What the GTM Intel MCP Server Does
The MCP server connects your AI agent directly to the GTM Intel startup funding database — 19,000+ companies that raised funding, enriched with CEO names, emails, LinkedIn profiles, HubSpot detection, and company summaries. Your agent can search, filter, and analyze funding data in real time.
This is the same dataset that powers our programmatic funding pages — exposed as tools your agent can call autonomously.
Available Tools
Once connected, your agent can use these 7 tools:
get_recent_funding
Retrieve recent funding events with optional filters for days, industry, stage, and country.
search_funding
Search funding events by matching against company name, industry, or location.
get_funding_by_company
Get all funding events for a specific company by name.
get_funding_stats
Get aggregated statistics — total events, unique companies, top industries.
get_industries
List all unique industries in the funding database.
get_funding_stages
List all unique funding stages present in the database.
get_countries
List all unique countries where funded companies are located.
Installation
The MCP server runs as a local binary. You build it from source, then point your AI client at it.
1. Clone and Build
git clone https://github.com/dburk/gtmintel.git
cd gtmintel
# Build the MCP server binary
cargo build --bin mcp_server --release
2. Set Up Environment
Copy the binary wherever you like (e.g., ~/bin/gtmintel-mcp). The server needs a DATABASE_URL pointing to the GTM Intel Supabase Postgres instance. Contact [email protected] for access credentials.
3. Configure Your AI Client
Claude Desktop
Edit ~/Library/Application Support/Claude/claude_desktop_config.json:
"mcpServers": {
"gtmintel-funding": {
"command": "/path/to/mcp_server",
"env": {
"DATABASE_URL": "postgresql://..."
}
}
}
}
Claude Code
Set the CLAUDE_CODE_MCP_SERVERS env var or add to your project config:
{
"mcpServers": {
"gtmintel-funding": {
"command": "/path/to/mcp_server",
"env": { "DATABASE_URL": "postgresql://..." }
}
}
}
Cursor
Add to your project's .cursor/mcp.json:
"mcpServers": {
"gtmintel-funding": {
"command": "/path/to/mcp_server",
"env": { "DATABASE_URL": "postgresql://..." }
}
}
}
Codex (opencode)
Add to your project's .codex/mcp.json:
"mcpServers": {
"gtmintel-funding": {
"command": "/path/to/mcp_server",
"env": { "DATABASE_URL": "postgresql://..." }
}
}
}
Example Queries
Once connected, your agent can answer questions like:
- "What AI startups raised funding in the last 30 days?"
- "Search for recently funded fintech companies in the UK."
- "Show me all funding rounds for companies in San Francisco."
- "What industries have the most funded startups?"
- "Get details on Anthropic's funding history."
- "Which funded startups use HubSpot?"
- "List companies that raised over $100M."
Current Limitations
- Funding data only. The MCP server covers the funding database (19K companies). Lead-gen signals, CRM sync, email sequences, and the review queue are not yet exposed as MCP tools — coming next.
- No auth built in. Authentication is delegated to the MCP transport layer. Anyone with the binary and database URL has full read access.
- STDIO transport only. The server uses standard input/output — no HTTP/SSE transport yet. All major AI clients support this.
Pricing
The MCP server is free to connect and use for cached reads from the database. Fresh scans, enrichment, and premium data lookups consume credits on paid plans. See pricing for details.