Data Sources & Methodology
Innovator Lens aggregates public U.S. federal data into research-ready company profiles. Every field carries a badge showing where it came from and how much to trust it. We surface names and roles from public grant, patent, and publication records; we do not provide personal contact details, sell contact lists, or facilitate outbound marketing.
A hard external source — government data or SEC filings. Authoritative.
Computed by Innovator Lens (inference from other data) — not an external fact.
From the company itself (e.g. its website). Useful for recency, but may be marketing — treated as lower-confidence than filings.
Validated sources
Company records
ValidatedSource: SBIR.gov + SAM.gov ↗
Built from SBIR awardee records; government identifiers enriched from SAM.gov.
Investments (raises)
ValidatedSource: SEC EDGAR — Form D ↗
Parsed from SEC EDGAR Form D filings, matched to the company by issuer name.
Market signals
ValidatedSource: SEC EDGAR — 8-K ↗
From SEC 8-K material-event filings; event type and sentiment are NLP-classified.
Government contracts
ValidatedSource: USASpending.gov ↗
Direct from the USASpending.gov awards API, matched by UEI / DUNS / name.
Publications
ValidatedSource: PubMed + arXiv ↗
From PubMed and arXiv, matched by principal-investigator name and affiliation.
Patents
ValidatedSource: USPTO (PatentsView) ↗
From USPTO data via PatentsView, matched by patent assignee.
Derived sources
Technology sectors
DerivedSource: Derived from grants
Computed from this company's grant titles & abstracts by keyword scoring against a 15-sector taxonomy — not an externally assigned label.
Competitors
DerivedSource: Derived from sectors
Inferred from companies that share technology sectors — an automated signal, not a curated list.
Predictive score
DerivedSource: Derived model
A temporally-validated model over grant/signal/enrichment features — a probability estimate, not a fact.
Deep Research
DerivedSource: Local LLM (Qwen)
Synthesized from public web sources by a local LLM, then quote-verified against the source — a computed signal, not an external fact. Primary-source facts (gov/edu/SEC) are support-checked; secondary sources are adversarially refuted.
AI briefing
DerivedSource: Local LLM (Qwen)
A "10-minute briefing" synthesized by a local LLM from this company's own signal graph (grants, funding, patents, publications, contracts, trials, technologies, competitors, market signals and verified research facts). A computed summary grounded strictly in that data — not an external fact; funding and competitor lists are taken directly from the underlying records.
Opportunity score
DerivedSource: Derived — your workflow
A heuristic score ranking companies by recency-weighted intent signals (recent SBIR, funding, hiring, patents, market signals) against the criteria you define in your workflow — a prioritization estimate, not a fact. Personalizes over time from your feedback.
Self-reported sources
Website intel
Self-reportedSource: Company website
Summary, products, investors, funding and key people taken from the company's own website. Self-reported — useful for recency, but may be marketing; validated filings (e.g. SEC Form D) take precedence on conflict.
Spotted something that looks wrong? Email hello@innovatorlens.com and we'll review it against the source.