Broadcom brought AI aplenty to the latest release of its Automic Automation IT control system.

The latest V26 iteration of the platform embeds agentic AI capabilities into orchestrated workflows, such as those related to VMware Cloud Foundation (VCF) integrations in the program. These agents are plugged into Automic Automation through the use of a newly-integrated model context protocol (MCP) server, where it could then determine how to allocate compute, storage, and networking resources based on real-time conditions. In addition, users are able to plug in and run their own chosen large language model (LLM).

Agent orchestration is made possible by a new AI Job Types feature, which brings AI-driven reasoning directly into workflows as governed, auditable objects. Broadcom said its “intelligent control plane" sees non-deterministic AI behavior operating within deterministic control frameworks, "preserving traceability, policy enforcement, and role-based access."

In effect, Automic acts as both MCP client and server, positioned as a central control plane for an increasingly disjointed ecosystem of AI agents and MCP-enabled services. Broadcom pointed to native Python deployment, agentless execution options, and AI-aided workflow creation as reducing reliance on fragmented orchestration and scripting solutions.

Automic Automation V26 comes after last year's updates to VMware Avi Load Balancer (Avi), with the multicloud application delivery platform previewing MCP controls. In the preview, traffic using MCP is protected by a web application firewall, while the system keeps sessions persistent and enforces authorization, narrowing the attack surface created by AI workloads.