Mixed-language slang returned zero results. A non-engineering team deployed AI in 3 days to decode queries, raising search yield to 94%.
A Hong Kong real estate platform serving 1.5 million monthly active users in a fragmented market where listings range from parking spots to vacant lots.
High-intent buyers often input complex queries mixing English acronyms, Chinese characters, and slang, such as "TKO 廣明苑 1 bed $3K." The legacy...
“In the past, property listings were scattered. We want to address the gap between user expectations and quantifiable clicks that connect searchers with representing agents through technology in order for their property search process to be more efficient. Ultimately, we want to align demand with supply to ensure that no enquiry is left behind.”
Online real estate portal for buying, renting, and listing properties in Hong Kong.
Enterprise software, cloud infrastructure, and consumer electronics platform.
Related implementations across industries and use cases
Property details were buried in text. AI now extracts attributes to enrich data by 40%, driving a 98% search match rate.
Millions of requests choked self-hosted search. Hybrid vector search now powers 10k daily AI tasks with 40% higher accuracy.
Basic scripts failed complex queries. Now, an AI agent on WhatsApp automates property searches and viewing bookings instantly.
Rigid filters forced buyers to act like admins. Now, they describe a "cottage feel," and AI scans millions of listings to find a match.
Local modeling limited output to 5 projects. Now, 2,000 users build their own models on a shared platform, scaling production 12x.
Lab supply orders were handwritten in notebooks. Digital ordering now takes seconds, saving 30,000 hours for research annually.
Experts spent 15 minutes pulling data from scattered systems. Natural language prompts now generate detailed reports instantly.