DocuLens methodology: put the evidence beside the conclusion
DocuLens is an AI-assisted research workspace over real earnings-call transcripts, SEC filings, and SEC/XBRL facts. Its central design choice is simple: a useful answer should leave a path back to the record that supports it.
What goes in
The workflow works from public-company source material. SEC EDGAR provides the filing record for 10-Ks and 10-Qs, including the filing text and tables used by the filing workflows. SEC/XBRL facts provide structured historical values where the relevant concept and period are available. Earnings-call transcript coverage depends on the source: DocuLens can fetch transcript exhibits filed with the SEC and supports uploaded or pasted transcripts when a company does not file one there.
How a research request is scoped
A request is anchored to a ticker or company, a topic or question, and—when appropriate—a year or quarter range. The scope matters because “revenue” in one fiscal period is not automatically comparable with “revenue” in another, and a statement about a future quarter should not be graded against the wrong result.
How evidence is retrieved
DocuLens searches the available transcript and filing material for passages relevant to the requested topic. Retrieval may use literal terms and semantic similarity to find related language, while workflow-specific checks help keep context blocks, tables, notes, and transcript exchanges attached to the right source. The exact retrieval path varies by workflow, so a Core Research report and a narrow question extraction are not the same product behavior.
How conclusions are organized
Research outputs separate evidence from synthesis. A report may combine a quoted passage, a filed number, a structured SEC fact, and an explanation of how those pieces fit. Guidance vs. Actual uses the time axis between an original forward statement and a later result. Promise Tracking is for qualitative commitments. What Changed compares recent transcript discussions. The filing explorer preserves selected source tables in the Excel workflow with a Source Map.
What the system does not claim
- It does not guarantee that every company's transcript is publicly available in the SEC record.
- It does not turn a directional comment into a precise forecast.
- It does not treat a missing mention as proof that a project was abandoned.
- It does not invent historical numbers when an SEC/XBRL fact is unavailable or ambiguous.
- It does not provide personalized investment advice, a price target, or a buy/sell recommendation.
How to use the output responsibly
Inspect the cited source, check the reporting period and definition, and challenge the synthesis when the evidence is thin or conflicting. If the answer matters, open the linked filing or transcript and verify the passage yourself. The best use of an AI research tool is to shorten the path to primary material while keeping human judgment in the loop.
Start with the research guides, then use the transcript analyzer or SEC filing analyzer for the question in front of you.