IBM ties agentic workloads to data, automation and hybrid cloud

IBM's AI operating model reflects a shift from deploying AI features to running AI continuously. As Gunnar put it, successful AI use means workloads that are continuously being utilized, which pressures enterprises to be always on. That has implications for how existing systems are integrated.

Gunnar described the model as bringing together four components — data, workloads, automation and hybrid cloud — that enterprises may already obtain from different providers. The emphasis on integration with SAP and Salesforce systems suggests the challenge is less about replacing core systems and more about connecting them to new agentic workflows. Governance is a central part of that connection: Gunnar argued that moving governance closer to the data, including access rights and provenance, helps agents make better decisions.

Human judgment remains part of the model. Gunnar said prompts can convey some employee knowledge, but capturing tacit knowledge is an ongoing challenge, and essential systems still rely on people in the loop. The interview was conducted ahead of IBM TechXchange 2026, suggesting the company intends to continue developing these themes with clients.