NexArt launches Evidence Bridge for AI execution logs

NexArt has launched Evidence Bridge, a managed ingestion layer that turns existing JSON event data into cryptographically verifiable AI execution evidence without requiring schema changes. The product arrives as the EU AI Act’s logging and retention rules take effect, targeting organizations that need traceable records from systems already in production.

Why it matters: - Evidence Bridge could reduce the friction of AI compliance by letting organizations reuse logs they already generate. - The product targets teams that need traceable execution records without rebuilding event schemas that also feed analytics, observability and audit workflows. - The launch lands after Article 12 of the EU AI Act took full effect on 2 August 2026. - Article 12 requires high-risk AI systems to technically allow automatic recording of events over the system’s lifetime at a traceability level appropriate to the system’s purpose. - Article 26(6) requires deployers to retain those logs for at least six months.

What happened: - NexArt launched Evidence Bridge on 10 September 2026. - Evidence Bridge is a managed ingestion layer that converts existing JSON events into structured, cryptographically verifiable AI execution evidence. - The service is available now at bridge.nexart.io. - NexArt said the system works without requiring changes to existing schemas. - Evidence Bridge can create Certified Execution Records from event data already produced by production systems.

The details: - Most production AI systems already record execution identifiers, timestamps, requested models, inputs, outputs and governance decisions. - Those fields often live in different formats across systems, with one application using runId, another using execution.id, and provider information stored in metadata. - Governance decisions may also sit in separate systems. - Evidence Bridge lets organizations keep their existing JSON and define a mapping that explains how each field should be interpreted as execution evidence. - The mapping process starts with representative JSON. - Evidence Bridge suggests mappings between source paths and NexArt evidence semantics. - A human review step follows, because matching field names does not prove meaning. - NexArt said a field named model could mean a requested model, a routing alias, a fallback, or a provider-reported value. - The reviewed mapping becomes an immutable profile revision. - Applications then send the same JSON to a Bridge ingestion endpoint. - If the event includes the required semantics, Evidence Bridge creates a Certified Execution Record and verifies its integrity locally. - If the event lacks required semantics, Evidence Bridge marks it ineligible instead of inventing missing information. - Evidence Bridge preserves an Origin Trace for every mapped field. - The Origin Trace shows the value, the source path, any constant used, the transformation applied and the active profile revision. - Profiles are immutable and revisioned, so historical evidence stays tied to the rules that produced it.

Between the lines: - NexArt is positioning Evidence Bridge as an evidence-mapping layer, not a new logging standard. - The product’s core pitch is that cryptography can secure evidence integrity without forcing teams to redesign working production systems. - NexArt also draws a clear line between integrity and truth. - A Certified Execution Record shows that protected information stayed consistent with the artifact’s cryptographic identity. - A NexArt Node attestation adds signed evidence that the Node verified and accepted the artifact. - Neither proves that the originating application’s claims were true. - If an application logs the wrong model name and Evidence Bridge maps it faithfully, the record can remain cryptographically intact while still containing an incorrect claim. - NexArt said its certification is not an EU AI Act conformity assessment and does not certify ISO/IEC 42001 compliance. - The company also said Evidence Bridge is not a replacement for direct SDK integration. - Records can be created and verified locally without Node attestation. - When signed attestation is required, the record is submitted unchanged to a NexArt Node.

What's next: - Organizations can start using Evidence Bridge now through the bridge.nexart.io service. - NexArt said direct SDK integration remains available alongside the new ingestion path. - The company is likely to position Evidence Bridge for teams building audit-ready AI systems that need verifiable execution records without altering existing pipelines.

The bottom line: - NexArt is betting that the easiest path to AI evidence is to map existing JSON into certified records, not to rebuild production systems from scratch.