OutSystems report finds only 26% of enterprises ready for AI agent conflicts

A new report from OutSystems, produced in collaboration with CIO Dive and KPMG, finds a wide gap between enterprise confidence in AI agent governance and actual operational readiness. The report, titled “Engineering Business Outcomes: The Shift to Agentic Systems,” says 82% of executives believe their organizations can manage, integrate, and monitor AI agents as they move into production, but only 26% feel well prepared to resolve conflicts across agents, systems, and automated workflows.

The findings are based on a survey of more than 850 IT decision-makers. That sample size gives the report a broad view of how enterprises are approaching agentic AI, even though the research comes from OutSystems, a vendor with a commercial interest in the governance market.

The conflict problem is easy to underestimate. AI agents are increasingly able to act across applications and workflows, not just answer questions in a chat window. When several agents touch the same business system, they can collide over permissions, task priority, data consistency, and accountability. Those are operational conflicts that traditional IT approval processes were not designed to resolve in real time.

Most enterprises are still relying on familiar controls. According to the report, 83% of respondents use existing IT, security, or compliance processes, specialized reviews, or centralized policies and monitoring to oversee agents that can act autonomously. Only 5% have implemented automated controls, audit trails, and formal human review points.

The regional picture is uneven. In Asia-Pacific, 39% of respondents say they are very well prepared to resolve AI agent conflicts, compared with 29% in Europe and 18% in North America. Asia-Pacific also leads on automated governance controls: 14% report having automated controls, audit trails, and human review points in place, versus 3% in Europe and 2% in North America.

Tiago Azevedo, CIO of OutSystems, said the gap is where the real work happens. “Having put agents to work across the business, I’ve learned that safety comes from building control into the platform, not reviewing it afterwards,” he said. The implication is that governance has to move into the runtime environment and the agent’s call path, rather than sitting in a policy document or a periodic review.

Michael Harper, managing director at KPMG, framed the solution in terms of execution. “To bridge the gap between high confidence and true operational readiness, enterprises need the ability to execute policy-as-code and runtime guardrails,” he said. That means turning governance rules into enforceable code and applying them while agents are running, so autonomy does not come at the expense of security or reliability.

The report’s commercial context matters: OutSystems and KPMG both sell tools and services in this area. Still, the core finding is hard to dismiss. As agentic AI moves from pilots into production, the next competitive question may be less about how many agents a company can deploy and more about whether it can audit, constrain, and override them when they disagree. Enterprises should watch for runtime governance features, clear human escalation paths, and vendor claims that can be tested in production.