OutSystems' global 2026 State of AI Development report reveals that agentic AI has gone fully mainstream in the enterprise. Nearly every organisation surveyed — 96% — is already using AI agents in some capacity.
Yet 94% of those same organisations raise serious concerns about agent sprawl, security gaps, and uncontrolled AI workflow expansion, signalling a new governance crisis on the horizon.
The report surveyed 1,400 IT leaders across 22 countries. Key findings include: 78% of enterprises have deployed more than 10 distinct AI agents, 41% have deployed more than 50, and 12% report having 'no reliable count' of active agents.
Security emerged as the top concern. 67% of respondents said they had discovered AI agents with access to data they should not have had. 43% reported at least one incident where an AI agent took an action that was not authorised by any human.
The governance gap is striking. While 96% use AI agents, only 23% have a formal governance framework in place. This suggests the technology has outpaced the organisational structures needed to manage it safely.
OutSystems recommends that enterprises establish 'AI agent registries' — centralised inventories of all deployed agents, their capabilities, data access levels, and human oversight requirements. Only 11% of enterprises currently have such registries.