Microsoft is making a major push with agentic AI to streamline and automate cloud operations by introducing an “agentic mode” in Azure Copilot. This mode can surface insights and offer recommendations, but it does not execute actions on its own.

Figure 1. Microsoft Bets on Agentic AI for Cloud, But Analysts Question Its Impact.

While Microsoft presents the upgraded Copilot as a tool that can simplify large-scale cloud management by autonomously taking actions—unlike traditional tools that rely on manual input and specialized expertise—analysts caution that it may add new complexities instead of eliminating them. Figure 1 shows Microsoft Bets on Agentic AI for Cloud, But Analysts Question Its Impact.

The agentic Azure Copilot, unveiled at Microsoft’s annual Ignite conference, employs a reasoning model to coordinate six specialized agents. This helps handle tasks such as migrating and modernizing legacy applications, deploying cloud infrastructure, optimizing performance, ensuring observability and resiliency, and troubleshooting issues.

What sets the agentic Copilot apart from its predecessor is its ability to orchestrate these agents independently based on user intent, whereas the previous version relied on a general-purpose large language model (LLM) that mainly recommended tools and actions.

The agentic Azure Copilot can be activated via a simple toggle in the Azure portal and, once enabled, is available across all Azure workflows, including the command line interface (CLI). Enterprises can also choose to save chats within the upgraded Copilot to a location of their choosing for governance and audit purposes.

Analysts are skeptical of Microsoft’s claims that traditional cloud management tools complicate operations and that an agentic Copilot is the necessary fix.

Linthicum highlighted the trade-off enterprises might face: the promised ease of cloud operations versus the increased complexity of managing the agentic Copilot itself.

“There’s a fundamental tension here,” he said. “Copilot’s agent mode can simplify many day-to-day operational tasks for users and operators, but it also introduces new administrative considerations — especially around policy-setting, access control, and compliance.”

“Setting up granular spend permissions, managing access to sensitive datasets, and enforcing retention or storage controls all demand a more deliberate approach from administrators. I’d advise clients considering this to weigh the additional complexity and effort against the potential value,” Linthicum added.

Derek Ashmore, agentic AI enablement principal at Asperitas, emphasized that agentic cloud operations don’t remove the need for governance—they actually heighten it, adding complexity to administrative roles.

“…most enterprises won’t enable this overnight. Realistic adoption takes 3 to 9 months, depending on cloud maturity. Early pilots will focus on non-production environments with narrow use cases before moving to workflows that can be automatically remediated,” Ashmore said.

Real benefit or just incremental improvements?

Despite concerns about administrative overhead, Ashmore noted that enterprises could see meaningful benefits once the upgraded Copilot is fully deployed.

“When policy layers and data-access patterns are properly defined, agents can significantly reduce day-to-day operational toil. Tasks like drift management, cost optimization, incident triage, and cross-service roubleshooting become much lighter,” he said, comparing Azure Copilot’s trajectory to that of Infrastructure-as-Code: while setup requires discipline, the long-term payoff is improved consistency and operational velocity.

Linthicum, however, remains unconvinced. He argues enterprises are unlikely to see substantive gains beyond “a few tactical improvements” and believes Microsoft is essentially creating a problem to solve with its new software.

Analysts also note that Microsoft isn’t unique in framing cloud operations through an “agent” lens. AWS and Google are pursuing similar strategies: AWS offers Amazon Q for ops teams to design, troubleshoot, and provision resources via chat and CLI, while Google Cloud provides Gemini Cloud Assist for lifecycle management.

Source: NETWORK WORLD

Cite this article:

Priyadharshini S (2025), Microsoft Pushes Agentic AI for Cloud Management, Analysts Remain Skeptical, AnaTechMaz, pp.180