Landlord Negotiation Simulator

The idea

The first lease negotiation many renters experience is also the first one they have ever had to prepare for. At QHacks 2025, we built a simulator where someone could practice that conversation before money, housing, and a real landlord were involved.

The product placed the user in an interactive negotiation and returned AI-generated landlord responses as the conversation developed.

Designing the practice flow

I designed the end-to-end experience in Figma before building the interface. The flow needed to make the scenario understandable, keep the conversation moving, and avoid burying a first-time renter under controls.

The point was practice, not legal advice or a guaranteed script. The interface had to feel low-stakes enough for someone to try a response, see what came back, and keep going.

Building it in 36 hours

Frontend

I built the responsive interface with Next.js, React, and TailwindCSS. The conversation view had to stay readable as messages accumulated and still work on a smaller screen.

Backend

A Flask backend connected the interface to OpenAI for the simulated landlord responses. Keeping that work behind an API boundary let the frontend focus on the conversation state and presentation.

What shipped

By the end of QHacks, we had a working negotiation loop that connected the Figma flow, responsive frontend, Flask service, and AI responses.

The project taught me that conversational products need more than a text box. The framing, expectations, and transitions around each response determine whether the interaction feels like useful practice or random generated dialogue.

Limits

This was a hackathon simulator, not a source of legal guidance. A longer build would need clearer jurisdictional boundaries, stronger safety language, and a more deliberate way to evaluate the quality of a negotiation.