HelpRent Da Nang: AI rental agent, matching engine and agent-facing API

Rental supply in Da Nang is spread across Facebook groups, local classifieds and building managers, mostly in Vietnamese. Newcomers can't compare prices, can't talk to owners in their language, and AI agents have no clean source to query.

- AI assistant Mai acts as a rental agent: understands requests in any language an LLM supports, narrows the whole city down to the two or three places worth a look, answers questions on deposits, contracts and scams, and with one tap sends the owner a message in the owner's own language on Zalo or WhatsApp. Works on the site and in Telegram. - Matching engine under Mai; search also by a screenshot of a map with an area circled, read by a vision model. - Aggregation and normalisation of public listings into one catalogue, with median rents by size and district recomputed nightly. Site interface in English, Vietnamese and Russian. - MCP server with four read-only tools (search listings, get listing, find buildings, market stats) and a REST API with an OpenAPI spec, no key or account needed. - Pay-per-request for agents: a free daily allowance, then HTTP 402 with an x402 challenge settled in USDC on Base or Solana; Solana Pay for people. - Read-only by design: no write tools and no contact details; agents send the person to the listing page. Machine-readable throughout: llms.txt, schema.org on listings, ItemList on catalogues.

One searchable layer over a fragmented rental market, usable in any language; prices anyone can cite from the market page; a live example of an AI agent paying for data over x402.