Legacy data, not AI models, is slowing enterprise adoption
Enterprise AI data readiness is the condition in which an organization’s systems, architecture, and governance make data accurate, connected, and trusted enough to support AI models at scale across business workflows. For many large organizations, decades of intertwined legacy systems and unstructured data remain the main barrier to that state, more than any limit in AI models themselves. Data is scattered across aging applications, with little consistency in definitions, ownership, or quality. That lack of AI-ready data stops agentic AI and copilots from reaching beyond pilots into daily operations. Instead of insight flowing into workflows, teams rely on manual exports, side spreadsheets, and one-off integrations. Until enterprises fix these foundations through legacy system modernization and strong enterprise data governance, investments in AI platforms, GPUs, or new models are likely to deliver isolated experiments rather than scalable value.
Inside the IBM–ServiceNow plan to modernize legacy systems
IBM and ServiceNow have announced a multi-year collaboration that treats legacy modernization as a precondition for AI at scale, not a side project. Their joint application modernization approach uses IBM Bob, Enterprise Application Runtime for Java, and IBM watsonx.data to scan and refactor legacy systems so organizations can evolve existing applications instead of replacing them wholesale. According to ServiceNow’s John Aisien, “Most enterprises have the ambition to deploy agentic AI, but lack the foundation to run it at scale.” The idea is to free core business logic and data locked in old platforms, then expose it through the ServiceNow AI Platform as workflows that support AI agents. This strategy aligns modernization work with AI outcomes, turning technical debt into a structured path toward AI-ready data and hybrid cloud deployment rather than a risky rip-and-replace exercise.
Enterprise data governance at the workflow layer
AI-ready data is not only about access; it depends on governance built into the workflows where AI acts. IBM and ServiceNow plan to extend ServiceNow Workflow Data Fabric with IBM watsonx.data and ServiceNow Data Catalog to bring data quality, observability, and master data management closer to execution. That means AI agents working in IT, finance, or HR use consistent definitions and trusted business context, rather than conflicting data extracts. Raj Datta of IBM notes that “AI adoption at scale requires more than access to models. It requires rethinking the systems, data and governance that support them.” By tying enterprise data governance to workflow orchestration, the collaboration aims to keep data AI-ready as it moves, enforcing controls on how information is cataloged, shared, and used in decisions across hybrid cloud deployment and on-premises estates.
Autonomous IT operations as the execution engine for AI
Modernized applications and governed data only matter if AI can trigger reliable action. The IBM–ServiceNow collaboration targets this execution gap with autonomous IT operations. Planned solutions integrate Red Hat Ansible, IBM Bob, Instana, HashiCorp Terraform, and HashiCorp Vault into ServiceNow IT workflows to detect, remediate, and resolve issues before they affect services. Instead of AI stopping at alerts, workflows can coordinate changes across infrastructure automation, observability, provisioning, and secrets management. This stack turns signals from AI models into closed-loop actions, moving operations toward autonomous IT operations rather than incremental scripting. As these capabilities mature, enterprises can apply similar patterns to business processes: AI identifies risk or opportunity, and orchestrated workflows execute governed responses across distributed systems, regardless of where they run in a hybrid cloud deployment.
What enterprise leaders should do now
The IBM–ServiceNow roadmap, with solutions expected in the second half of 2026, underlines a clear message for CIOs and CDOs: treat data readiness as the starting line for AI, not an afterthought. That means prioritizing a realistic legacy system modernization program, mapping where critical data sits, and deciding which platforms will serve as workflow and data “control towers.” Enterprises should define their enterprise data governance model early, including ownership, cataloging, and quality rules, so that AI-ready data is available when agents arrive. They should also identify where autonomous IT operations can yield quick wins, such as incident remediation or change management, and prepare process owners for AI-driven workflows. Investing heavily in models and infrastructure without doing this groundwork is likely to create well-funded pilots but limited impact. With governance and modernization in place, AI can extend across the enterprise rather than remain stuck in experiments.
