A sweeping trust deficit is holding back artificial intelligence (AI) deployment in corporate America, even as investment surges and early pilots multiply across industries.
Just 6% of companies fully trust AI agents to autonomously handle their core business processes, according to a Harvard Business Review Analytic Services study of 603 global business and technology leaders conducted in July and released Tuesday. The research, sponsored by Workato and Amazon Web Services, reveals a stark contradiction: enthusiasm for AI is widespread, but confidence in letting it run critical operations remains scarce.
The pattern reflects deep-seated caution: Nearly half of those polled confine AI agents to limited or routine tasks, while 39% restrict them to supervised scenarios or peripheral workflows. Companies are willing to experiment, but remain reluctant to grant AI unsupervised authority over finances, customers, or workforce decisions.
Despite this wariness, adoption is accelerating. Nine percent of organizations have already fully deployed agentic AI systems—those capable of making decisions and taking actions with minimal human oversight. Half are piloting or exploring use cases, and 86% expect their investment to increase over the next two years.
Yet the infrastructure supporting this ambition lags dangerously behind. Only 20% say their technology systems are fully prepared to support agentic AI for core processes. Just 15% feel confident about their data readiness, and a mere 12% believe their risk and governance controls are adequate. The study classifies 27% of organizations as readiness leaders, 50% as followers, and 24% as laggards.
Companies deploying AI are seeing real benefits such as improved productivity, cost reduction, and enhanced customer experience but gains consistently fall short of expectations. Security concerns dominate the barrier list, with 31% citing cybersecurity and privacy worries as primary obstacles. Data quality anxiety, unprepared business processes, and infrastructure limitations follow closely.
In response, 74% of organizations are implementing or planning “enterprise orchestration” — connecting systems, data, and applications through a governed framework designed to safely power AI at scale.
The human element may prove decisive. Kim Huffman, chief information officer at Workiva, warned that required change management and reskilling efforts have been underestimated. Nearly half (44%) of organizations are prioritizing employee training in AI oversight, while 39% focus on building governance frameworks.
Industry observers see 2026 as a pivotal year. John Callery-Coyne, chief technology officer of ReflexAI, predicts the shift from experimental to foundational AI, with organizations facing mounting pressure to demonstrate return on investment. He anticipates agentic AI will evolve from programmer tools into mainstream personal assistants, intensifying debates around privacy, transparency, and accountability.
Despite current hesitation, 72% of respondents believe AI benefits outweigh risks. As enterprises invest in orchestration, governance, and workforce preparation, the trust gap may narrow — potentially transforming AI agents from experimental helpers into trusted stewards of core business operations.
