Let agents find US doctors with checkable evidence over MCP
A developer built a remote MCP server with 1.2M US physicians, Medicare procedure volumes and facts from hospital sites.
An open-source n8n copilot reads lead emails, assigns them to agents round-robin, drafts replies and sends nothing until the agent taps Send in Telegram.
It watches Gmail, extracts and validates leads with OpenAI plus rules, dedupes, stores them in Supabase, assigns agents, and nudges about unanswered leads at 30 minutes, 2 hours and 24 hours. Agents manage leads in natural language on Telegram.
MIT-licensed with workflows, schema and setup guide.
The human Send step keeps the AI from emailing clients on its own.
Our own starting point, written from the source. Adapt it to your setup.
Is this email a real estate lead? If yes return JSON {type: sale|rental, intent: viewing|requirements|general, name, phone, property_ref}.A developer built a remote MCP server with 1.2M US physicians, Medicare procedure volumes and facts from hospital sites.
A hotel analyst built an n8n workflow to check housekeeping staffing needs with data when staff call in sick.
A builder made an n8n workflow where AI drafts customer email replies and you approve, edit or reject them.
A hobbyist spent months getting ChatGPT and Claude to write LDraw, the instruction language for LEGO CAD models, and open-sourced the tooling.
Five real AI agent use cases each morning, each with a recipe and a source. Plus a Sunday roundup. Free.