IBM's Bruno Aziza: agent governance will break

Aziza frames the problem as one of scale rather than intent. Giving employees tools to build agents creates "both a management challenge and an opportunity to improve their work," and in his telling the gap is structural: employees will innovate whether or not a platform exists to keep up. His proposed answer is IBM's AI operating model, which he describes as connecting agent orchestration, data integration, hybrid operations and sovereignty, with alerts so organizations can respond when problems arise.

On sovereignty, Aziza's test has four parts — data, technology, operations and regulation — and he warns that an incomplete answer can push a company "down a silo that will be detrimental to your organization." IBM's Sovereign Core is the product meant to span those dimensions, with more than 200 compliance frameworks mapped to controls, according to the company. Aziza applies the same logic to partner evaluation: buyers should examine how a partner's capabilities work together across the enterprise rather than assume a single offering covers the full scope.

IBM is using TechXchange 2026, running Oct. 26–29 in Atlanta, to put those implementation questions in front of practitioners. Aziza says the agenda is dominated by customer content, with examples from Citigroup, CVS Health and DoorDash, and that the conference is deliberately built around diverse customer environments rather than one vendor's cloud or application. His reasoning, in his own words, is that in a changing world customers turn to peers they trust and ask how a system was designed, and that "building for the dimensions of hybrid, is the only way you can win today."