AI adoption is entering a more demanding phase. For many UK organisations, access is no longer the immediate constraint. AI tools have been approved and many workflows already incorporate AI. The challenge now is turning that availability into repeatable, confident usage that delivers measurable operational value.
We've previously examined how fragmented data and unclear ownership prevent enterprise AI projects from reaching production. Now we are looking closely at what needs to change when available AI technology has not yet become part of how the organisation consistently operates.
Jump To
- AI Growth Is About Depth, Not Reach
- Why Deployment Does Not Automatically Create Adoption
- Sector Case Study – Fintech & Insurance
- Workforce Readiness Determines the Value Created
- Five Questions To Ask Before Scaling Adoption
- Embedding AI With the Right Execution Capability
AI Growth Is About Depth, Not Reach
The number of organisations using AI is rising quickly. Office for National Statistics data shows that the proportion of UK businesses with ten or more employees using at least one AI technology increased from approximately 12% in late 2023 to around 35% in June 2026. PwC's 2026 AI Jobs Barometer reinforces this, showing that financial and professional services are among the sectors recording stronger growth in AI-related hiring.
However, the ONS research exposes a substantial depth gap. Among adopting businesses, only 10% described their use as extensive. Just 15% said more than half of employees used AI in their daily work, while only 11% reported that most of their workforce had received AI-related training.
Meaningful AI adoption at scale becomes visible when AI is used repeatedly within important workflows, is linked to operational outcomes and is fully understood by the people responsible for reviewing its outputs. The differentiator is not access, it is execution.
Why Deployment Does Not Automatically Create Adoption
A technically successful deployment can still result in limited or inconsistent use.
This often happens when AI is added to an existing process without redesigning the process itself. Employees receive access but little role-specific guidance. They may not know when an output can be trusted, when it should be challenged or who is accountable when it affects a customer or operational decision.
Ownership can also become divided between technology, data, risk and business teams. Each function controls part of the system, but no single leader owns the complete outcome. Meanwhile, success is measured through implementation milestones rather than adoption, processing time, quality, confidence or customer impact.
Enterprise AI adoption therefore requires operating-model decisions. Organisations must define where AI sits within the workflow, how exceptions are handled, who can override recommendations and how informal employee use will be brought within proportionate governance.
Without those decisions, deployment simply creates technical availability.
Sector Case Study – Fintech & Insurance
The current state of AI adoption in fintech and insurance illustrates just how technological maturity can outpace organisational depth.
Research from the University of Cambridge Judge Business School found that 47% of fintech respondents had reached advanced stages of AI adoption, compared with 30% of traditional financial institutions. Fintechs were also ahead in agentic AI adoption, at 57% compared with 45% of established institutions.
Similarly, the latest published Bank of England and FCA breakdown found that 95% of responding insurers were using AI. Across financial services, firms expected the median number of use cases to increase from nine to 21 within three years.
However, while technology supports faster experimentation and integration across onboarding, fraud detection, payments, customer support, affordability assessments and financial guidance, that speed must be matched by accountability, resilience and consumer protection.
This is underlined by FCA consumer research which has found that one in five UK adults would be open to AI making financial decisions for them, particularly in complex areas such as debt advice, pensions and investments. These are not guaranteed outcomes but they nevertheless demonstrate why firms must establish boundaries, permissions and redress before autonomous activity becomes more common.
Governance and change capability must develop alongside technical capability rather than being treated as controls imposed after the event. Integration also matters. New capabilities must work with established policy and claims platforms, while leaders retain visibility of third-party models and infrastructure dependencies.
Workforce Readiness Determines the Value Created
The next stage of AI adoption is fundamentally a workforce challenge.
The Financial Services Skills Commission describes AI as a structural, system-wide shift affecting business models, operations, tasks, roles and skills. It suggests that many roles could experience gradual automation of approximately 30% to 50% of their component tasks. That does not mean the roles will disappear. It means work will be redesigned, with greater emphasis on human judgement and oversight.
Organisations will need capability across several connected groups:
- Leaders who can align investment with commercial and operational priorities.
- Data and engineering professionals who can build and maintain reliable systems.
- Architects who can integrate AI with existing technology estates.
- Business analysts and product owners who can translate operational problems into workable use cases.
- Risk, legal and governance professionals who can provide proportionate oversight.
- Programme and change leaders who can redesign workflows and support adoption.
- Domain professionals who can assess outputs using sector knowledge and judgement.
Five Questions To Ask Before Scaling Adoption
Before expanding AI use, leaders need a clear view of where AI adoption is already happening, what outcomes it should improve and who is accountable for the results. These five questions provide a practical framework for scaling AI with greater confidence.
- Where is AI already being used? Map approved systems and informal employee use. Leaders need an accurate view of current behaviour before they can manage it.
- Which workflow or business outcome should change? Avoid targets based only on licences, prompts or use-case numbers. Define the operational improvement required.
- Who owns the output and the consequences? Clarify responsibility for validation, overrides, errors, escalation and customer impact.
- Which capabilities should be built, hired or accessed temporarily? Separate immediate delivery gaps from the skills and ownership the organisation must retain over the longer term.
- How will adoption and value be measured? Consider repeated usage, processing time, exception rates, quality, employee confidence, customer outcomes and risk incidents.
Embedding AI With the Right Execution Capability
Different stages of AI adoption may require different resourcing approaches.
Where contract recruitment addresses immediate or time-bound delivery gaps, permanent recruitment builds retained capability and institutional knowledge. Resource-as-a-Service provides greater control and visibility across a defined requirement, while Talent-as-a-Service supports scalable, longer-term hiring.
SPG Resourcing brings expertise across Change, Transformation and Data, supported by technically certified recruitment consultants, structured internal technical interviews and a transparent client and candidate process. Assessment is based on delivery context, technical understanding and the outcomes a role must support, not keyword matching.
