The State of AI in the Enterprise has revealed that 99% of organisations now use AI in some form:
- 83% run agents
- 42% integrate agents into complex workflows
- 19% run agents autonomously at scale
But adoption is not the same as value. And value is not evenly distributed.
According to the report, which surveyed 1,640 IT decision-makers across the United States, United Kingdom, France, and Japan, 50% leading-edge organisations report significant ROI — measurable improvement above 25% on the metric they track. Among early-stage organisations, that figure is 11%.
Businesses capturing the most value from AI today have built the right foundations to make their models work. That includes secure content connections, effective governance, and architecture flexible enough to keep upgrading as AI stacks evolve.
3 foundations that set AI leaders apart
The report identifies three foundations that distinguish the leading edge from everyone else.
Frontrunners have connected content to AI
96% of organisations say it is important (or very important) that agents access company-specific content and knowledge. Yet only 36% have actually connected agents to trusted internal content across many use cases.
The challenge is making enterprise knowledge accessible, usable, and trustworthy for the agents that depend on it. Most enterprises do not have an AI problem first. They have a content problem. Business context (the contracts, policies, meeting notes, and documents that tell AI how your organisation actually works) lives in unstructured data.
Analysts consistently estimate that the vast majority of enterprise data — often cited at 80–90% — exists in unstructured form. When that content is fragmented across systems, AI remains impressive but unreliable. The result is output that looks polished but requires heavy revision because the AI lacked the business context to get it right.
Organisations that have acted on this are already pulling ahead: 63% of leading-edge organisations describe their unstructured data as an active competitive advantage they put to work with AI. Among early-stage organisations, that figure is 26%.Context is the new currency of enterprise AI.
Governance ensures security as AI scales
Organizations reporting established or advanced governance frameworks rose from 24% in 2025 to 73% this year, but only 39% have comprehensive visibility across sanctioned and unsanctioned AI use, and just 34% have formal standards governing how agents access company data. Moreover, 49% of organisations have already experienced an AI-related data exposure incident.
And yet the governance paradox is real: 76% of respondents say their current governance requirements are slowing their ability to deploy agentic AI — while 93% agree that better governance would help them move faster over time.
Governance designed for human workflows and retrofitted onto agents slows AI down. Governance purpose-built for agents — with defined permissions, comprehensive visibility, and controls that make sanctioned AI tools more useful than the alternatives — speeds it up. Governance that accelerates AI is governance purpose-built for agents — with defined permissions, comprehensive visibility, and controls that make sanctioned AI tools more useful than the alternatives.
Combining governance, permissions management, and enterprise knowledge access into a single operational approach makes AI trustworthy enough to deploy at scale.
Architecture flexibility is key
68% of respondents are concerned about being locked into a single AI provider, but the average number of AI tools officially adopted has risen to 3.3.
Many organisations spent the last decade managing SaaS sprawl. Now they’re managing agent sprawl — a proliferation of point solutions, disconnected workflows, and overlapping capabilities that creates new complexity rather than reducing it.
The leading edge is building so it doesn’t have to bet on which model wins.
Hiring numbers that challenge headlines
Against the prevailing narrative, 58% of organisations expect their total headcount to grow over the next three years — rising to 79% among the most mature. Only 9% say AI agents are primarily eliminating roles today. In EMEA, where labour market regulation and works council requirements shape how organisations deploy automation, this finding carries particular weight: the agentic transition is not a headcount reduction story, it is a capability reshaping one.
The roles being added fastest — AI agent operators, governance professionals, workflow automation specialists — barely existed two years ago. The workforce being built around the agentic enterprise will be larger, differently shaped, and concentrated around capabilities that did not exist on most org charts just a few years ago.
What this means for EMEA enterprises
The findings reflect a pattern EMEA organisations will recognise, but with a distinctly regional dimension. EMEA enterprises operate across multiple regulatory environments simultaneously: GDPR, NIS2 in Germany, and the EU AI Act taking operational effect across the bloc.
In the UK, enterprises operate under UK GDPR — the same regulatory DNA, a distinct framework. Fragmented, ungoverned content is not only an AI bottleneck; it is a live compliance risk.
For DACH organisations in particular, localised compliance proof points are not optional. For French and Southern European markets, the architecture flexibility argument carries specific weight: the ability to bring your own AI models and avoid vendor dependency is a strategic preference.
And across the bloc, Europe’s sovereignty conversation is expanding beyond data residency to something more fundamental: knowledge control, permission governance, and operational resilience. This is the argument that distinguishes EMEA’s AI challenge from the global picture. Data residency was the first frontier — where data lives.
The next frontier is what happens to it: which agents can access it, under what permissions, with what audit trail. For enterprises operating across multiple jurisdictions, that is the condition which determines whether AI can be deployed at all.
Read the full State of AI in the Enterprise report
