Updates to legacy IT automation tools over the past two months have begun connecting agent AI to systems that are central to the most critical enterprise workflows, from ERP to mainframes.
Businesses have been using workload automation and orchestration tools for decades, dating back to the days of enterprise IT before cloud computing and DevOps. However, the tools have since been expanded to address these new trends and have recently been adapted to generative and agent AI as well. Tools in this category often automate repetitive tasks and reliably connect disparate and complex software platforms into broader business workflows, said Dan Twing, an analyst at Enterprise Management Associates (EMA).
“Workload automation is the glue that moves from one application domain to another, preserving processes across different environments,” said Twing. “Every major organization, any large enterprise, has some level of capability.…For the last 20 years, the only way the cloud has grown is by having that support and automation behind it.”
Version 26 of Broadcom’s Automic software, released April 8, includes a new Agentic AI job type that enables workload automation tools to act as Model Context Protocol (MCP) servers and connect traditional IT orchestration to AI agents. Automic and other tools from Broadcom’s Agile Operations division can connect critical application data from ERP, mainframes, and core banking systems to Broadcom’s private AI infrastructure portfolio based on VMware.
BMC’s Automic competitor Control-M added an AI assistant and workflow creator on March 18 and is working with early adopter customers on AI agent-driven workload automation. On April 8, BMC released a statement on the direction of AI agent support in its Automated Mainframe Intelligence (AMI) product, extending zAdviser Enterprise mainframe reports generated by AI to include distributed systems applications.
On a deeper level of legacy infrastructure, IBM on April 2 announced an agreement with semiconductor maker Arm to bring cloud and mobile applications running on low-power processors to IBM Z and LinuxOne environments through virtualization.
Workload automation enables backward compatibility for enterprises.
Dan Twing EMA Analyst
What all of these updates have in common is that they position products that enterprises have been using for decades as a trusted way to add deterministic orchestration and governance to improve the reliability and security of non-deterministic AI workflows. As such, these vendors join a cacophony of other IT marketers making similar pitches to businesses. But Twing says these vendors already have access to very important customer systems, which could be attractive to larger companies.
“Workload automation brings the old world and the new world together and allows them to coexist. And we don’t have one old world and one new world. We have 15 different age groups,” Twing said. “Client/server capabilities are still there, early cloud architectures are still there. Workload automation is what enables backward compatibility for enterprises.”
Broadcom marshals workload acquisition
Broadcom’s Automic is built around jobs, which are software components that execute commands across a variety of environments, including operating systems, databases, enterprise applications such as SAP and Oracle’s E-Business Suite, file transfers, and web services. Version 26 adds an AI agent job type to coordinate and integrate AI agents into existing workflows. A new job type wraps Automic’s role-based access control, logging, and auditing protocols around AI agent activity for governance. Version 26 adds more natural language-driven workload execution interfaces to the Python-based code assist tool for building data pipelines.
rajeev kumar
“The user enters a simple prompt, and the LLM connected to Automic generates a workflow plan with the configuration and ground rules that are part of the product,” said Rajeev Kumar, head of workload automation products at Broadcom.
For example, a business analyst might want to pull data from Salesforce every day at 6 a.m., move that data to BigQuery for analysis, then move it to Looker to create a report, generate an AI summary of that report, and email it to the CEO. Kumar said Automic can now generate a workflow plan for that, identify the jobs and workflow elements that need to be created to achieve that outcome, present it to the user for review, and then deploy with user approval.
“AI is helping software engineers,” Kumar said. “We’re focusing on non-software engineers, business analysts, who have traditionally built these workflows but relied on ad-hoc tools.”
Broadcom has amassed an extensive hardware and software business over the past decade, making it one of the most important AI infrastructure vendors for enterprises to consider, said IDC analyst Stephen Elliott.
“People don’t realize how much of the Internet traffic goes through Broadcom hardware,” Elliott said. “VMware is just one part of that larger infrastructure software group. Don’t forget the CA part, the Symantec part, the networking software part, the chip software.”
BMC takes a prudent and strategic perspective
In its March update, BMC’s Control-M added support for AI agents from partners such as CrewAI, LangGraph, and Snowflake Cortex, as well as Jett AI Advisor and Workflow Creator. BMC CTO Ram Chakravarti said support for multi-AI agent orchestration is still in development.
“We are working on this in two phases. In the core product, you can now call individual agents based on pre-built integrations and inject them into your workflow,” Chakravarti said. “In parallel, we are co-innovating with some of our lead customers towards meaningful use cases where bespoke agents are orchestrated with the Control-M core and tailored with add-on capabilities such as managed file transfer for AI and federated data exchange.”
