EquityNest

Inspiration

We, and many other college students across the country, feel worried about rising costs and whether we’ll ever afford the futures we’re working toward. Since 2020, home prices have surged, first-time buyers keep getting older, rates swing, and the rare “good deal” is gone before most people even see it. We built EquityNest because working buyers shouldn’t be priced out by guesswork or hidden models. If undervalued homes exist, people deserve a clear, fair way to find them first. So we turned messy real-estate data into something simple you can trust: attom’s price side-by-side with our equitynest value, a quick analysis of repair and costs, and plain-english guidance to help take away from the stress of finding a home.

What it does

EquityNest AI scans listings and public records to flag properties priced below their true value, then presents a simple, trustworthy picture of the opportunity. For each home, we show an ATTOM price side-by-side with our EquityNest value, which is a more accurate estimate produced by our calculator. EquityNest also utilizes a chatbot that provides you with any help that you may need. An integrated AI agent prioritizes the best candidates, explains exactly why our value differs from ATTOM’s, and guides next steps, from tightening search criteria to drafting a negotiation game plan.

How we built it

We built EquityNest on a 3-agent AI architecture powered by Google’s Gemini models, combining real estate expertise with cutting-edge AI. Agent 1 manages customer interactions, Agent 2 performs professional-grade property analysis using standard appraisal methods, and Agent 3 monitors listings for opportunities. With ATTOM’s property data and a custom comparables engine, our system normalizes price per square foot and adjusts for key factors like condition, lot size, and timing. A multi-factor scoring engine evaluates price, cash flow, market conditions, and risk, while structured Pydantic schemas ensure consistent outputs. On the frontend, a clean HTML/CSS/JS marketplace features filterable grids, map integration, and real-time chat that translates complex analysis into plain English, making investment insights clear and accessible.

Challenges we ran into

Working as a four-person team running into frequent merge conflicts, so we adopted a clean branching strategy and stayed disciplined about reviews. Tuning valuation across neighborhoods was hard, and we had to balance raw margin with confidence so results felt credible rather than noisy. On the front end, refactors occasionally broke interactions; we stabilized them with loading states, consistent event handling, and small tests around key components. Integrating the AI agent also required thoughtful prompts and guardrails so explanations stayed concise, consistent, and aligned with the math.

Accomplishments that we're proud of

We shipped an immersive marketplace that feels fast, looks polished, and tells a clear story about each property. The Deal Analyzer calculates ARV ranges, rehab costs, and net margin in a way that matches what users see on the cards. The brand and UI feel professional and approachable, and the AI agent adds real value by turning numbers into guidance without overwhelming the user. Most importantly, the Equity Score aligns with the analyzer’s outputs, so when the agent recommends a deal, the rationale and the math line up.

What we learned

We learned to divide work well and stay calm under pressure. We started each task from a clean branch and talked early when conflicts came up. We saw that clear explanations build trust. Showing the comps, a confidence label, and a one-line reason helps more than chasing tiny gains in accuracy. The AI agent works best when it repeats those same explanations in plain language. We also learned how much small UX touches lift the product. Skeleton loaders, simple toasts, and friendly copy make the app feel faster and more human.

What's next for EquityNest

Next, we will forecast the seller’s exit month by combining metro and ZIP-level trends with seasonality, adjusting for liquidity and insurance risk, and broadening our data sources to include school, crime, and photo-quality signals. We plan to let the AI agent run deal-finding sweeps on a schedule, deliver daily equity alerts by city or ZIP, generate shareable deal packets, and offer guided onboarding so new users can run their first search with coaching and understand exactly how the system reached its recommendation.