Evidence is not a file store. It is a dependency of every decision.
AURELIS treats evidence as structured, version-aware records that connect claims to sources, AI assets, policies and decisions. When a model, permission or dependency changes, the question is not only whether evidence exists—but whether it still supports the claim being made.
Follow one decision back to its proof.
Select an item to inspect its metadata. This is a clearly labeled synthetic example demonstrating the intended data model, not live customer evidence.
What a useful record needs to say.
These fields support traceability and review. Not every evidence source can supply every field; missing values should remain explicit rather than be invented.
Provenance
Record ID, source system, collection method, collector identity where available, timestamp and integrity metadata where implemented.
Applicability
Asset ID, exact version, environment, relevant dependency, claim or metric, and the conditions under which the evidence applies.
Freshness & verification
Verification state, expiry or review interval, conflicts, superseded records and the reason evidence may no longer support a claim.
Policy lineage
Requirement, policy condition, assurance case and decision that consumed the evidence.
Impact links
Related model versions, vendors, tools, permissions, assets and assessments that may require reassessment.
Resolution
Finding, owner, remediation, verification evidence, closure state and resulting reassessment.
Preserve the chain from observation to action.
Accept a source record or customer-provided observation.
Map it to a stable schema without silently changing meaning.
Check required fields, scope, provenance and available integrity data.
Link the record to assets, versions, claims and policies.
Evaluate whether evidence remains applicable after material change.
AURELIS does not turn an uploaded document into proof of compliance by itself. Conclusions depend on source quality, validation rules, customer context and configured integrations.