NUU
A rental search for the Australian market: renters describe what they want in their own words and get scored matches.
Client
NUU Systems
Industry
PropTech / Real Estate
Timeline
MVP Complete
Year
2025
Services
Brand Strategy · UI/UX Design · Design System · Frontend Development · Backend Development · AI/ML Integration · Database Architecture
01 / The Challenge
Understanding the problem
Renting in Australia is competitive, and searching is slow: listings are spread across platforms, and search filters ask for technical criteria rather than what renters actually want, such as 'a beachy vibe near good coffee shops' or 'somewhere quiet for my family near trains to the CBD'.
02 / The Approach
How we tackled it
We designed the search as a conversation instead of a form. GPT-4o acts as a property concierge, turning plain-language answers into structured preferences, and a weighted scoring model rates each property on seven factors: location, budget, features, amenities, transport, property type and bedrooms. The visual identity borrows from industrial operating systems, to set NUU apart from typical listing sites.
03 / The Solution
What we built
A full-stack platform: a GPT-4o conversation finds what a renter wants in two or three exchanges, and the matching model scores properties on seven factors, with a bonus for exceptional matches. The front end pairs a Three.js architectural scene and scroll-driven animation with a dark, industrial design language. The back end runs on Vercel serverless functions and Supabase PostgreSQL, with pgvector for semantic search and PostGIS for location queries.
Tech Stack
04 / The Outcome
Results & impact
The MVP is complete and live at nuu.agency. A renter describes what they want in their own words, sees scored matches with the reasons behind each score, and refines from there. Vector search is set up so live property feeds can be connected next.
2–3
Exchanges to a match
7
Scoring factors
