CloseSure | Devpost

Inspiration

Every year, first-time homebuyers face the same impossible choice. Spend $300 to $600 on a home inspection before knowing anything about the property, or go in completely blind. When you're browsing 5 to 10 homes, you can't afford to inspect them all. And most first-time buyers don't know what to look for. For example, they don't know if a pre-1980s house has lead paint.

We realized there's a massive gap between browsing on Zillow and paying for a full inspection. No free tool exists that gives buyers property-specific risk information at the browsing stage. We built CloseSure to fill that gap.

What It Does

CloseSure is a free pre-screening tool that helps you quickly understand a property’s history before buying. You simply enter a property address, and CloseSure pulls real public permit records while combining them with the home’s age and other details. Using Google’s Gemini API, it creates a clear, plain-English report that highlights three key areas: Good Points, showing updates and improvements backed by permit dates, Bad Points, flagging risks like missing permits or age-related issues, and Questions to Ask, giving you targeted queries to bring to your agent or the seller, all tailored specifically to that property.

How We Built It

Backend: The system is built with Python and FastAPI and includes a permit lookup service that queries public municipal records by address. An address parser validates and normalizes user input before the AI service structures the permit data into a prompt for Google’s Gemini API. The Gemini prompt is carefully designed to return structured JSON with Good Points, Bad Points, and Questions, all grounded in real, verified data to give users a clear and actionable property report.

Frontend: The frontend is built with Next.js and Tailwind CSS and features four main pages: a landing page, an address input page, an AI-powered report, and an inspector discovery page. The AI report uses color-coded sections with green for good points, red for bad points, and blue for questions, along with category tags for each finding, such as Structural, Roofing, Plumbing, or Electrical. Users can hover to expand detail panels for progressive disclosure, while scroll-triggered animations and glassmorphism styling add a polished, interactive feel throughout the site.

Challenges We Ran Into

We faced several challenges while building the project. Permit data availability was a key concern, as not every municipality has digitized, accessible records, for the hackathon, we scoped the project to Los Angeles but designed the architecture to be modular for expansion to other cities. AI accuracy was another priority, ensuring that Gemini’s analysis stayed grounded in real permit data rather than generating plausible-sounding but incorrect findings, which we addressed by structuring prompts to reference specific permit dates and types. Finally, scoping for 36 hours required hard decisions, we had ideas for dashboards, saved reports, and property comparisons, but chose to focus on polishing a single end-to-end flow rather than delivering multiple half-finished features.

Accomplishments That We're Proud Of

Some of the highlights from the project are a fully working end-to-end flow that takes users seamlessly from entering an address to receiving an AI-generated report backed by real permit data, a report page that makes complex information feel simple and approachable for first-time buyers, and an ADU/permit verification feature that flags unpermitted additions, something no other tool currently offers. All of this was built and integrated by a four-person team within just 36 hours.

What We Learned

Some key lessons we learned from the project are that scoping aggressively is crucial, as one polished feature is far more valuable than several half-finished ones, and agreeing on data contracts early helps when building the frontend and backend in parallel. Beyond the technical side, we gained a real sense of how the real estate world works, what buyers actually need from inspections, and where the current home owning process falls short.

What's Next for CloseSure

Future plans for the platform include a saved reports dashboard that lets users compare multiple properties side by side and integration with Zillow and Redfin to enable pre-screening directly from property listings. We also aim to expand permit coverage across more cities and municipalities beyond Los Angeles and optimize the experience for mobile by converting hover interactions into tap-to-expand. Additional goals include forming inspector partnerships for a referral-based revenue model that keeps the tool free for buyers, as well as incorporating historical sales data and neighborhood risk scoring to provide deeper property intelligence.