FlipperZillow

• 

  aerial house render

Inspiration

We're in our twenties and homeownership already feels out of reach. Leveraging AI-powered semantic search, powerful graphical processing units, and transcribing complex conversations became a something we could leverage to help our future.

What it does

FlipperZillow takes any address, pulls real listing photos via RapidAPI, runs them through an AI vision pipeline to analyze each room's condition and value signals, and delivers a fully narrated 3D spatial tour with an honest, AI-powered breakdown of a property's strengths and red flags.

How we built it

We started super simple, expo marker to whiteboard, drawing out ideas of key pillars we wanted to integrate. Initially, we had 3 services, a frontend with some type of search engine, a compute engine for determining the best house option, and lastly a AI-powered chatbot for interpretation. As we hashed out ideas, we increased to 9 pillars, focusing on the pursuit of leveraging the technologies we were provided.

Challenges we ran into

Our biggest challenge was finding a working dataset, an initial promising GitHub dataset had no address ties, Zillow rate-limited and banned our scraper, and Redfin turned out to be equally off-limits, leaving us back at square one until Jake, the ACM president, pointed us back toward RapidAPI, which we'd nearly overlooked entirely and ended up being the solution that unlocked the whole pipeline without violating a single terms of service.

Accomplishments that we're proud of

We are proud of our multi agent pipeline that came together to tie all of these technologies together. Our problem solving rigor through all of the difficulties that came with developing/finding an API for sourcing the data. Most of all, staying strong and staying awake for the whole hacking duration.

What we learned

How much exposure to different tools and resources you can get your hands on if you put your mind to it (and feel the pressure of a deadline). Spending more time on planning than coding is critical to a strong foundation. Most importantly, do not bite off more than you can chew, adding more features != better product.

What's next for FlipperZillow

FlipperZillow will eventually become a tool for realtors and housing market goers to tour both the inside and outside in a 3D rendered space, with focus on identifying flaws within overvalued properties as well as finding overall undervalued properties.

Built With

• amd
• depthv2
• elevenlabs
• fastapi
• python
• rapidswitch
• sam3d