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.
Real estate · Do this yourself
Run a real estate agency’s lead follow-up from Telegram.
✅ MIT-licensed with workflows, schema and setup guide.
Who it’s for
Real estate agents, investors, property managers and renters or buyers.
Adapt it to your industry
- Agent: listing descriptions, social posts and follow-up emails
- Property manager: tenant question answers and maintenance logs
- Investor: a quick deal check using public numbers
Step by step
- Pick one task: a listing write-up, a buyer follow-up, or a property comparison.
- Open ChatGPT, Claude or Grok and paste the prompt below with the property details (no private client data).
- Ask for 3 versions in different tones and pick the best.
- Check every fact (size, price, features) against your listing; AI can get numbers wrong.
- Save your best prompt as a template for the next listing.
- Track which posts or emails get replies, and ask the AI to improve the weak ones.
Ready-to-copy prompt
Is this email a real estate lead? If yes return JSON {type: sale|rental, intent: viewing|requirements|general, name, phone, property_ref}.Start with AI
Opens the AI with this guide and the prompt filled in.
- Cost
- $0–20/month (free tiers or one AI plan)Our estimate
- Time
- 2–6 weeksOur estimate
How to make money from it
Pricing idea: If you help other agents, charge per listing package (for example $50–200) or a monthly fee.
Finding your first 10 customers
- Offer free listing write-ups to 3 agents in your office.
- Share before/after examples on LinkedIn or Instagram.
- Ask your brokerage to share it with the team.
Ideas, not promises. Results vary.
Details: source, numbers and tags
What they did
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.
How it went
MIT-licensed with workflows, schema and setup guide.
Worth knowing
The human Send step keeps the AI from emailing clients on its own.
- Difficulty
- AdvancedOur rating
- Time to build
- Not reportedOur estimate
- Monthly cost
- Not reported
- Revenue
- Not reportedWe only show revenue the source states
- Engagement
- 2 Reddit pointsfetched Oct 15, 2026
Original builder recipe
- n8n
- OpenAI API
- Supabase
- Telegram bot
- Gmail
- Connect listings, calendar and messaging.
- Let the agent handle routine questions and scheduling.
- Keep a human on negotiations and compliance.
- Log every interaction.
Tags
Read the original
- Reddit: www.reddit.com/r/VibeCodeDevs/comments/1wa7y5y/opensource_ai ↗ (opens in new tab) Sep 7, 2026 · 2 points (fetched 2026-10-15)
More ideas like this
Check whether you really need replacement staff before calling them in
A hotel analyst built an n8n workflow to check housekeeping staffing needs with data when staff call in sick.
Draft email replies with AI and keep a human approval step
A builder made an n8n workflow where AI drafts customer email replies and you approve, edit or reject them.
Generate LEGO models from a prompt
A hobbyist spent months getting ChatGPT and Claude to write LDraw, the instruction language for LEGO CAD models, and open-sourced the tooling.
The Daily 5.
Five real AI agent use cases each morning, each with a recipe and a source. Plus a Sunday roundup. Free.