RocketFlow Documentation

RocketFlow lets you build AI chatbots trained on your own content — websites, documents, and FAQs — and deploy them as an embeddable chat widget, through messaging integrations, or via the API.

Quick start

1. Create a chatbot in the dashboard and give it a name and personality.
2. Train it by adding knowledge sources: website URLs, documents, or FAQs.
3. Embed it on your site with one script tag (copy it from Dashboard → your chatbot → Embed):

<script
  rel="preload"
  type="module"
  data-chat-id="your-chatbot-id"
  src="https://widget.getrocketflow.io/rkt-chat-loader.js"
  crossorigin="anonymous"
></script>

Full walkthrough: Getting Started.

Explore the docs

• Features — skills, automations, guardrails, CSAT, human handoff, auto-retrain, insights, webhooks, white-label.
• Integrations — Slack, Telegram, WhatsApp.
• Settings — API keys and team management.
• MCP Server — manage your chatbots from Claude, Cursor, and other AI assistants via the Model Context Protocol.
• Session Initialization — how the chat SDK manages sessions, for developers integrating @rkt/chat-client.
• Troubleshooting — common issues and fixes.

How the widget works

The loader script renders a chat button on your page and opens the chat UI in an iframe served from widget.getrocketflow.io. Chat messages are sent same-origin inside that iframe and proxied to the RocketFlow API server-side, so embedding requires no CORS setup on your site. Conversations, analytics, and leads appear in your dashboard in real time.