Disclosures
How we use AI
Some research is drafted by a model from cited sources. Most of it is approved by a person before it is published. Names in our deepen queue are the exception: their scorecards publish automatically from cited SEC filings, with an AI-generated label and no human pre-approval. The AI mark on a section links to this page.
What the model drafts
AI drafts research notes, the tech explainer, roadmap items, risks, the promise tracker, the summary of public posts, and scorecard rationale. The draft is written from cited SEC filings and transcripts. It is not a live market feed and it is not a price target.
Claims are checked against those sources. A sentence that is not in the cited filing or transcript does not get published as a sourced claim.
Reviewed and not yet reviewed
Not every AI-assisted section has been approved by a person.
Deepen queue. Names in the deepen queue move from basic to full scorecards automatically. The model drafts moat, stack importance, customer concentration, cycle and the thesis from the company's latest SEC annual report. Each value must quote that filing word for word, and automated checks block price targets, buy or sell language, new red flags and removed kill criteria. If those checks pass, the scorecard publishes with an AI-generated label and no human pre-approval.
Everything else. Research drafts for names outside the deepen queue are approved by a person before publishing and carry the reviewed label. Automatic publishing is limited to the deepen queue. AI transcripts of official earnings-call replays are machine transcriptions, marked as such, with a link to the replay.
Not investment advice
Nothing on this site is investment advice. We set no price targets. Scores are rule-based checklists on our coverage, not a recommendation to buy or sell. Do your own research.