Housing · AI agents · Recommendation systems

HomeSwipe Agent

An AI apartment-search agent that learns from swipe behavior and ranks listings with structured data.

Most Shippable · DayDreamers × MentorMates
CASE STUDY03/ 15

Interface-inspired project artwork · not a product screenshot

OVERVIEW

From a difficult problem to a usable system.

HomeSwipe Agent, built under the hackathon name RentOS, turns apartment hunting into a personalized decision workflow. Users review listings through a swipe-based interface, while the system extracts structured listing information, observes preferences, enriches listing data, and improves its recommendations.

THE PROBLEM

What had to change

Apartment listings are fragmented, inconsistent, difficult to compare, and filled with unstructured information.

THE SOLUTION

How the product responds

The application converts listings into structured data, tracks user decisions, scores apartments against individual preferences, and surfaces stronger matches.

TECHNICAL APPROACH

How the system fits together

Listing content is extracted into a consistent schema before entering a preference-scoring pipeline. Swipe signals update a user preference profile, while analytics and observability make the agent workflow easier to understand and debug.

KEY FEATURES

What the product can do

  • Swipe to save or pass
  • Preference learning
  • Structured listing extraction
  • Recommendation scoring
  • Listing enrichment
  • Analytics events and tracing
  • Responsive apartment cards