Cisco Adds Agent Builder to Cloud Control Studio for Custom AI Operations Workflows

Cisco announced Agent Builder for Cloud Control Studio, a new capability designed to let enterprise teams customize and extend Cisco Cloud Control with AI agents.

A builder for operational agents

In a Cisco Blogs post, Cisco said Agent Builder is part of Cloud Control Studio and is intended to help teams create AI agents that work with enterprise tools, reusable skills, and operational workflows. Cisco describes Cisco Cloud Control as a platform for managing AI-ready data centers, and Cloud Control Studio as an extensibility layer for adapting that platform to specific enterprise environments.

According to Cisco’s product page for Cloud Control Studio, users can build custom agents, connect tools, and encode runbooks as skills. The company positions this as a way to move common operational procedures into governed agent workflows rather than leaving them only as manual documentation or one-off scripts.

Tool connections, reusable skills, and MCP support

Cisco’s announcement says Agent Builder lets teams connect third-party tools and define reusable skills that agents can use across workflows. The Cisco Blogs post also says the product supports the Model Context Protocol, or MCP, which is increasingly used to connect AI systems with external tools and data sources.

Cisco’s product page describes these integrations as part of Cloud Control Studio’s role as an extensibility engine for Cisco Cloud Control. In Cisco’s framing, the builder is meant for teams that want agents to interact with existing enterprise systems while staying within a managed operational environment.

The emphasis on reusable skills is notable because it suggests Cisco is focusing on repeatable operational tasks rather than only conversational assistants. Cisco says teams can encode runbooks as skills, which could allow organizations to standardize procedures for infrastructure, operations, or support scenarios inside the Cloud Control environment.

Testing and lifecycle controls

Cisco says Agent Builder includes sandbox testing, giving teams a place to test agents before publishing them into broader use. The company’s blog post also describes a versioned lifecycle, which indicates that agents and skills can be managed through stages rather than released as unmanaged experiments.

The announcement also points to governance and observability features. Cisco says Cloud Control Studio includes governance controls, while Cisco’s product page says agents can be governed with AI Defense and monitored with Splunk AI Agent Monitoring. Those references connect Agent Builder to Cisco’s wider portfolio, including security and observability products.

Cisco’s materials do not provide independent performance benchmarks or customer deployment metrics for Agent Builder. The available sources describe the product capabilities and intended use cases, but they do not establish how widely the tool is being used or how it compares with agent-building platforms from other vendors.

Why it matters

Enterprise AI agent deployments often depend less on the model alone and more on how safely an agent can access systems, follow approved procedures, and be monitored after release. Cisco’s announcement reflects that shift: the company is presenting Agent Builder as an operations-focused environment with integrations, testing, governance, and lifecycle management.

For Cisco customers already evaluating Cisco Cloud Control, Agent Builder may provide a structured way to tailor agentic workflows to internal tools and runbooks. For the broader market, the announcement is another example of infrastructure and networking vendors adding AI agent development features directly into operations platforms rather than treating agents as standalone applications.

Cisco’s public materials frame Agent Builder as a way to extend Cisco Cloud Control, not as a general-purpose AI development platform. The clearest claims from Cisco are that teams can build custom agents, connect tools, create reusable skills, test in a sandbox, use MCP support, and manage publication through governed lifecycle controls.