Article | November 20, 2026

Beyond "stay or migrate": Bringing agentic AI to Oracle classic applications   

A conversation from Oracle AI World in Las Vegas, with John Boccuzzi, president of ISG Research, and Mark Simpson, global leader of DXC's Oracle and Salesforce practice

For years, enterprise IT leaders have faced a binary decision about the Oracle classic applications that still run their businesses: stay where they are or undertake a major migration. Agentic AI is creating a third path: modernize now, while retaining the freedom to migrate when it makes business sense.

The third path

These classic systems include JD Edwards, PeopleSoft, Siebel and E-Business Suite, and many have supported mission-critical operations for 20 to 25 years. Moving to SaaS may ultimately be the right destination, but a large-scale transformation can take years and significant investment. Staying put without creating new value is not compelling either.

In his discussion with John Boccuzzi, president of ISG Research at Oracle AI World, Mark Simpson, global leader of DXC's Oracle and Salesforce practice, described a third path: keep the proven application as the trusted system of record and add an agentic AI layer of digital workers around it. "HR runs through these, your supply chain runs through these, all your finance runs through these," Simpson explained. "These are not systems you want to get wrong."

Bring in the agents

Mark Simpson, global leader of DXC's Oracle and Salesforce practice, and John Boccuzzi, president of ISG Research, met at Oracle AI World 2026 to discuss what enterprises can do with the classic Oracle applications — JD Edwards, PeopleSoft, Siebel and E-Business Suite — that still run their businesses. One opportunity: Add an agentic AI layer of "digital workers" on top of your system of record.


What a digital worker is

A digital worker goes beyond an individual AI agent. It combines specialized agents, enterprise data, workflow, security, governance and human oversight to run an end-to-end process, while the underlying application remains the system of record.

For example, a natural-language experience can sit over a PeopleSoft back end so employees can check a leave balance or look up a policy without replacing the core system. As Boccuzzi put it, the original investment "wasn't a bad investment. It was the best investment at the time," and organizations can add "more versatility up on top."

Why it matters now

"We move away from the entrapment of 'modernization means move to SaaS,'" Simpson said. "We're now moving towards 'modernization means how do I drive value for my business?'" Oracle Fusion may still be the right destination, and DXC has helped many customers transform through it, but customers can now choose the path and timing that create the most value. In practice that means:

  • Continuous patching: Digital workers review release notes, run regression tests and apply security updates, supporting the shift from quarterly to monthly patching.

  • Faster payback: Boccuzzi noted that some AI-overlay investments of roughly $500,000 to $2 million could generate meaningful returns within six to seven months; actual results vary.


DXC proof points

  • 80+ agents already built and grouped into digital workers

  • About 1 million users served by agentic, natural-language access to a decades-old PeopleSoft payroll and HR environment

  • Multi-application platform combining people, enterprise data, workflow, controls and security, including Salesforce alongside JD Edwards or Siebel

  • 25 to 30 years of heritage implementing these applications and guiding clients through outsourcing, tooling and transformation eras


How to start

Simpson advises starting small with a tangible source of business pain, and beginning with the outcome rather than the technology. "Let's start with one digital worker," he said. "Let's just look at one process. Let's look at that end to end... Let's get it working."

Modernize now; migrate when it makes business sense. Identify one process where a digital worker could remove friction, improve responsiveness or reduce operational effort, then use that outcome to shape the broader roadmap, one digital worker at a time.