Powered by an interwoven integration with Knowledge Enrichment for content that is readied for AI consumption, this Knowledge Discovery release delivers improved response quality, smoother day-to-day use, and tighter administrative oversight. Whether you’re an end user hunting for insights or an admin safeguarding compliance, you’ll notice the impact the moment an Agent opens. Available TODAY with Knowledge Discovery 1.1 release.
At a glance
- Higher-precision answers driven by Knowledge Enrichment’s data-curation pipeline plus broader file-format support and LiteLLM guardrails
- Richer context up front thanks to an overhauled Focus panel and granular metadata filters
- Real-time quality indicators for administrators with clear warnings before swapping language models
- Point-and-click model switching to Claude 3, Llama 4 (Maverick or Scout), or AWS Nova
- Governance baked in: prompt-injection defenses, policy-compliance checks, and targeted authorization fixes
Cleaner, enhanced Agent scoping
- Streamlined Focus panel surfaces the most relevant context as soon as an Agent launches.
- Expanded metadata filters let users zoom into exactly the content slice they need, faster.
Advanced Agent management & administration
- Live quality indicators let administrators track answer performance in real time.
- Transparent pre-change warnings appear when you swap language-model providers.
- Streamlined agent settings make it easier for admins to adjust scopes and permissions.
Platform & architecture upgrades
- 600 + file formats supported: XLSX, TIFF, PPT, audio transcripts, and more.
- LiteLLM guardrails defend against prompt injection, policy violations, and hallucinations.
- One-click model switcher empowers admins to move between Claude 3, Llama 4, or AWS Nova without code.
