Enterprise AI: Strategy, Reality and Risk

1. AI is Spreading Fast – But it’s Still Not Fully Embedded

“Workforce access to AI has expanded by 50% in just one year – yet enterprise AI remains underutilised and its productivity and innovation potential is still largely untapped.” Deloitte

AI access across organisations has grown quickly. Deloitte reports that workforce access to approved AI tools has increased by around 50% in just one year. Nearly 60% of workers now have access to AI – but fewer than 60% of those people actually use it day to day.

Even more telling, only one in four organisations has managed to move 40% or more of their AI experiments into production. In other words, AI is everywhere – but it’s not yet fully part of how work gets done.

Sound familiar? Here’s what you need to do:

Provide clarity

For AI to become genuinely useful, organisations need to move beyond simply giving people access to tools. Teams need clarity on why AI matters in their role, how it fits into day-to-day work, and where it adds value. This means shifting away from isolated experimentation and towards shared learning, practical use cases and visible examples of AI being used well. When real-world application is recognised and normalised, confidence and adoption follow.

Establish ownership & accountability

Leaders play a critical role in closing the gap between experimentation and impact. That starts with being explicit about what “production-ready” AI looks like in their organisation and setting expectations that pilots should lead somewhere. Ownership for AI adoption must be clear, with senior leaders accountable for moving initiatives out of testing and into operational use.

Put down the scatter-gun

From a technology perspective, organisations need platforms and infrastructure that support scale, security and consistency. Fragmented tooling slows progress and increases risk. Standardising approaches where possible, and creating clear pathways from pilot to deployment, helps ensure AI efforts don’t stall before delivering value.

In short: Access is a good start – but impact only comes when AI is embedded into everyday work.

2. Productivity Gains are Real – Deeper Change is Rarer

“AI transformation reveals productivity for most, business reimagination for a few.” Deloitte

Most organisations are already seeing productivity and efficiency gains from AI. That’s the good news.

The challenge is that far fewer are using AI to rethink how their business actually works. Only 34% of respondents say AI is driving deep business transformation, rather than incremental improvement.

To turn this around, leaders need to:

Champion curiosity

Many teams are comfortable using AI to work a little faster, but less comfortable questioning long-standing ways of working. A cultural shift is needed to encourage curiosity, challenge and experimentation focused on outcomes, not just efficiency. When people feel supported to rethink processes rather than simply speed them up, deeper transformation becomes possible.

Lead from the top

Leaders need to signal that productivity gains are only part of the story. By setting ambitions that include customer value, service redesign or new ways of operating, leaders give permission to think bigger. Regularly reviewing where AI is being applied – and where it could go further – helps keep transformation on the agenda.

Say goodbye to siloes

Technology choices should enable end-to-end change, not isolated automation. Organisations benefit most when AI tools are connected to core systems and informed by reliable data, using insight from AI to identify where processes can be redesigned ensures technology investment supports meaningful change.

In short: Efficiency is useful – transformation is where competitive advantage appears.

3. AI Skills Are Growing, But Work Design is Lagging

“84% of companies have not redesigned jobs around AI capabilities.” Deloitte

Despite widespread discussion about automation and future skills, 84% of organisations have not redesigned jobs or work structures around AI. Most effort has gone into training and awareness – not into rethinking roles, responsibilities or workflows.

Here’s where we’re seeing the gaps:

AI training isn’t intelligent or tailored enough

Building AI capability is about more than training courses. Organisations need to support ongoing learning and involve people directly in shaping how AI fits into their roles. When employees have a voice in how work evolves, AI feels less like something being done to them and more like a shared capability.

Leaders aren’t leading on AI

Redesigning work around AI is a leadership responsibility. Leaders need to create space for teams to explore how roles may change and be honest about what that means. Clear communication, investment in leadership transformation and thoughtful planning help reduce uncertainty and build trust as responsibilities evolve.

“AI as an add-on”

AI tools should be deployed in ways that complement human strengths rather than compete with them. By redesigning workflows with AI embedded from the start, organisations can ensure technology supports better outcomes. Insights generated by AI can then be used to continuously improve how work is structured.

In short: Without redesigning work, AI’s impact will always be limited.

4. Where AI is Built and Controlled Now Matters More

“Sovereign AI is about more than technology ownership – it’s about strategic independence.” Deloitte

Deloitte highlights the growing importance of sovereign AI – where organisations consider where AI is developed, where data sits and who controls it. More than three quarters of organisations say the location of AI development influences their technology choices.

When thinking about sovereign AI, consider the following:

This is everyone’s responsibility

Sovereign AI raises questions that go beyond technology teams. Organisations need a shared understanding of data governance, regulatory expectations and regional considerations. Bringing these conversations into the open helps teams make informed, responsible decisions.

Leaders need to deepen their understanding

Senior leaders must be clear about their organisation’s stance on data control and compliance. Aligning AI decisions with broader risk, legal and governance frameworks ensures choices made today stand up over time.

Flexibility – built in

Technology selection should consider transparency, data residency and long-term flexibility. Systems need to adapt as regulations change, with governance built in from the outset rather than added later.

In short: AI strategy now includes questions of control, compliance and independence.

