Legacy data as the real barrier to enterprise AI
Legacy data and systems are the intertwined technical debt that stops enterprises from scaling AI, because outdated applications, siloed databases, and weak governance prevent AI models and agents from accessing consistent, trusted, and timely information across business workflows. Most large organizations already experiment with AI pilots, but they run into a wall when they try to move beyond isolated proofs of concept. Historical systems were built for transactions, not for AI-ready data governance, observability, or model-driven automation. Data lives in overlapping schemas, incompatible formats, and tightly coupled applications that cannot be changed quickly. This makes enterprise data readiness more important than access to any specific foundation model. To scale AI, enterprises must treat legacy system modernization and data preparation as one program, not separate projects, and connect them directly to how AI is deployed, monitored, and governed in production.
IBM and ServiceNow: Making legacy modernization the AI foundation
IBM and ServiceNow are framing legacy system modernization as a condition for enterprise AI rather than a parallel initiative. Their multi-year collaboration focuses on modernizing aging applications, improving AI-ready data governance, and bringing autonomous operations into IT workflows. Decades of deeply interconnected legacy systems remain, in their words, “the biggest barrier to moving fast on AI,” so the plan is to evolve existing systems instead of replacing everything at once. IBM tools such as IBM Bob, Enterprise Application Runtime for Java, and IBM watsonx.data will scan and refactor legacy applications so they can support modern, AI-enabled workflows while still running on the models customers choose. The partnership also extends ServiceNow’s Workflow Data Fabric with IBM enterprise data capabilities. That combination aims to keep data AI-ready—governed, observable, and cataloged—as it moves into ServiceNow-powered workflows, making the platform an orchestration layer for work, data, and AI agents.
AI-ready data governance moves into the workflow layer
The IBM–ServiceNow collaboration highlights how AI-ready data governance is shifting from back-end data platforms into the daily workflow layer. By integrating IBM watsonx.data and related tooling with ServiceNow’s Workflow Data Fabric and Data Catalog, the partners aim to support data quality, observability, and master data management inside the same environment where employees and AI agents work. IBM notes that “AI adoption at scale requires more than access to models. It requires rethinking the systems, data and governance that support them.” That means controlled data definitions, visibility into data lineage, and clear rules for how information flows into AI-driven decisions. As autonomous IT operations are added—using components such as Red Hat Ansible, Instana, HashiCorp Terraform, and HashiCorp Vault inside ServiceNow workflows—the governance framework extends down into infrastructure as well. The result is a more consistent control plane for both data and operations, which is essential for regulated industries.
Unisys and Rafay: Orchestrating AI across hybrid and regulated environments
While IBM and ServiceNow focus on modernization and governance, Unisys and Rafay Systems target infrastructure orchestration for AI workloads across hybrid and regulated environments. Rafay provides a self-service platform for infrastructure orchestration across public, private, and edge locations, including Kubernetes orchestration, while Unisys contributes AI expertise and managed cloud services. Together, they offer a unified intelligent AI software layer spanning agents, models, and modular AI infrastructure. According to Unisys, only 36% of enterprises say they are ready to support large-scale AI workloads, so the partnership aims to close that gap by simplifying deployment, lifecycle management, and governance for AI-intensive and GPU-heavy workloads. By embedding Rafay’s governed, self-service capabilities into broader cloud and application operations, organizations can integrate AI workloads into existing environments and adopt agentic frameworks and workflows with more confidence and consistency.
End-to-end AI: From legacy modernization to orchestrated deployment
Taken together, the IBM–ServiceNow and Unisys–Rafay partnerships show that enterprise AI readiness spans three connected layers: legacy application modernization, AI-ready data governance, and infrastructure orchestration. Enterprises need end-to-end solutions that upgrade legacy applications without starting from scratch, prepare and govern data so it is ready for AI, and manage AI workloads reliably across hybrid cloud AI deployment models. Infrastructure orchestration platforms such as Rafay’s and workflow-centric control towers like ServiceNow’s are becoming critical in regulated settings, where security, compliance, and consistent operations are non-negotiable. The emerging pattern is clear: AI programs that ignore legacy systems and fragmented data will stall at the pilot phase, while those that treat modernization, governance, and orchestration as one continuous architecture will be better positioned to make AI operational, scalable, and sustainable.
