Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.chatgrid.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
What you’ll build

A structured research pipeline that ingests multiple sources, vectorizes them, and produces an AI-synthesized report with edges linking conclusions back to source material.
  • Batch-ingest 5+ web sources in one API call
  • Vectorize all sources for semantic search
  • AI-generated synthesis across all sources
  • Edges connecting findings to their source nodes

​
Prerequisites

  • ChatGrid API key (get one here)
  • Node.js 18+ or Python 3.10+

​
Step 1: Create a research board

curl -X POST https://api.chatgrid.ai/v1/boards \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "AI Agent Market Research - Q1 2026"}'

​
Step 2: Batch-ingest sources

Use the batch endpoint to add multiple source nodes in one call — faster and atomic. 
curl -X POST https://api.chatgrid.ai/v1/boards/{boardId}/nodes/batch \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "nodes": [
      {"type": "website", "data": {"url": "https://a16z.com/ai-agents-landscape-2026"}},
      {"type": "website", "data": {"url": "https://sequoia.com/article/agentic-ai-market-map"}},
      {"type": "website", "data": {"url": "https://techcrunch.com/2026/02/15/enterprise-ai-agents-funding"}},
      {"type": "youtube", "data": {"url": "https://youtube.com/watch?v=agent-market-deep-dive"}},
      {"type": "website", "data": {"url": "https://mckinsey.com/capabilities/ai/ai-agents-enterprise-adoption"}}
    ]
  }'
Save the node_ids — you’ll use them to create edges in step 5.

​
Step 3: Vectorize all sources

Vectorize each source so AI can search across them semantically. Run them in parallel for speed. 
for url in \
  "https://a16z.com/ai-agents-landscape-2026" \
  "https://sequoia.com/article/agentic-ai-market-map" \
  "https://techcrunch.com/2026/02/15/enterprise-ai-agents-funding" \
  "https://youtube.com/watch?v=agent-market-deep-dive" \
  "https://mckinsey.com/capabilities/ai/ai-agents-enterprise-adoption"; do
  curl -X POST https://api.chatgrid.ai/v1/boards/{boardId}/documents/vectorize \
    -H "Authorization: Bearer YOUR_API_KEY" \
    -H "Content-Type: application/json" \
    -d "{\"url\": \"$url\"}"
done
Vectorization runs asynchronously. For large sources, allow 10-30 seconds before querying.

​
Step 4: Ask AI to synthesize across sources

Create a chat and ask AI for a structured synthesis. It searches all vectorized content before responding. 
curl -X POST https://api.chatgrid.ai/v1/boards/{boardId}/chats \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "Research Synthesis"}'

curl -X POST https://api.chatgrid.ai/v1/boards/{boardId}/chats/{chatId}/messages \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "Synthesize the research on this board into a findings report: 1) Market size and growth, 2) Key players and positioning, 3) Enterprise adoption trends, 4) Gaps and opportunities. Cite which source each finding comes from.",
    "stream": false
  }'
Save the synthesis as a notepad, then create edges connecting it to each source for traceability. 
curl -X POST https://api.chatgrid.ai/v1/boards/{boardId}/nodes \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"type": "notepad", "data": {"content": "# AI Agent Market Findings\n\n[Paste synthesis here]\n\nGenerated from 5 sources."}}'

# Link findings to a source node (repeat for each source)
curl -X POST https://api.chatgrid.ai/v1/boards/{boardId}/edges \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"source": "{findingsNodeId}", "target": "{sourceNodeId}", "label": "sourced from"}'

​
What’s happening under the hood

The batch endpoint creates all source nodes in a single database transaction. Vectorization runs in parallel — each source is fetched, cleaned, chunked into ~500-token segments, and embedded using pgvector. When AI generates the synthesis, it performs a semantic search across all chunks from all sources, retrieves the most relevant passages, and generates a grounded response with citations. Edges are stored as first-class graph relationships, so anyone viewing the board can visually trace a finding back to its original source.

​
Next steps

Client Onboarding

Auto-generate knowledge boards for new clients

Sales Deal Room

Build deal rooms with competitive positioning