AI Reply Safety Gate, Multi-Stage Knowledge Fact Extraction, and Soft-Delete Catalog Protection
Added a pre-delivery AI reply safety gate and PII controls per agent, automated multi-stage knowledge fact extraction and verification, and soft-delete safeguards for products.
Key changes
- Configure content moderation and personal-data safety gates inside Agent Builder before AI messages reach customers.
- Extract and verify structured knowledge facts in multiple stages for more accurate product and policy answers.
- Protected product catalogs with soft deletion while streamlining training package imports in the UI.