The Model Context Protocol (MCP) began as an Anthropic initiative designed to bring order to the complex challenge of connecting large language models (LLMs) with external systems. It has since evolved rapidly into an important open standard for AI connectivity.
This is according to CASA Software – a leading SA digital transformation organisation that partners with global technology leaders, including Broadcom, to enable its customers to realise the value of AI-driven operations and streamlined automation. CASA explains how organisations using Broadcom Automic Automation, together with MCP, can move beyond passive chatbots toward governed, high-value AI agents.
Automic Automation is widely recognised as a market leader and is trusted to orchestrate business-critical processes for large enterprises. It enables companies to automate, manage, and scale complex IT workflows across hybrid, multi-cloud, and mainframe environments. Through centralised governance, role-based access control (RBAC), auditability, and controlled workflow execution, Automic helps businesses to operationalise AI safely within enterprise automation.
With Automic Automation V26, MCP is integrated into the platform through the Automation Assistant, AI Jobs, AI Connection objects, and MCP-enabled workflow patterns. This shift toward an open, vendor-neutral standard architecture has been recognised in the industry. In the 2025 EMA Radar for Workload Automation and Orchestration, Broadcom automation was named a Value Leader, while Automic earned Recognition for Excellence in Agentic Automation Enablement. “This recognition highlights Automic’s support for governed, reasoning-based AI and emerging standards such as MCP, positioning the platform not just as a tool for today, but as an execution layer for tomorrow’s AI-driven enterprise,” says Greg Harrowsmith, CASA Software Pre-Sales.
“Major industry players have since backed the protocol and in December 2025 Anthropic donated MCP to the Agentic AI Foundation, a directed fund under the Linux Foundation. This strengthened its position as an open, vendor-neutral standard. This allows enterprises to expose approved Automic workflows as governed automation capabilities, or skills, that authorised agents can call while retaining RBAC, audit, and security controls,” Harrowsmith confirms.
He notes that the MCP ecosystem has expanded rapidly, with published MCP servers connecting AI models to tools, data, and applications in real-world enterprise contexts. “This matters because MCP bridges the gap between an AI model’s static training data and the dynamic reality of the enterprise,” he says.
Beyond data access, MCP can enable tool-based actions where those actions are explicitly exposed and authorised. “By giving AI agents the ability to execute approved tasks, MCP helps transform passive chatbots into capable agents that deliver genuine operational value,” he adds.
Harrowsmith says the solution features an integrated intelligent assistant that goes beyond a simple help bot; it empowers users to interact with technical documentation, query real-time system data, analyse execution output, filter and sort lists using natural language and perform selected task-level actions through MCP-enabled calls. “Intelligent list filtering is one of its standout capabilities. Instead of manually constructing complex filter logic, users can simply describe the data they need. This transforms a traditionally tedious administrative task into a seamless, conversational command.”
Centralising the AI ecosystem
Harrowsmith says the shift extends far beyond the intelligent assistant. “Thanks to AI Jobs, ASK_AI, AI Connection objects, Automation.AI, and MCP tool integration, Automic enables teams to embed AI-assisted decisioning into governed workflows. Think of it as governance by design: LLMs can support natural-language inputs, workflow parameters, and execution branching within controlled automation processes. This capability reaches its full potential when paired with MCP, with Automic acting as a governed orchestration layer for the ecosystem,’ he says.
An illustrative enterprise workflow could involve an AI-powered Automic job using approved MCP tools to read a work item from an enterprise service or work-management platform. It can retrieve relevant context from an approved internal knowledge source, summarise the issue, and pass structured output to downstream Automic workflow steps for classification, routing, or escalation. “This centralisation helps move AI from an isolated black-box experiment to a governed automation pattern, where approved tools, data access, execution scope, and audit requirements can be managed explicitly,’ says Harrowsmith.
Automic as a skill provider.
In a truly agent-based ecosystem, value flows in two directions: consuming intelligence from approved models and executing governed actions. Automic is designed to support both.
Broadcom positions Automic as both an MCP client and MCP server, allowing approved external agents and LLM-enabled systems to interact with Automic workflows through governed, standardised interfaces. This allows authorised agents to call approved Automic workflows or tools that administrators have intentionally exposed.
Harrowsmith highlights CASAs wealth of technical IP and experience spanning over three decades: “SA businesses wanting to harness the power of AI should be speaking with us. As organisations transition from simple chatbots to complex, autonomous multi-agent workforces, this recognition confirms a critical truth and that is that if you want AI that works safely and effectively, you need Automic,” concludes Harrowsmith.