Tag: agentic-ai
All the articles with the tag "agentic-ai".
When Agents Succeed and Systems Fail
An agent can close a case while the customer is still blocked. A practical architecture for agentic investigations with durable state, governed tools, trusted context, and independently verified outcomes.
Permission Is Not Approval: How to Authorize AI Agent Actions
An AI agent can have valid permissions and still make the wrong decision. An architecture for independent checks and action-bound authorization.