If your AI use cases are not aligned with your overall digital business strategy, your AI pilot will end up being a science experiment.
Federated data exchange is a process by which query equipment can access potentially sensitive data in a partner’s infrastructure and vice versa, without exporting the information outside the corporate network. This can be an important part of starting work with a new partner. One of Control-M’s pilot customers reduced the federation data exchange process using AI agents from 30 days to less than 12 hours, Chakravarti said. He declined to name the customer or specify the size of the company, other than to say it was a “very large” company.
“We have already built and continue to build many of the pieces of the puzzle, but what’s important to us is how can we deliver predictable results using Control-M, with its dependency management, reliable handoffs, SLA compliance, and traditional set of complexity management,” said Chakravarti. “If the AI use case is not aligned with the overall digital business strategy, the AI pilot will end up being a science experiment.”
Dan Twing
BMC has also undertaken extensive portfolio rationalization over the years, most notably spinning off its IT service management and operations management business and its workload and mainframe automation business into separate companies last year.
“Broadcom still needs to extend [agentic AI support] “BMC didn’t have those challenges. While Broadcom added SaaS two years ago, they had other challenges, such as going through SaaS modernization,” Twing said.
AI expands mainframe modernization opportunities
Broadcom began integrating generative and agent AI into mainframe management by adding MCP servers to its Rally Agile development and Endevor change management software, which supports mainframes alongside distributed systems. Broadcom also supports the open source Zowe framework for hybrid cloud mainframe integration, including Zowe MCP Server. Finally, IBM’s WatchTower observability tool includes AIOps capabilities for mainframes.
BMC’s AMI tools update in April included the AI-driven zAdviser development productivity monitoring tool’s Enterprise Application Analysis report. The existing AMI AI Assistant now includes integration with Mainframe Knowledge Hub and Knowledge Expert Chat, allowing you to retrieve information from sources such as runbooks, tickets, log files, and previous incident resolutions.
BMC’s AMI direction statement goes beyond AMI descriptions and recommendations to learn from past incidents and move towards autonomous AI agent-driven workflows for system and performance diagnostics, development workflows, security validation, and operational recovery.
With this statement of direction, BMC is taking a more holistic and thoughtful approach to AI in mainframe modernization than Broadcom, said Steven Dickens, CEO of HyperFrame Research.
stephen dickens
“Broadcom has put an MCP server on a mainframe and connected it to a bunch of legacy applications, so you can kind of interrogate them through the MCP server. This seems to me more like table stakes than a comprehensive deployment of AI,” Dickens said. “BMC is thinking, ‘How do we bring in support data? How do we bring in Redbooks and support databases and knowledge bases? And then how do we explain the code in a broader way of thinking and automate operations?'”
Will history repeat itself with IBM and Arm?
In Dickens’ view, BMC has the most ambitious mainframe strategy, but IBM also has concert AIOps software that supports System Z automation and control of mainframe hardware, which it achieved with its recent support deal for Arm chips.
IBM made a similar effort to integrate x86 chips into its zBX systems more than a decade ago, and while they can now support most major enterprise workloads, there are some known issues in areas such as storage resource management and, in some cases, support for third-party applications, Dickens said.
Further opening up the platform to Arm chips could provide new avenues for third-party app compatibility, Dickens said, and there is strong incentive for both sides to make the Arm integration work.
“No matter what anyone says about mainframes, mainframes are highly available, resilient, performant, and the fastest processors on the market,” he said. “Arm now has access to instruction set collaboration with hundreds of chip developers and architects. IBM has some advantages in this area.”
But Dickens said he doesn’t expect to see a shippable result from the partnership until the next-generation System Z launches, likely in 2028, given IBM’s typical three-year System Z release cycle. The latest z17 system was launched in April 2025.
On the workload automation front, IBM’s workload automation was placed in the “Strong Value” category in the October 2025 EMA Radar Report on Workload Automation and Orchestration. It ranked below Control-M and Automic, which are among the tools included in EMA’s top “Value Leaders” category, along with Stonebranch, HCLSoftware, Beta Systems, and Redwood.
But overall, Dickens said, IBM and Red Hat have powerful hybrid cloud agent AI tools that can compete for enterprise buyers.
“When you look at Red Hat, you look at the OpenShift integration that they’re doing on the mainframe. IBM is not just having conversations about mainframe tools, they’re having more comprehensive hybrid IT kind of conversations,” Dickens said.
Beth Pariseau, senior news writer at Informa TechTarget, is an award-winning IT journalism veteran. Any tips? send an email to her or connect linkedin.
rajeev kumar
Dan Twing
stephen dickens
