CONSHOHOCKEN, PA — Most large organizations have moved beyond piloting AI agents, but only about one-third trust the decisions those systems make, according to a Forrester Consulting study commissioned by Boomi, highlighting operational and governance challenges that continue to hinder enterprise AI deployments.
The survey of 409 IT and technology decision-makers across North America, Europe and the Asia-Pacific region found that 86% of organizations have deployed AI agents beyond the pilot stage, while only 34% reported confidence in the actions those agents take.
The findings suggest companies are pressing ahead with production deployments even as governance and systems integration lag. Organizations identified as operating in “agentic chaos” — the bottom quartile for readiness across governance, integration and API management — reported moving AI agents into production 77% of the time despite lower operational maturity.
Those organizations incur an average of $2.1 million in additional costs from compliance penalties, customer losses, operational downtime and rework, according to the study.
By comparison, organizations classified as having “agentic control” were more cautious and substantially more confident in their AI deployments. Fifty-five percent reported high confidence in their agents’ actions and decisions, compared with 22% of organizations in the lowest-readiness group.
The research points to enterprise systems integration, rather than advances in underlying AI models, as the strongest differentiator between organizations that trust their AI agents and those that do not.
Organizations with stronger operational readiness were nearly three times as likely to indicate that well-managed application programming interfaces (APIs) determine whether an AI use case proceeds. They also reported broader adoption of integration platform as a service (iPaaS) technology, with 46% using iPaaS for agentic workflows compared with 25% of organizations in the lower-readiness group.
The gap widened when organizations evaluated the technologies required to build AI agents. Eighty-six percent of respondents with higher operational readiness identified iPaaS and API management capabilities as important to deploying AI agents, versus 58% among organizations classified as operating in agentic chaos.
Steve Lucas, Boomi’s chairman and chief executive officer, argued the findings indicate that enterprise AI performance depends more on data connectivity than model sophistication.
“This research confirms what we’re seeing everywhere: the trust problem with agentic AI is really a data problem,” Lucas said. “Agents can only be trusted to act on data that’s been properly activated, connected, and governed.”
The study also found organizations with stronger governance were more likely to centralize oversight of the Model Context Protocol (MCP), a standard that enables AI agents to connect with enterprise systems, and to align AI and integration teams under a single operating model.
Among organizations with higher operational readiness, 59% reported productivity improvements from AI agents, while 51% cited increased innovation, 46% reported creating reusable business capabilities and 45% reported automating repetitive tasks.
Based on the findings, Forrester recommended organizations align AI and integration teams, establish centralized governance for AI agents and deploy orchestration capabilities that allow AI systems to securely connect with enterprise data, applications and other agents.
Support the local news that supports Chester County. MyChesCo delivers reliable, fact-based reporting and essential community resources—free for everyone. If you value that, click here to become a patron today.