AIRz

The question

Property searches often start with rigid filters. AIRz explored a more conversational experience, connecting natural-language requests with property information and location context.

My contribution

I was part of the IE capstone team building AIRz for Zyonia, alongside Kylie Koshet, Karim El Achkar, Julio Varela López-Jacoiste, Maya Demeure and Sosina Tefera. The architecture and capabilities shown here are our collective work; the public announcement does not break down individual component ownership.

The approach

• Agent orchestration across strategy, execution and summarisation.
• Retrieval-augmented generation using Azure AI, Google Maps and custom web search.
• Text and voice semantic search, vector embeddings and hybrid memory.
• Administration tools and mapping validation.

What came out of it

The team presented AIRz as its IE capstone for Zyonia. The public project announcement documents the approach; adoption and performance metrics have not been provided.

Explore the source.

Inside the work.