No brand sets out to accumulate a dozen AI agents. It happens because the first one worked.

Order-status inquiries get automated, the numbers look good, and then billing wants an agent. Marketing wants an agent. The CCaaS platform ships one. Meanwhile, two business units quietly build their own. Each decision makes business sense on its own, but together they create agent sprawl.

In the new Opus Research paper, “How Brands Can Govern the Next Wave of AI Agents,” VP and Lead Analyst Ian Jacobs argues that this is where the conversation must shift from deployment to governance. Once brands start calling agents “teammates” and “workers,” they have effectively conceded the point: a workforce needs training, policies, escalation paths, and performance measures.

A single agent is a project. Thirty agents are a distributed workforce operating in your name.

The paper builds on the five-layer AI Agent Control Plane: journey and intent state; identity and consent; policy and guardrails; knowledge governance; and evaluation and testing. It makes the case that the fifth layer is essential to long-term AI-agent success.

That layer includes four forms of agentic testing: pre-deployment testing, continuous monitoring, audit and observability, and cross-agent load testing. It also identifies the practices that distinguish teams managing agents effectively from those waiting for customers to expose edge cases.

The report is available now. Our thanks to Cyara for underwriting this research.

Use the form below to receive an email and download this free Opus Research whitepaper, “How Brands Can Govern the Next Wave of AI Agents”.

Categories: White Papers, Conversational Intelligence, Intelligent Assistants, Intelligent Authentication, Articles