What's new

All 10 categories shipped as importable campaign templates with Indic variants.

Bundle Trust Lab certification evidence with Operations Platform runtime traces in a single auditor-ready PDF.

Auto-fallback to Claude Sonnet when Sarvam-1 throughput degrades; configurable per evaluator.

EU AI Act bundle generation no longer queues behind unrelated jobs.

Run agent policies against historical traces; see what would have happened before deploying.

TrustLayer-maintained Inspect fork with sandboxed Python tools and platform-data access.

All audit events now cryptographically chained; verifiable via CLI.

Full pillar/control tree mapped; readiness scoring, gap analysis, filing-ready evidence.

Submit multiple artifacts under shared evidence collection.

A/R/D/N keys now work even after dragging the split pane.

8,400 new test cases across Hindi, Bengali, Tamil, Telugu, Marathi, with code-switched variants.

Trust-lab-conducted vendor evaluations for Anthropic, OpenAI, Google, Sarvam, Cohere.