An AI value-investing engine's honest read on any stock — quality, a sector-aware fair value, the bear case, and what would make it veto — including, clearly, when the answer is no.
Not a chatbot guessing. A pipeline that runs the same late-Buffett discipline on every name — and shows you all of its work.
Fourteen independent AI models debate every name — a bull case, a bear case, second opinions. When they disagree, we say so instead of faking a consensus.
A discounted-cash-flow estimate routed by sector — banks, REITs, biotech and software are valued on their own terms — with bear, base and bull scenarios, not one fragile number.
Earnings-quality and filing-distress screens (Beneish M-score, accruals, late filings, restatements, comment letters) can block a buy outright — the prize here is avoiding the blow-ups.
Position size follows conviction, Kelly-style — small when the case is thin, larger only when even the bear case is cheap. No oversized bets, no concentration.
A small, experimental account, managed live by the same engine. Every position, signal and reason is public on the live desk.
Outcome tracking began June 25, 2026. Every decision is journaled with timestamps. No returns are claimed until the data matures. First filter-vs-random results expected late July 2026.