Chapter 4. The Agentic Reliability Architecture
Most AIOps initiatives that fail do so because they added intelligence to existing tools rather than designing the system that intelligence can safely operate within. Model quality, adoption, incentives, ownership, and data access all matter too, but architecture is the factor a decade of pattern observation keeps landing on as the one teams underinvest in and then can’t retrofit. The teams that wire smart components onto existing toolchains find themselves explaining, 18 months in, why the agent’s behavior is inconsistent, why trust is eroding rather than growing, and why the same class of incident keeps producing different responses. The teams that design the system first do not have those conversations. They have different conversations, harder ones, but the harder conversations are the ones that lead somewhere.
Autonomy is not a feature you enable. It is a property that emerges when a system is structured to observe itself clearly, reason within defined constraints, and act without exceeding delegated ...
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