Core intelligence artifact
Findings that keep the evidence attached.
A WITNESS Finding is not just an AI answer. It is an evidence-backed conclusion whose supporting observations, alternatives, uncertainty, limitations, and reasoning path remain inspectable.
The distinction
Explanation is cheap. A justified finding is harder.
General-purpose AI can produce a coherent explanation from incomplete context. WITNESS is designed to investigate first: reconcile the evidence, compare signals, form an inference, challenge it against alternatives, and preserve the conditions that could change the conclusion.
What a Finding preserves
Observation
What was actually detected or measured before interpretation.
Evidence
The sources and signals the conclusion depends on.
Competing explanations
Other plausible causes considered before settling on the leading explanation.
Uncertainty
What remains unknown, disputed, incomplete, or weakly supported.
Finding
The conclusion that best fits the available evidence at the time of investigation.
Change conditions
The new evidence or conditions that would materially alter the conclusion.
Traceability
The reasoning path remains visible.
WITNESS separates source reports, WITNESS analysis, model-generated inference, and canonical evaluations so an answer cannot silently become its own evidence. Findings are designed to be inspected, challenged, and revised when new evidence arrives.