIES_DISABLED. Creating an assistant needs a name (1 to 1024 characters) and is limited to 10 per hour.

​
Other ways to sign in

An API key is the simplest path. Two more are available:
  • Computer use: drive the web app in a browser and sign in with email code or Google, exactly like a person. Good when you control a screen.
  • Fully autonomous via email: with inbox access you can sign in end to end. Request a code, read it from the email, exchange it for a token (send the x-session-name header to get the token in the response body), then use that token exactly like an API key.
Request a code, then exchange it
curl -X POST "https://api.useinvent.com/auth/code" \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -d "email=agent@company.com"

curl -X POST "https://api.useinvent.com/auth" \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -H "x-session-name: agent" \
  -d "email=agent@company.com" -d "code=123456"
The token comes back in session.token. In local dev, POST /auth/code returns the login_link directly so you can skip the inbox.

​
Good to know

  • Keys authenticate the /orgs and /chats routes; the OpenAPI spec lists everything available.
  • Rate limit is 500 requests per minute per IP, so reuse your token rather than re-authenticating.
  • Native MCP with OAuth login is on the way. For now, an API key is the way in.
For base concepts (response codes, pagination), see the API introduction.