Enterprises across North America, Europe and APAC have rushed to deploy AI agents, but confidence in what those agents are actually doing lags far behind, according to new research commissioned by Boomi.

A Forrester Consulting study of 409 director-and-above IT and technology decision-makers across the three regions found that 86 per cent of organisations have moved beyond the pilot stage for AI agents, yet just 34 per cent say they trust the actions their agents are taking. The gap reflects what Forrester terms ‘agentic chaos’ – deployment that has outpaced the governance needed to support it safely.

The cost of moving too fast

Organisations in the bottom quartile for operational readiness – spanning governance, integration and API/MCP management – are pushing ahead regardless of the gaps: 77 per cent of this group are moving agents into production anyway, exposing themselves to an average of US$2.1 million in added costs from compliance fines, lost customers, downtime and rework. Organisations with stronger readiness, by contrast, report far higher confidence, with 55 per cent expressing trust in their agents’ actions and decisions compared to just 22 per cent among those still in chaos.

“This research confirms what we’re seeing everywhere: the trust problem with agentic AI is really a data problem,” said Steve Lucas, Chairman and CEO at Boomi. “Agents can only be trusted to act on data that’s been properly activated, connected, and governed, and most companies deployed agents before they did that work. The ones who did it first are the ones getting real value now.”

Integration, not smarter models, is the differentiator

The study identifies integration as the clearest line separating enterprises that trust their agentic AI from those that do not. Decision-makers with strong operational readiness were three times as likely as their less-prepared peers to say reliable, well-managed APIs determine whether they pilot a use case at all. Integration platform as a service (iPaaS) adoption showed the widest gap of any method surveyed: 46 per cent of organisations with strong readiness use iPaaS to support agentic workflows, against just 25 per cent of those in agentic chaos.

The gap widens further when it comes to building agents in the first place. Eighty-six per cent of organisations with strong readiness say iPaaS and API management capabilities are important, compared with 58 per cent of those in chaos – the single widest gap Forrester measured in the entire study. Some organisations are now running as many as 200 agents, a symptom Forrester calls “POC/pilot purgatory,” where ambition stalls because agents were never properly connected to the enterprise systems they need to act on.

Where the payoff shows up

Among organisations with strong operational readiness, 59 per cent reported productivity gains from their agentic AI deployments, 51 per cent reported increased innovation, and 46 per cent gained reusable capabilities they could apply elsewhere in the business. Forrester recommends that organisations looking to close the trust gap align their AI and integration teams under one operating model, implement a control plane to govern agents, and introduce an orchestration layer connecting agents to data, applications and each other.

The full study, titled The Agentic AI Readiness Gap: Insights From Leaders At Organisations Scaling Agentic Pilots To Production, is available on Boomi’s website.