real-estate-agent

It is an Agent with Long Term Memory of a property buyer's preferences and available houses in their location. And flexibility to find properties based on specific requirements. To save properties and schedule property tours. It is a functional Agent that showcases Capella's diverse data platform capabilities: operational/KV, vector, geospatial and built in caching, to find, serve and store unique properties for a buyer at speed.

An Agent with Long Term Memory of:

• Available Properties
• Buyer's Preferences
• Saved Properties
• Scheduled Tours

Single Agent Architecture (for now)

Routes:

1. Direct to LLM - us.meta.llama4-maverick-17b-instruct-v1:0, for general queries e.g. mortgage approval process.
2. CB Vector Search Tool - finding properties based on cosine similarity of text embeddings.
3. Gmaps API/Geospatial Search - using lat, lon of property addresses, to find nearby schools and restaurants.
4. Realtime web search using Tavily API - for current housing market trends, average house prices.

Americas AI Challenge Team:

1. Kevin Farley
2. Dan James
3. Seong Cho
4. Jake Wood

Models Used:

• Text Embedding Model: Titan Text Embeddings V2 (1024 dimensions). Similarity Metric: Cosine for text similarity in a RAG application.
• Large Language Model: us.meta.llama4-maverick-17b-instruct-v1:0

Couchbase Capella Data Structure:

• Bucket: properties

  • Scope: 2025-listings
  • Collection: united-states

• Bucket: profiles

  • Scope: buyers
  • Collection: 2025

• Bucket: profiles

  • Scope: tours
  • Collection: 2025