How this site runs

1. The data comes straight from the MLS. Listings arrive through MLS GRID’s licensed data feed — the same source the MLS itself publishes — refreshed every few hours. Nothing here is typed in by hand or scraped off another site.
2. A private mirror keeps it current. Each refresh updates our own copy of the listings; when the MLS changes or removes a listing, the mirror follows in the same cycle, so what you see tracks what’s actually on the market.
3. Every market number is computed, not estimated. Medians, counts, and the other market figures are calculated by the database in plain arithmetic over real records — an AI model never writes a market number. If the data isn’t there yet, the page shows nothing rather than a made-up figure.
4. The whole site rebuilds itself. When the data changes, every page is regenerated from scratch and redeployed — timestamped, so you can see how fresh each number is. There is no live database call between you and a stale page; the page you load was already rebuilt with the latest data.
5. A watchdog checks it hourly. A separate monitor confirms the feed is healthy every hour. If a refresh ever fails and the data goes stale, the site says so out loud — a visible banner — instead of quietly showing yesterday’s numbers as if they were today’s.

The broker’s entire monitoring duty is a short weekly digest of the pipeline’s health. The machine does the watching; a person reads one email.

• It never publishes on its own. The studio’s tools research and draft the guides and neighborhood chapters — but they never hit publish. Every change to what this site says is reviewed, edited, and merged by the broker before it goes live, one change at a time. No words reach you that a person didn’t read, approve, and put his name to.
• It never makes up a market number. Prices, medians, and counts are arithmetic done by the database over real MLS records. A model is never in that path. Missing data renders as an honest blank, not an estimate.
• It never invents an answer. The AI concierge answers only from this site’s own published knowledge base — it cannot reach beyond it, and it cannot guess. When it doesn’t know, it says so and points you to the broker.
• It never profiles who should live somewhere. Every neighborhood page describes the place — its features, its flood story, its housing stock — never the kind of person who “belongs” there. That’s a fair-housing line the content is built to and checked against automatically.
• It never stores private party data. Seller and occupant private fields from the MLS feed are stripped at the source, before anything is saved — they never reach this site or any tool behind it.
• It runs under a cap and a log. The AI concierge operates under a hard spending limit with every query recorded — so it can be audited and can never quietly run away. (Site analytics already collect no personal information.)