Six production AI products built for analysts who need to revise theses, not generate them. Deterministic-first reasoning. Provenance on every metric. Mechanism-only signals labeled as such — never dressed up as conviction.
Each product is a working system you can see today, not a slide. Direct portal links where the product is shrink-wrapped; request access where the analytical depth needs a 30-minute walk-through first.
LiveEquity researchBuild the thesis on any US stock. Reads filings, valuations, and macro context — and tells you when the data revises the call.
Why this is different: Deterministic-first reasoning. Every number traces to its source filing or quote API. The model never invents prices.
LiveForensic microcapsForensic microcap research. 200+ tickers screened weekly for dilution, runway, and gatekeeper red flags — with cited deep-dive on any name.
Why this is different: 16-node forensic framework. Verdicts grade evidence honestly — validated drivers vs. mechanism-only commentary vs. tested-null context.
LiveBiopharma researchPoint-in-time biopharma research — people-and-effort signals, board-backing scoring, valuation vs. base-rate LoA, interactive board-network graph.
Why this is different: officer_share — the 0.77-AUC tell — surfaced as a first-class signal. Every metric carries its source date. Tier-labeled evidence on every node.
LiveReverse-DCFReverse-DCF valuations with XBRL-grounded inputs and provenance on every number. Edit any assumption; see the implied growth back out.
Why this is different: Sector-equivalent inversions: financials get implied ROE growth, REITs get implied AFFO growth, banks get implied dividend growth. Excel export.
LiveMulti-factor screensGA-evolved screening strategies for US equities — smart-money flow, earnings catalysts, multi-factor combos with walk-forward validation.
Why this is different: Out-of-sample Sharpe ratios reported per strategy. Strategies that overfit on training data are visibly flagged, not hidden.
LiveStrategy briefsAsk a decision question, get an analyst-grade strategy brief in ~5 minutes — multi-step reasoning over the context you provide, every claim cited.
Why this is different: Live multi-step reasoning pipeline. Cost transparent per run (~$0.10-0.25). Reports rendered as cited markdown, not vibes.
LiveForensic verdict engineTicker in, verdict out. Scores ~150 deterministic forensic rules plus ~44 LLM judgments against a company's filings; produces a composite score, conviction, and cited decision memo.
Why this is different: 209-rule library. Every fired rule is cited with evidence. Composite score is a scalar from −5 (bearish) to +5 (bullish), with per-family severity tiers.
These aren't marketing claims. They're rules our agents check on every build — when one is violated, the commit is blocked.
Valuations and ratios are computed by code, not generated by language models. The LLM provides the narrative; the math is verifiable.
Every number on a dossier carries its source date. If a price is from a stale cache, you see the date inline. Missing dates render a visible 'date unknown' badge, not a blank.
Each finding is graded: validated drivers vs. mechanism-only commentary vs. tested-null context. We tell you how much each signal is backed — not just that an AI said it.
When the data doesn't support a claim, our products abstain. You see a dash, not a plausible-looking guess. The model is allowed to say 'I don't know.'
The 6 above are our public products. We also build bespoke AI systems for clients — investment intelligence, structured credit, news graphs, simulation. Book a 30-minute call and we'll map your data to a working prototype.