RentoAI

Inspiration

Struggle of outstation students to settle in a new place.

What it does

Our web app helps outstation students find ideal accommodations like PGs and rentals effortlessly. Powered by GenAI, it generates SQL queries based on user preferences to fetch relevant listings. A chatbot provides instant property details, while a premium model allows sellers to promote listings for better visibility, ensuring a seamless housing search experience.

How we built it

GenAI: Generates SQL queries based on user preferences. SQL Database: Fetches relevant properties efficiently. Chatbot: Engages users, answering queries about properties.

Challenges we ran into

Building this platform wasn’t all smooth sailing—we hit some real challenges along the way, but each one pushed us to think deeper and refine our solution.

Integrating PostgreSQL & Managing the Database Setting up PostgreSQL with Prisma and TypeORM sounded simple in theory, but in practice, we ran into query optimization issues and struggled with data relationships. Fetching relevant properties efficiently was tricky, especially as the dataset grew. We solved this by fine-tuning our SQL queries, indexing key fields, and caching frequent searches to make responses faster.

Making the Chatbot Actually Useful At first, our chatbot felt more like a FAQ bot than an intelligent assistant it was either too generic or got stuck on complex queries. The challenge was getting it to understand real user needs and respond with relevant property details. We refined it by integrating LangChain, improving the prompts, and feeding it contextual data so it could return meaningful, accurate responses instead of vague or incorrect answers.

Building User Trust & Engagement One of our biggest challenges was making users trust the listings and feel confident in using the platform. People are often skeptical about online rental platforms due to fake listings and outdated data. To tackle this, we prioritized verified listings, added user reviews, and introduced premium features for serious sellers, ensuring that buyers see only genuine, quality options.

Balancing Free vs. Paid Features We wanted to monetize the platform but also keep it accessible. If we locked too many features behind a paywall, users might leave; but without a revenue model, sustainability would be an issue. We found a middle ground by offering free basic searches while giving premium users boosted visibility and exclusive features, making it fair for both sides.

Accomplishments that we're proud of

Making the AI functional

What we learned

We dived into all the tech stack in detail. it was our first time working with OpenAI.

What's next for RentoAI

We'll work more on the revenue model

Built With

• clerk
• gptmini
• neon
• nextjs
• openai
• postgresql
• prisma
• react
• shadcn
• tailwind