Architecting AgentOps Needs CHANGE

TL;DR

The paper addresses the challenge of operating Agentic AI systems whose behavior continuously evolves post-deployment, highlighting the inadequacy of traditional DevOps and MLOps in such non-deterministic contexts. It introduces CHANGE, a six-capability framework ($\mathbb{C}$ontextualize, $\mathbb{H}$armonize, $\mathbb{A}$nticipate, $\mathbb{N}$egotiate, $\mathbb{G}$enerate, $\mathbb{E}$volve) to operationalize AgentOps and manage co-evolution among agents, infrastructure, and human oversight. The main contributions are the formalization of the six capabilities, their integration into an AgentOps platform, and guidance for evaluating alignment and evolution through simulations and controlled deployments, illustrated via a running Alice-and-Bob scenario. The work provides a practical path toward disciplined governance of evolving autonomous systems, enabling safer, more reliable Agentic AI deployments by turning change into observable, governed dynamics rather than an uncontrollable source of risk.

Abstract

The emergence of Agentic AI systems has outpaced the architectural thinking required to operate them effectively. These agents differ fundamentally from traditional software: their behavior is not fixed at deployment but continuously shaped by experience, feedback, and context. Applying operational principles inherited from DevOps or MLOps, built for deterministic software and traditional ML systems, assumes that system behavior can be managed through versioning, monitoring, and rollback. This assumption breaks down for Agentic AI systems whose learning trajectories diverge over time. This introduces non-determinism making system reliability a challenge at runtime. We argue that architecting such systems requires a shift from managing control loops to enabling dynamic co-evolution among agents, infrastructure, and human oversight. To guide this shift, we introduce CHANGE, a conceptual framework comprising six capabilities for operationalizing Agentic AI systems: Contextualize, Harmonize, Anticipate, Negotiate, Generate, and Evolve. CHANGE provides a foundation for architecting an AgentOps platform to manage the lifecycle of evolving Agentic AI systems, illustrated through a customer-support system scenario. In doing so, CHANGE redefines software architecture for an era where adaptation to uncertainty and continuous evolution are inherent properties of the system.

Figures (1)

• Figure 1: The $\mathbb{CHANGE}$ framework illustrated through agent interaction scenarios