Overview
Turn your documentation into an interactive chat experience. Visitors ask questions in natural language and get instant, accurate answers grounded in your actual content.
Pricing: Free Embed modes: Inline, Iframe, Popup
Visual styles
- Floating Chat — compact bubble in the corner that expands into a chat window
- Inline Messenger — full-sized chat interface embedded directly in page content
Configuration
| Setting | Type | Default | Description |
|---|---|---|---|
| Display Name | String | Assistant | Name shown at the top of the chat |
| Persona | Enum | Docs | Preset: Support, Sales, Documentation, or Custom |
| Suggested Qs | Array | [] | Up to 4 clickable starter questions |
| Temperature | Range | 0.7 | AI creativity vs. factual precision (0.0 - 1.0) |
| Max Context | Number | 5 | Maximum document snippets per query |
Features
- Hybrid RAG search — combines keyword matching with AI context retrieval for precise answers
- AI persona templates — pre-configured prompts for Support, Sales, and Documentation use cases
- Edge-native streaming — near-instant responses powered by edge AI
- Suggested questions — one-click chips that guide users to key information
- Source citations — answers reference specific documents for transparency
Use cases
- Support documentation — instant answers to technical questions from your manual
- Sales intelligence — specialized agent that handles pricing and feature queries
- Knowledge bases — make internal wikis and guides searchable via natural language
- Product docs — help users find integration guides, API references, and tutorials
Embed example
Popup (recommended — floating chat bubble):
<script src="https://cdn.widget.best/embed.js"
data-widget-id="your-doc-chat-id"
data-mode="popup">
</script>Inline:
<script src="https://cdn.widget.best/embed.js"
data-widget-id="your-doc-chat-id">
</script>Document indexing
Upload your documents (PDF, Markdown, plain text) through the dashboard. Documents are automatically:
- Chunked into semantic segments
- Indexed with vector embeddings
- Stored at the edge for fast retrieval
Updates to documents re-index automatically. No manual reprocessing needed.
Technical details
- Documents chunked and indexed on Cloudflare’s edge network
- RAG architecture: retrieval-augmented generation for grounded, accurate responses
- Streaming responses for near-instant perceived latency
- Supports display rules for conditional visibility