The industry OS · private alpha
Real estate
runs on ANVIL.
One data spine. Every role in the deal — flipper, wholesaler, contractor, lender, realtor — working the same address, the same numbers, the same truth. No copy-paste. No leaks.
Chapter one · the problem
A flip passes through six people before the keys change hands. Wholesaler. Investor. GC. Lender. Realtor. Buyer. Each one works in a different tool. The deal leaks at every handoff. ANVIL stops the leak.
One address
A property enters ANVIL once.
A wholesaler distributes a deal. It lands in the investor's inbox. Same row. Same ID. Same address hash. Nobody types anything twice.
Every role sees their view
One dataset. Five surfaces.
The flipper sees the underwriting. The GC sees the SOW. The lender sees the draw request. The realtor sees the listing spec. They're not separate systems synced through email — they're the same system, styled per role.
Truth compounds
Every close teaches the next deal.
Sold price hits the comp engine. Actual rehab hits the estimator's calibration. Wholesaler accuracy hits their reputation score. The database gets smarter with every property that closes.
- Marketplace effect · every role brings the next
- Multi-role data spine · address entered once
- Outcome database · every close teaches the next
- Reputation as infrastructure · accuracy tracked, forever
- Legal data foundation · no vendor risk, compliant by design
Choose your surface
Two live today. More coming.
Flippers.
Underwrite, rehab, and close houses. Full pipeline from wholesaler text to closing wire — on one screen, in your rhythm.
Wholesalers.
Distribute deals to verified investors. Build reputation on accuracy — the closest thing to a moat wholesalers have.
Building in public
Every week. Publicly.
Monthly investor overview and LP partner statements, then ten improvements behind them — eval harness, network reputation, deal→offer→SOW, Fair-Housing audit trail. (#376, #377, #378)
Every dollar figure moved from decimal to integer cents across four phases — shadow columns, dual-write, writer sweep, then the reader migration behind a CI drift gate. (#369–#372)
Phases 2 through 5: real glass chrome, kinetic card and button primitives, the /today mesh canvas, the Deal Radar polar plot. Every surface breathes now — subtle, purposeful, never noisy. (#347–#358)
AI Deal Thesis, URAR waterfall, Match Score radar — then save-as-preset and compare-two mode on top. Comparable-property analysis reaching the ceiling of what a CRM can do. (#327, #328, #338)
Wholesalers can distribute deals to verified investors, with verification badges, saved-search alerts, and reputation baked into every send. (#280–#290)
→ See the full changelog
“I’ve flipped 40+ houses. Every tool I tried did five percent of what I needed, or forced my workflow into the shape of theirs. So I built the tool where nobody has to copy anything, and the way we actually run a flip.”