Build a multi-agent real estate workflow prototype

A builder used Claude to help create a system where agents handle listing, pricing, and buyer questions for real estate sales.

Done with Claude· Advanced· Mar 2026

What they did
They spent a weekend making SixPercent.ai, a real estate setup with several agents: SellerBot, BuyerBot, TxBot, and MarketBot. Claude handled most of the coding, producing about 27,000 lines across 41 files. The system used OpenClaw for agent file access, web browsing, APIs, memory, Redis event triggers, file state, and task handoffs, while people still made the final call on important decisions and handled self-guided showings.
How it went
The project had only recently gone live, so there were no finished deal results to report. The builder later said the site had bogus stats copied in from an older cached version, and that Claude's output had to be checked repeatedly to catch invented details.
Tips
If you try this, keep the first version limited to mock data and simulated conversations. Make sure a person reviews anything binding before it goes out, and audit the agent's output carefully for made-up details.

Try it yourself with Claude Code

Help me design a prototype real-estate agent workflow for [property type] using [my tech stack]. Create separate roles for [seller tasks], [buyer questions], [pricing research], and [transaction steps], and use mock data only. Show me how the agents should hand off work, store state, and log decisions. Stop and ask for my approval before any step that would send a message, publish a listing, delete data, or submit anything.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.