Chapter 6. From Change to Confidence: Agentic Delivery Pipelines
If incident response is where autonomy is tested under pressure, delivery is where it is tested under intent. Most reliability failures do not begin during incidents. They begin during change: a configuration tweak that was right yesterday and wrong this morning, a rollout that moved faster than understanding, or a deployment that was technically correct but contextually wrong because the system it landed on had drifted since the change was authored. The change itself is rarely the whole bug. Most often, a meaningful part of it is that the system the change met was different from the system the change was tested against, and nothing in the pipeline noticed.
This chapter is the second proving ground because delivery is, in one important sense, the easier of the two workflows. SRE teams already accept heavy automation in their pipelines. Most modern engineering organizations run CI/CD with deterministic build steps, gated promotions, and automated rollback hooks. The substrate is already in ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access