RivalLens

Methodology

How RivalLens produces grounded intelligence

RivalLens treats a monitored page, uploaded document, and AI output as untrusted inputs. A useful conclusion must remain traceable to a source, capture time, workspace, and analysis version.

Capture and scope

Users confirm the page region and monitoring intent before a new source is created. Checks preserve the last valid baseline when a locator is uncertain; they do not silently expand to the full page.

Evidence and claims

Evidence records retain source URL, capture time, content hash, and access scope. Facts cite evidence. Inferences are labeled as inferences. When evidence is insufficient, RivalLens should abstain rather than fill gaps with a plausible claim.

Signals and quiet periods

A signal is delivered only after confidence, evidence, scope, preference, and deduplication checks. Quiet periods are reported as no verified move, with coverage health and research recommendations—not as evidence that nothing changed anywhere.

Feedback and revisions

Feedback can improve future matching rules. It does not rewrite historic evidence or change the meaning of an already persisted analysis. Corrections create a later version.