IBM i: A Natural Platform for Agentic AI

Adam Shedivy and Jack Woehr at PowerUp 2026

At the April 2026 COMMON PowerUp conference in New Orleans, IBM’s Adam Shedivy presented a compelling vision of IBM i as an “agentic-native” operating system. Adam observed that IBM i’s integrated architecture provides what agentic AI needs: business data, operations, security, authority, and auditability.

Let’s see what makes IBM i an agentic-native platform.

Integration is the key

While useful for humans, the integrated nature of the IBM i platform becomes even more potent for agentic AI. Here, the operating system, Db2 database, object model, user profiles, job management, command environment, security model, audit journals, and system services are part of the same platform.

As a result, authorized agents can observe, understand, and take action, all within established, governed interfaces, rather than a loose collection of disconnected tools.

SQL services add even more consistency for agents

Using the growing collection of IBM i Services and Db2 for i Services, agents can query system information with SQL instead of relying on green-screen commands.

For agents, SQL Services create a consistent interface to the platform that can help answer questions such as:

• Which jobs consumed the most CPU overnight?
• Which user profiles have authority settings that deserve review?
• What changed on the system recently?
• Why did a particular order, invoice, or batch process fail?

Security Must Be Enforced by the Platform

Agentic AI becomes risky when an agent can act without boundaries. Although a prompt can suggest that an agent not access payroll data, a better design is to run the agent under an identity that cannot access payroll data.

IBM i has a mature security model that controls access in a consistent manner throughout the database and the operating system itself, with additional database safeguards such as Row and Column Access Control (RCAC). These controls are enforced universally for all users and applications, including AI agents.

A new view of IBM i in the agentic era

As we all move forward modernizing our applications, we leverage IBM i as a near-ideal environment for putting agentic AI to work, even being “agentic-native,” as Adam Shedivy puts it.

We can help you get started with AI with IBM i. Get in touch for an AI strategy session or for any other assistance in turning old AS/400 perceptions to modern IBM i power.