Tag: eval suite
-
Agentic AI Change Management: A Prompt Edit Is a Production Change
Agentic AI change management treats every prompt, model and tool edit as a production release. Here is what counts as a change and how I gate it.
-
Agentic AI Code Review: Intent, Not Diff
What is agentic AI code review? Agentic AI code review is how a team checks code that an agent wrote before it ships. The human stops reading every line and reviews the intent, the evals and the scope of the change, while review agents scan the diff first. One thing stays fixed: a named person…
-
Agentic AI Non-Determinism: Repeat the Boundary, Not the Route
What is agentic AI non-determinism? Agentic AI non-determinism is the fact that the same agent, given the same task twice, takes a different route each time. The prompt is the same, and so is the model. But the steps, the tool calls, and sometimes the answer are not. So a test that expects the same…
-
Agentic AI Governance Controls: 10 Questions, 10 Answers
What are agentic AI governance controls? Agentic AI governance controls are the checks built around an AI agent that decide what it may do. Some only guide the agent. Others can stop it. You cannot govern an agent with a policy document. You can govern it with controls around it. New here? I publish one…