Attention is scarce
Institutional analysts receive more information than they can review and must reconstruct history manually.
Evidence-linked disclosure intelligence
Find What Changed. Verify the Evidence. Decide What to Investigate.
Pure News Intelligence helps analysts compare corporate disclosures, trace every material change to its source, and turn selected findings into testable research questions.
- Evidence-linked disclosure changes
- Full prior-context review
- Transparent research limitations
- No automatic trade recommendations
The research journey
A negative result produced a more defensible workflow.
The project did not validate a new trading signal. It did reveal a practical need for evidence-linked review.
The alpha thesis failed
Locked tests did not show incremental value from 10-Ks or enriched expected-language models.
Discovery sharpened the problem
Six interviews emphasized prioritization, trust, provenance, integration, and human judgment. These interviews informed the design; they did not validate adoption.
The project pivoted
The product and quantitative research were separated around a narrow SEC filing workflow.
The MVP works
It compares consecutive 10-Ks, checks earlier disclosures, and exposes exact supporting evidence.
The next test is a pilot
A controlled analyst workflow study should measure time, recall, trust, corrections, repeat use, and memory value.
Experimental capability
Turn a Verified Change into a Testable Research Question
For selected disclosures, the Experimental Research Idea Engine proposes economic mechanisms, analyst questions, affected entities, and falsifiable hypotheses. A skeptical-review stage searches for missing prior evidence and unsupported reasoning before anything is shown.
Grounded hypothesis
Attach atomic claims to full-prior evidence and require a falsifiable research question.
Skeptical challenge
Search for omitted context, incompatible quantities, and unsupported causal mechanisms.
Analyst decision
Accept, edit, reject, or escalate. The system never substitutes for analyst judgment.
8cases reviewed
2publishable hypotheses
6safe failures
0actionable trade views
This feature is experimental. It generates research hypotheses, not validated trading signals or investment recommendations.
Coverage readiness
View Universe Coverage Preview
Explore which companies have filing and permitted news metadata indexed for future staged analysis.
Frozen demonstration
The claim boundary is part of the product.
The interface preserves point-in-time evidence and keeps system interpretation, analyst judgment, and later outcomes visibly separate.
150companies and consecutive 10-K pairs
995evidence-backed disclosure changes
629high-confidence changes
1,990verified excerpt-offset checks
Final research classification: INSUFFICIENT_EVIDENCE. The product demonstration is operational; the risk analysis was inconclusive. No alpha, causal, trading, customer-adoption, or institutional-validation claim is supported.
Product focus
One user, one accountable decision.
Human judgment remains the final decision point.
Target user
Institutional research analyst
Needs to focus attention, compare history, verify evidence, and explain conclusions.
Buyer and gatekeeper
Head of Research + CTO
Need measurable workflow value, integration, security, governance, and auditability.
Value proposition
Read the right thing first
Understand what changed and preserve the supporting record without weakening analyst judgment.
Disciplined innovation
Evidence changed the project rather than being tuned away.
The research did not prove alpha, but it helped reveal a better innovation opportunity.
Original ideaPredict returns
→Empirical testLock the comparison
→Null resultReject the claim
→Customer discoveryRefine the problem
→Working productEvidence-linked review
→Analyst pilotTest workflow value
Start with the evidence
Follow the recommended demo path.
Review ranked changes, inspect contrasting cases, then audit the research results and limitations.