Evidence you can inspect. Limits you can see.
Universr enables companies to act on real world risk. To do that responsibly, we separate evidence, context, signal and explanation — and we never present one as another.
Only real source checks can become evidence
Evidence is reserved for verified source checks that actually returned a result. When nothing verified is available, we say so plainly rather than filling the gap.
AI explanation is not evidence
AI helps explain what a result means in plain language. It never creates observations, events or evidence on its own.
No fabricated results
We do not invent satellite results, observations, events or evidence packs. Every claim is traceable to a labelled source.
How each source is used
NASA FIRMS
Evidence
Active-fire detections. FIRMS can be verified fire evidence when the real API returns a source check — including an honest result when there are no detections.
Weather
Context only
Current weather helps orient a snapshot. It is never presented as risk evidence.
Mapbox
Location and context only
Used for geocoding and maps. It locates a place and provides context, never evidence of an event.
AI explanation
Explanation only
AI is used to explain results in plain language. An explanation is not evidence and is labelled as such.
Planet
Premium
Planet is a premium source and is not presented as event evidence from metadata alone.
Earth Engine, Sentinel, Landsat
Scaffolded
These remain scaffolded and are not shown as verified evidence unless real verified output exists.
Explore, Monitor and Continuity
Explore is a one-off public snapshot. Monitor is ongoing tracking of a single site. Continuity is portfolio-level risk management across many locations and dependencies. The same evidence standards apply to all three.