Last week I shipped something I'm genuinely excited about: a Model Context Protocol server that lets you search 130M+ B2B contacts and enrich profiles with verified emails & phone numbers — directly from Claude, Cursor, Windsurf, or any MCP-compatible AI tool.
No REST calls. No Postman. Just ask your AI assistant in plain English.
What It Looks Like
Imagine typing this into Claude:
"Find me marketing VPs at SaaS companies in Austin with 50-200 employees"
And getting back real, verified contact data — names, titles, emails, direct dials, LinkedIn profiles — right in the chat. That's what Bytemine MCP does.
Why MCP?
If you haven't been following, MCP (Model Context Protocol) is Anthropic's open standard for connecting AI assistants to external data sources. Think of it as USB-C for AI tools — one protocol, many connections.
Most B2B data providers still require you to:
Log into a dashboard
Run a search
Export a CSV
Import it somewhere else
With MCP, the data comes to you — inside the tool you're already working in.
Setup (2 Minutes)
Option 1: npx (Fastest)
npx bytemine-mcp --api-key YOUR_API_KEY
Option 2: Claude Desktop
Add this to your claude_desktop_config.json:
{
"mcpServers": {
"bytemine": {
"command": "npx",
"args": ["-y", "bytemine-mcp", "--api-key", "YOUR_API_KEY"]
}
}
}
Restart Claude Desktop. Done.
Option 3: Direct HTTP (Cursor, Windsurf, etc.)
Point your MCP client to:
https://bvjmtgaxijpyasjtaqiv.supabase.co/functions/v1/mcp-server
Transport: Streamable HTTP
Auth: Pass your api_key as a parameter.
Available Tools
The server exposes two tools:
people_search
Search 130M+ contacts with 20+ filters:
- Job titles, seniority, departments
- Company name, industry, revenue, employee count
- Location (state, city)
- Skills, interests, schools
- Has phone, has email, has direct dial
contact_enrich
Get the full profile for a specific person:
- Verified work & personal email
- Direct dial, mobile, work phone
- Full job title, seniority, department
- LinkedIn, Twitter, Facebook
- Company details & address
Each operation costs 1 credit. Enrichment only charges on a successful match.
Real Use Cases
Here are prompts I use daily:
Prospecting:
"Find CTOs at healthcare startups in New York with 10-50 employees"
Pre-meeting research:
"Enrich this LinkedIn profile: linkedin.com/in/janedoe — get me her email and phone"
List building:
"Search for HR directors at companies with 500+ employees in California"
Competitive research:
"Find engineers at [competitor] who have 'machine learning' in their skills"
How I Built It
The MCP server runs as a Supabase Edge Function using mcp-lite + Hono. The architecture is simple:
AI assistant sends a tool call via Streamable HTTP
Edge function validates the API key and checks credits
Queries the Bytemine contact database (130M+ records)
Returns structured results back to the AI
The whole server is ~200 lines of TypeScript. MCP's spec is clean enough that you don't need heavy frameworks.
Get Started Free
The Developer Suite is free for 12 months — 500 credits/month, full API access, no credit card required.
Sign up at bytemine.ai
Go to Settings → API Keys → Create a key
Run npx bytemine-mcp --api-key YOUR_KEY
Links:
🐙 GitHub
📚 API Docs
🌐 Bytemine.ai
If you're building MCP servers or doing anything interesting with B2B data + AI, I'd love to hear about it in the comments. 🤙
Koko Zarov
CEO of Bytemine
Top comments (0)