5. Agentic AI is Coming Fast, But Governance is Not

“Autonomous AI agents are racing into the enterprise, but oversight is lagging.” Deloitte

AI agents – systems that can act with a degree of autonomy – are expected to grow rapidly, with 74% of organisations planning to deploy them within two years. Yet only 21% say they have mature governance in place.

As agentic AI becomes more capable, the risk isn’t just about what these systems can do – it’s about what happens when they’re deployed without clear guardrails or mature governance.

Autonomous systems that can make decisions, trigger actions or interact with other systems at speed introduce new operational, ethical and reputational risks if oversight is unclear. Without agreed boundaries, accountability and escalation routes, organisations can quickly lose visibility and control over how decisions are being made.

The result is often a reactive approach – stepping in only once something has gone wrong – rather than building confidence, trust and resilience from the outset.

For agentic AI to deliver value safely, governance needs to be designed alongside the technology, not added later in response to incidents.

Here’s where to start:

Human gatekeepers

As AI systems gain more autonomy, organisations need clear, shared understanding of where that autonomy is appropriate. Reinforcing the importance of human oversight and responsible use helps teams feel confident rather than cautious about adopting new capabilities.

Autonomy + accountability

Leadership teams must define who is accountable for autonomous systems and how decisions are overseen. Cross-functional governance – involving legal, risk, technology and people specialists – ensures agentic AI is deployed thoughtfully and safely.

Don’t just ‘set and forget’

Agentic systems require monitoring, auditability and control. Organisations should ensure AI-driven decisions can be reviewed and explained, with safeguards in place before scaling use across the enterprise.

In short: Speed matters – but governance matters more.

6. Physical AI Is Becoming Business as Usual

“Physical AI is rapidly becoming integral to operations worldwide.” Deloitte

Physical AI – robotics, autonomous systems and smart machines – is already in use in 58% of organisations, and Deloitte expects that figure to reach 80% within two years.

Physical AI raises a special set of considerations, because its impact is felt in the real world, not just on screens or in systems. When AI is embedded into robotics, machinery or autonomous environments, questions of safety, reliability and human oversight become central.

Organisations need to think carefully about where responsibility sits, how people can intervene when something doesn’t behave as expected, and how these systems are monitored over time. Physical AI also brings operational teams, engineers and digital specialists into much closer collaboration, which can surface gaps in processes and decision-making if roles aren’t clear. Getting this right requires more than technical readiness – it calls for joined-up leadership, clear accountability and a shared understanding of how intelligent systems should behave in live environments.

Here are some things to consider:

Collaboration is key

When AI moves into physical environments, shared responsibility becomes essential. Encouraging collaboration between digital teams and operational colleagues helps ensure safety, ethics and trust remain central.

Accountability from the outset

Leaders need to be clear about accountability for safety and compliance. Involving operational leaders early in AI decision-making ensures physical systems are introduced with appropriate oversight.

Joined up tech

Reliable infrastructure, clear safeguards and human intervention mechanisms are critical when AI operates in the physical world. Integrating data from these systems into wider governance frameworks supports consistency and control.

In short: Physical AI changes how work happens – and needs careful leadership.

7. Strategy is Ahead of Reality

“Leaders feel more strategically ready for AI than operationally ready.” Deloitte

According to the report, 42% of leaders say they are strategically ready for AI, compared to only 30% who feel operationally ready. In this environment, problems tend to surface quickly.

Ambitious plans can stall as teams struggle with data quality, skills gaps or systems that aren’t designed to support AI at scale. This disconnect often creates frustration – leaders expect progress, while teams feel under-equipped to deliver it. Over time, trust can erode as AI initiatives fail to move beyond presentations and pilots, reinforcing scepticism rather than momentum.

Closing this gap requires leaders to pair strategic intent with practical investment in infrastructure, governance and capability, ensuring the organisation is genuinely set up to turn vision into action.

If you feel confident about your AI strategy, but less confident about your organisation’s ability to deliver it, consider the following:

Vision isn’t enough

Ambition needs to be matched with a strong delivery mindset. Building accountability into AI programmes and valuing execution as much as vision helps close the gap between strategy and reality.

Leaders must invest

Leaders must align AI ambition with investment in skills, systems and support. Prioritising delivery over presentation ensures strategies translate into real organisational capability.

Prepare the ground

Operational readiness depends on strong data foundations, capable infrastructure and the right skills. Addressing gaps through targeted investment, training and governance ensures organisations are equipped to deliver on their AI plans.

In short: Strategy works when organisations are ready to deliver it.

Why This Research Matters – and What Comes Next

Deloitte’s 2026 report shows that AI is no longer experimental – but it also shows how much work remains. Organisations are moving fast, but many are still constrained by culture, leadership capability and operational readiness.

At Mix Works, we see this every day. AI success isn’t just about technology – it’s about people, decisions and how work is structured.

If your organisation is looking for support with AI-led change, leadership capability, cultural readiness or governance, we can help. Our work focuses on making AI practical, responsible and aligned with how people actually work.

Get in touch to explore how we can support your leaders and teams through the next phase of AI adoption – with clarity, confidence and care.