01CHALLENGE
What the project had to solve.
A real estate agency: over 300 properties for sale and rent, dozens of active buyer and renter inquiries, deals at various stages.
The pains are typical for the market. Properties were overpriced on the owner's emotions — a listing sat for weeks without a single viewing, and nobody saw that the problem was the price. Realtors matched listings to inquiries by hand, keeping in their heads who's looking for what. Properties, inquiries, and deals lived in separate spreadsheets — there was no single picture. And assessing location and liquidity rested on a particular realtor's experience, not on data.
For an agency where income is the commission on closed deals, every overpriced listing that sits is frozen time and lost commission.
02SOLUTION
What we built and why.
Property catalog: sale and rent. 312 properties with filters by district, rooms, area, price, floor, and condition. On each card — a status (sale / rent / exclusive / urgent), the responsible realtor, and an AI flag for whether the price is within market.
A property card with everything about the deal. A photo gallery, specs, documents, viewing history, price history, owner data, and a "similar properties nearby" block. "Schedule a viewing" and "Link to an inquiry" buttons — the property goes into work right away.
Client inquiries and their stages. A database of buyer and renter needs: what they're looking for, budget, district, rooms, stage (new / matching / viewings / negotiation), and realtor. You can see who's at which stage of the funnel.
Property map. All properties on a city map with filters and a price heatmap — the realtor shows the location in the context of the district rather than as a list of addresses.
Deals dashboard. Active properties, new inquiries, deals in progress and their value, closed this month, average time to sell, and trends. The owner sees the funnel and the money on one screen.
AI inside the system — the core of the product. AI property valuation separately assesses the location (transit, schools, shops, environment, safety) and the building (year, type, number of floors, parking, liquidity) across 50+ factors, computes a fair price and gives a verdict ("overpriced by 3–4%"), and recommends the optimal price to close a deal in roughly 30 days. AI district ratings give an attractiveness index across dozens of factors — the realtor argues the price with numbers. AI inquiry matching finds the best properties for a budget and district in seconds instead of a manual search. AI insights on the dashboard automatically flag problems: "3 properties overpriced by 8–12%," "a listing with no viewings for 14 days."
How we built it: a CRM from scratch for the agency's processes + AI modules, an interactive map with a price heatmap, data on the client's server. Development was led by our developer paired with AI-assisted development.
