What Tedros AI does today
Everything on this page is live in the product. Run a search to see it on real data.
One search, four sources
Enter a city and state or a ZIP code. Tedros AI queries Zillow and Realtor.com by place, works out the map area those listings cover, then queries Redfin and Apartments.com across that same area, all in parallel.
Cross-site de-duplication
Most homes are listed on more than one portal. Addresses are normalized (street suffixes, directions, unit numbers) so the same home becomes a single card that shows each site's asking price and link.
Price-mismatch detection
When the same home carries a different price on different sites, often because one portal has not picked up a price cut yet, the card is flagged with the difference, and you can sort by the biggest gaps.
Market KPIs
Median asking price, median price per square foot, median days listed, number of price reductions, and how much of the inventory appears on multiple sites.
Rental comps and gross yield
Live apartment-community rents from Apartments.com, broken down by bedroom count, compared against median asking prices for homes with the same bedroom count to estimate gross rental yield.
AI market brief
Google Gemini (via Vertex AI) reads the computed statistics and writes a short plain-English summary of the market. It is instructed to use only numbers present in the report.
CSV export and search history
Download merged listings or rental comps as CSV for your own models. Signed-in users get a dashboard with their recent searches and daily usage.
Built on our own data infrastructure
Tedros AI runs on Google Cloud: the web app on Cloud Run, accounts and usage in Firestore, the AI brief on Vertex AI, and our own listing-data services on Compute Engine. Search results are cached for a few hours so repeated searches are fast, and per-user rate limits keep load on the listing sites reasonable.