A former IBM exec now taking the lead on "agentifying the mainframe" at BMC argues that bringing mainframes to AI is backwards: the best approach is not to force mainframe systems into the AI world, he says, but to bring AI to them.
After 24 years at IBM, where he served as vice president of product management for IBM Data and AI, Tarun Chopra joined BMC in October as senior vice president and general manager of Intelligent Z Optimization and Transformation. He also held leadership roles at IBM encompassing IBM Z mainframes and oversaw the integration of HashiCorp tools with the Z platform.
Companies operating critical mainframe systems can't afford to compromise the performance and reliability that's made them irreplaceable for decades, Chopra said in an interview on the IT Ops Query podcast this week. Nor can they repeat the mistakes of the past, when mainframes failed to integrate effectively with distributed and cloud-native systems.
"In the past, the innovation happen[ed], and then 10 to 15 years later, it [came] into the mainframe ecosystem," Chopra said. "By that time, the perception already builds: mainframe is old, mainframe is legacy. Maybe this time we have learned our lessons from the cloud wave and the client-server waves."
Attempts to eliminate the mainframe entirely in favor of new technology haven't completely succeeded, because few distributed systems can match the raw processing power of the mainframe or always directly replace critical programs written decades ago, he said.
In Chopra's view, mainframe modernization increasingly focuses on improving the integration of surrounding systems with the sticky platform, rather than undertaking a cost-prohibitive process of replacing it – only to, at best, replicate its existing capabilities.
"If you are a CIO or a CTO, why are you going to spend billions of dollars on re-platforming something when the value proposition to your end clients is the same [at the end]?" he said.
Lessons from HashiCorp and IBM Z
Instead of making mainframes AI-native, Chopra said, it's AI agents that need to become mainframe-native. And he has the recent experience of bringing HashiCorp's Terraform and Vault into the IBM Z fold to back it up.
"What Terraform is doing for mainframe is applying standard infrastructure management knowledge into the mainframe," he said. "So when you have a distributed [systems] person who's working on Terraform and has to operate in the mainframe, he or she doesn't have to understand mainframe – he or she just needs to use the Terraform language and create an object on the mainframe."
BMC's strategy to "agentify the mainframe," by encapsulating mainframe management expertise for use by AI agents behind MCP servers, tools and APIs officially kicked off earlier this year with the issuance of a statement of direction for its Automated Mainframe Intelligence (AMI) product.
The company has begun to execute on that strategy under Chopra's direction. On Oct. 20, it launched BMC AMI Agentics, which uses AI to advise IT operators on investigating mainframe issues, decipher legacy automation scripts and guide performance and optimization decisions.
"To me, the real power of AI is … capturing that 60 years of knowledge [about mainframe systems] and helping facilitate and pass it on to the next generation," Chopra said. "That has much, much, much higher ROI than any [direct agentic] automation I will do on the [mainframe] platform."
Beth Pariseau, lead editor for infrastructure news at TechTarget, is an award-winning veteran of IT journalism. Have a tip? Email her or connect on LinkedIn.