86% Of Enterprises Deployed AI Agents, Only 34% Trust Them
- •86% of enterprises have moved beyond pilot AI agent deployments, but only 34% trust their performance.
- •Forrester found that operational chaos costs underprepared firms an average of $2.1 million in losses and rework.
- •Enterprises prioritizing iPaaS and API management integration are significantly more likely to report productivity and innovation gains.
A study conducted by Forrester Consulting on behalf of Boomi reveals that 86% of enterprises have moved beyond the pilot stage with AI agents, yet only 34% of decision-makers report trusting the actions their agents take. The survey involved 409 IT and technology executives from North America, Europe, and APAC. Organizations categorized as being in a state of "agentic chaos" face an average of $2.1 million in additional costs due to compliance issues, operational downtime, customer loss, and necessary rework. In contrast, organizations with "agentic control" report significantly higher confidence levels, with 55% trusting their agents' decisions compared to 22% among those in the chaos category.
The research highlights integration as the primary factor distinguishing successful enterprises from those struggling. Companies with agentic control are three times more likely to rely on well-managed APIs for their use cases. Adoption of integration platform as a service (iPaaS) is notably higher among leaders, with 46% using it to support workflows versus 25% of organizations in agentic chaos. Furthermore, 86% of those with agentic control view iPaaS and API management capabilities as critical to their operational readiness. While organizations in chaos often focus solely on refining AI models, 47% of leaders prioritize enhancing integration with existing tools, APIs, and applications.
Operational disconnects remain a major hurdle, with some businesses operating up to 200 disparate agents that lack connection to enterprise systems. Leaders with better control are more likely to have established centralized governance of Model Context Protocol (MCP), a standard enabling agents to connect securely to enterprise systems, at 46% compared to 32% for those in chaos. Additionally, 48% of leaders have aligned their AI and integration teams under a single operating model. These investments yield tangible benefits: 59% of organizations with agentic control report productivity gains, 51% report increased innovation, 46% have gained reusable business capabilities, and 45% have successfully automated repetitive tasks.