A poster built a platform where agents, including a SellerBot, handle listings, pricing strategy and buyer inquiries, built with Claude as the coding assistant.
A poster built a platform where agents, including a SellerBot, handle listings, pricing strategy and buyer inquiries, built with Claude as the coding assistant.
The builder says they spent a weekend creating SixPercent.ai as a multi-agent real estate system with SellerBot, BuyerBot, TxBot, and MarketBot. Claude wrote about 27,000 lines across 41 files, while OpenClaw gave the agents file access, web browsing, APIs, persistent memory, Redis-driven events, file-based client state, and handoffs; the person says humans still approve binding decisions and handle self-guided showings.
The post frames it as launched only a few days earlier, so it does not report completed transaction results. In comments, the builder says the legal close stays with title companies and attorneys, and later says they pivoted to a hybrid model in Costa Rica with five autonomous agents supporting human agents.
Big catch: the builder says fabricated site stats slipped in from a previous cached iteration, and that Claude's work needed repeated auditing to catch made-up details.
Help me build a prototype where AI agents handle parts of a real estate sale: a seller agent that suggests listing prices and strategy for [property details] and a buyer-inquiry agent that answers questions from [sample inquiries]. Use [my tech stack] and mock data only. You're done when I can run a simulated conversation end to end, and nothing should send real messages or publish real listings.
Score every demo request against your ICP, update HubSpot and ping the right rep in Slack within a minute of the form submit.
Every new Zendesk ticket gets a category, priority and sentiment tag plus a draft reply in an internal note before an agent opens it.
Copilot and Claude review each PR for bugs, missing tests and risky changes before a human looks, so reviewers can focus on design.