Lead generation over MCP
Model Context Protocol lets an AI assistant use Cold Leads as a tool. Instead of copying lists between a spreadsheet, a verifier and a mail client, you ask the assistant — and approve what it sends.
A typical session
You: “Here's the list from yesterday's meetup. Import it, check the addresses and draft a short follow-up.” The assistant imports the rows (import_contacts), verifies each address (verify_email), saves a template (save_template) and prepares a campaign draft for the tagged contacts (create_campaign_draft). You read the text and the recipient count, say “send it”, and the assistant launches the campaign (launch_campaign). Replies land in the CRM, where the assistant can read them (get_conversation) and draft answers.
Connect in a minute
ChatGPT signs in with OAuth. Claude Code and other HTTP MCP clients use a secret key; Claude Desktop, Cursor, Windsurf and VS Code can run the open-source stdio server. All the setup details, tools and examples are on the AI agents page.
Honest limits
The MCP server works with your own contacts: Cold Leads has no third-party lead database and no enrichment. verify_email checks one address per call (1 credit); for lists of thousands use the bulk API or the web app.
FAQ
Where is the full tool reference?
On the Cold Leads for AI agents page (/solutions/for-ai-agents), and the live list is returned by tools/list at the MCP endpoint.
Which plan do I need?
Business, $99/month with 10,000 credits.