RedInk.

RedInk is an AI financial anomaly analyst for the NASDAQ QQQ universe that automatically detects when management's narrative diverges from the actual financials — and proves it with verifiable computations, SEC filing citations, and visible trust scores on every answer. Built explicitly not as an AI-for-finance chatbot but as one tightly-scoped analytical product for the analysts who cover dozens of filings per quarter and don't have time to read every MD&A claim against the numbers.

The positioning is precise: "skeptical by default, works like an analyst, not a chatbot". The product's mission is management spin detection — flagging when filings say "strong growth" while revenue grew 2%, or "improved margins" while liabilities jumped 18%. Every claim the AI makes is proved with verifiable computations (the math is shown), cited to source (the SEC filing line item is linked), and trust-scored (the LLM-as-judge eval result is visible alongside the answer).

Four user-facing capabilities ship in V0: (1) Anomaly Leaderboard — top anomalies for the latest quarter across the QQQ universe, ranked so the analyst knows where to start (you can't read 30 filings in a day; the leaderboard tells you which ones look weird this quarter); (2) Company-Quarter Detail — anomaly score, top drivers, time-series chart of key metrics; (3) Ask a Question + Evidence Panel — analyst-style conversational Q&A grounded in retrieved context; (4) Eval Panel — LLM-as-judge results visible in-line for trust calibration.

B2C + B2B: free tier attracts retail investors and finance enthusiasts; the real money is B2B — equity research teams, hedge fund analysts, audit firms, and compliance departments who need automated filing analysis at scale. Riding the 30%+ growth wave in AI-powered financial tooling against incumbents (Bloomberg Terminal — dominant but expensive) with a fundamentally different positioning: not a terminal, a sharply scoped management-spin detector with receipts.