The pace of AI innovation means that conversations around enterprise AI adoption have already shifted significantly from managing stagnant isolated IT projects to scaling organisation wide agentic workflows across core business functions. As a result, businesses are grappling with what it really takes to embed agentic AI into the fabric of an organisation, and most importantly, make it work securely at scale.
Our 2026 report on agentic AI highlights that organisations are increasingly focused on execution rather than the theoretics. Questions around building trusted autonomous systems, maintaining oversight without undermining agility, and creating governance structures that allow innovation to accelerate rather than stall are now taking centre stage.
Encouragingly, investment appetite suggests organisations are committed to expansion, with nearly three-quarters planning to increase spending on agentic AI over the next year. However, increased investment alone will not guarantee meaningful business impact. Unless the barriers preventing scale are addressed, larger budgets may simply produce more pilots rather than deliver transformative business results. Across EMEA, those barriers are well understood. The core challenge is not a shortage of talent, but security and data privacy concerns, with more than half of organisations in the region citing both as the biggest obstacles to scaling agentic AI initiatives.
Moving beyond experimentation
Agentic AI cannot deliver its full potential while projects remain stuck in the pilot phase.
To date, most deployments have focused on IT operations, but that’s beginning to change. More organisations are applying agentic AI in repeatable, customer-facing areas like customer support, where the impact is even more visible. Even sectors that have traditionally been slower to adopt AI, including legal services, are expected to accelerate automation efforts over the next few years, showing that confidence in more complicated use cases is building.
What is becoming increasingly clear is that the capabilities used to streamline IT workflows can also transform customer experiences and drive commercial growth. Yet, as AI moves closer to customers and those core decision-making processes, the stakes naturally get higher, making robust oversight and safeguards more important than ever.
Trust as the foundation for scale
For organisations looking to move from pilot to production, security and privacy remain essential requirements. However, for many leaders, the greatest challenge is not the complexity of the technology itself, but trust.
Establishing trust and confidence requires clear boundaries for when an AI agent can act independently and when human oversight is needed. With agentic workflows, trust has become the ultimate control mechanism.
As it stands, nearly 70% of agentic AI decisions are currently verified by humans, and almost half of organisations conduct a human-led review of AI outputs as a verification measure. This reflects a deliberate middle ground. Organisations are neither embracing full automation nor reverting to entirely human-driven workflows.
Instead, a more balanced model is emerging. Human judgement and agentic AI are working together – AI executes with speed, while humans provide direction and guardrails. Rather than replacing human capability, agentic AI is proving most valuable as a tool that enhances and extends it.
As adoption deepens, leaders must understand and apply this division of responsibility. AI may execute tasks and actions, but humans must continue to set objectives, and boundaries, while critically, remaining accountable for outcomes.
The missing link in autonomous systems
Business observability is the foundation that makes sustainable human-AI collaboration possible, providing the transparency and traceability needed to build confidence at the human-AI interface.
As agentic systems grow more autonomous and interconnected, their complexity also increases. A small error in one model component – a hallucinated output or a misinterpreted prompt – can quickly cascade across applications and environments. The fact that many teams still manually review agentic AI communication flows highlights a critical gap in real-time, context-aware automation.
Without end-to-end visibility, organisations are forced into a reactive stance – diagnosing issues only after they’ve already introduced risk. Traditional approaches that rely solely on logging events or flagging anomalies after the fact are no longer sufficient, and modern organisations need systems that can identify hallucinations, predict downstream consequences, and intervene before issues escalate.
Turning ambition into business impact
Escaping pilot purgatory therefore requires organisations to build a foundation of trust that allows agentic AI to operate securely and effectively at scale. Trust cannot be treated as an afterthought or added later but embedded into systems from the outset through strong governance, oversight, and observability.
Ultimately, the future of agentic AI will not be defined by the volume of experimentation, but by the ability to operationalise successful initiatives. When trust is engineered into an organisation’s system, pilots become pathways to production and create long-term business value rather than remaining isolated pilots.
Joshua Clay
Joshua Clay is RVP Solutions Engineering at Dynatrace – leading the strategic, enterprise and acquisition segment of Dynatrace for solution engineering across UK&I.
