The client
A property agency in Bangkok specializing in the expat rental and sales market. Around twenty agents working a mix of WhatsApp, Line, and email enquiries that came in through their website, partner referrers, and a handful of property portals.
The problem
A typical Monday morning meant somewhere between forty and seventy new enquiries waiting across five different inboxes. Agents were spending the first two hours of the day reading messages, deciding who should handle each one, then searching the listings database for properties that might fit. Plenty of viable matches were getting buried, and prospects were going cold by lunchtime.
The approach
The build pulled every enquiry channel into HubSpot through Zapier, normalized the incoming text, and ran each message through an LLM to extract structured requirements: budget range, area, bedrooms, lease versus purchase, move-in window, family size, pets. With those tags in place, the system queried the listings database for the top three to five matches and posted them, along with the original enquiry text and the suggested matches, into the right agent’s Line inbox.
The agent stayed in charge of the actual reply. What changed was that they were now opening their morning to a sorted queue of qualified enquiries with three matching listings already pulled, instead of an unsorted inbox.
The result
Median time to first match went from around four hours to under five minutes. Conversion on the high-budget enquiries lifted measurably because the agency was now responding before competitors did. The agency’s principal stopped his Saturday morning ritual of triaging the weekend backlog, because the system had already done it.
"I came in expecting a chatbot. What we got instead was a system that quietly does the work we were already doing, just instantly. Twenty agents, sorted morning queues, every enquiry pre-matched to listings. The senior partners stopped running their old weekend triage."