IBM plans to apply AI to one of the biggest challenges facing mainframe customers: understanding how changes to complex systems affect the applications and services that depend on them.
Mainframe teams often rely on manual analysis, disconnected tools, and knowledge held by experienced employees to map dependencies and assess risks. As those specialists retire and modernization efforts accelerate, it can become harder to predict the impact of a change before it reaches production.
IBM’s proposed solution, Z Atlas, is designed to create an AI-powered map of an organization’s full z/OS environment, covering infrastructure, middleware, and applications. The system would use that view to help teams with work such as patch management, regulatory responses, and sharing operational knowledge.
The product is still in the statement-of-direction stage. IBM says Z Atlas will combine automated discovery, topology modeling, dependency analysis, natural-language interaction, and AI-assisted workflows to keep a continuously updated picture of the environment.
That information could help teams assess proposed changes before implementation. For example, a system programmer considering a middleware upgrade could identify which applications and operational processes might be affected, determine what should be included in testing, and coordinate validation across teams.
IBM says the goal is to reduce outages caused by changes that have not been adequately assessed and verified. It also sees Z Atlas as a shared intelligence layer for helping organizations understand, manage, govern, and improve their mainframe environments.
The announcement comes as customers weigh modernizing their mainframes in place against moving workloads to cloud or distributed systems. Gartner has warned that poorly planned migration projects can threaten business continuity, and expects many providers focused on mainframe exits to change their business models or leave the market by 2030. The firm has argued that, for many organizations, generative AI may be more useful for modernizing mainframe systems in place than for accelerating a move off the platform.
IBM has also pointed to growing AI activity among mainframe customers. In a July earnings call, CFO James Kavanaugh said nearly half of z17 customers were investing in AI capabilities with the Spyre accelerator. He added that customers deploying watsonx Code Assistant for Z were increasing MIPS capacity three times faster than those that were not.
IBM says customers interested in Z Atlas can register for a preview.