RentRightly

Inspiration

As college students, we have experienced how housing contracts can be very long, and it can take a long time to find the answer to a question like "Will I be able to bring my pet?". That's why we created RentRightly - an AI tool for getting clarification about lease and rental agreements.

What it does

• For student renters - we use AI to help you summarize and chat with your contracts. Our Q&A interface can give you answers to specific questions about your contract.
• For landlords - we use AI to find blind spots in your contracts that require clarification and to help you draft the perfect housing contract.

How we built it

We use Google Cloud vertex AI to convert PDF documents into OCR text. Then, we use the Cloudflare AI worker agents and OpenAI GPT-4 to summarize the document and let you ask questions about it. We used streamlit as it provides an easy user interface to host and build AI tools.

Challenges we ran into

Getting cloudflare AI worker agents to run was tough because they depend on open-source models (namely, Meta's Llama models). For the OpenAI we initially struggled with context window length since contract documents are often lengthy. As a solution, we ultimately shifted to using the latest GPT-4 models.

Accomplishments that we're proud of

We were able to ideate and build all of this in less than 24 hours and our prototype actually works!

What we learned

We learned about how to build and deploy streamlit apps, how to use Google Cloud APIs, and also the Cloudflare AI worker agents. We also, unfortunately, learned that getting a react web app interface for everything was too much to get fully functional in just 24 hours with our limited experience, so we resorted to Streamlit.

What's next for RentRightly

We can build out our web app's functionality to align it with our Streamlit application.

Built With

• cloudflare
• google-cloud
• openai
• react
• restful
• vertex