Overview
- Snowflake announced at its Summit on Tuesday a suite of features that position the company as an agentic AI platform focused on governed context, agent security, and native streaming.
- Horizon Context, Semantic Studio and Semantic View Autopilot form a shared semantics layer so AI agents and BI tools operate from the same trusted business definitions.
- Agent Identity plus new AI security posture and trust-center tooling give agents verified identities, role-based permissions and complete activity audit trails to support zero-trust access.
- Datastream, native Kafka compatibility, Snowpipe streaming updates and zero-copy integrations with SAP and Salesforce bring fresher, source-semantic data into Snowflake to cut data movement and speed agent workflows.
- Developer and runtime tooling—CoCo, CoWork, Snowpark and Adaptive Compute—aim to move agents from pilots to production with mixed availability across general release and public or private previews; the rollout follows a recent $6 billion AWS collaboration to help customers deploy AI.