Databricks puts AI agent tools under strict controls

Send us Fan Mail Databricks is pushing governance down to the tool layer, creating enforceable controls on what AI agents can actually do inside production systems. This shift matters because most real incidents come from over-permissioned tools, not model behavior. The episode explains how this new control plane works and why it changes the risk profile for firms deploying autonomous agents. For professional service leaders, the message is clear. AI is entering governed production use, and the firms that prepare their data, identity, and access layers will move faster with fewer surprises. Those who ignore the tool layer will struggle with preventable incidents. We also cover OneStream's finance-focused agentic layer, identity gaps in machine accounts, and a supply-chain warning from a CISA credential leak. Learn more at https://crestvale.io Support the show