Skip to main content
This is the fastest path from zero to a running Rayify project.

1. Get an API key

Sign up at rayify.ai, open Settings → API Keys, click Create new key. Copy the sk_your_key-formatted token.

2. Pick your integration

If you’re using Claude / Cursor / ChatGPT / any MCP-compatible client, configure the MCP server:
Restart your AI client. The MCP server appears in the tool list with 8 tools + 3 guided prompts. Try one of the prompts:
This lists your most recently-created Rayify projects without you writing any code.

Option B - Python SDK

Option C - TypeScript SDK

3. Watch the project run

Open https://rayify.ai/projects/<your-project-id> in your browser. You’ll see:
  • The cross-family Reasoning panel (3-7 specialist agents) thinking through the brief in parallel
  • The Researchers fleet pulling supporting evidence
  • The Defender pipeline verifying every citation as it lands
  • The Judge panel scoring each agent’s output for calibration
When all stages finish, the project produces a DecisionPacket with the structured deliverable: selected options, residual objections, minority reports preserved verbatim, evidence with provenance, and a replay link.

What’s next