Authored By Julia Hill, Consultant – Interim Accountancy & Finance, Stanton House

Artificial Intelligence has quickly moved from an emerging technology to a boardroom priority for finance leaders. Yet despite the headlines, I don't think the conversation among CFOs is any longer centred on whether AI has a role to play. That question has largely been answered.

The more pressing challenge is something else entirely: how do organisations move from isolated AI experiments to trusted, enterprise-wide adoption that delivers measurable business value?

It's a challenge reflected more broadly across UK organisations. Stanton House's recent AI Leadership research found that while 60% of leaders consider AI a top-tier board priority, only 12% have a fully implemented enterprise-wide AI strategy, highlighting the gap between ambition and execution.

That was exactly the focus of a recent Stanton House breakfast roundtable that I hosted alongside my colleague David Fleming, where we brought together CFOs and senior finance leaders to explore how organisations are unlocking productivity and quality gains through strategic AI adoption.

We were also joined by Matt Johnson, CFO at Kaluza, whose experience leading AI adoption within a fast-growing technology business brought a valuable practitioner perspective to the discussion. Rather than talking about AI in theory, Matt shared practical lessons from introducing AI into finance, balancing innovation with governance and navigating the realities of embedding new technologies into day-to-day operations.

What struck me most wasn't how different each organisation's AI journey was, but how similar the challenges had become. Regardless of sector or stage of maturity, the conversation quickly moved beyond AI hype and into a far more practical discussion about leadership, governance, data and implementation and the following themes emerged which I will explore in this paper:

It all begins with your data

• Start with the business problem, not the technology
• Plot Paralysis is real
• AI is already delivering real value
• There is so much choice when it comes to technology and approach
• The biggest challenge might be Cultural

Throughout the morning and across all of these topics, the message was clear. AI is no longer the difficult conversation. Operationalising it is.

The conversation started with data, not AI

One of the first things that struck me was that very little time was spent talking about AI tools. Instead, the conversation began with data.

I kept hearing the same message from finance leaders: AI will only ever be as valuable as the quality of the data feeding it. Before organisations think about intelligent agents or advanced automation, they first need confidence that the information underpinning those technologies is accurate, structured, accessible and properly governed.

Listening to the discussion, it became clear that many organisations are still investing significant time in cleansing and structuring their data before AI can deliver meaningful value. Years of disconnected systems, inconsistent processes and varying data standards mean that, for many businesses, preparing data remains one of the largest pieces of the transformation effort.

There was also broad agreement that ownership is critical. Data quality cannot sit solely with technology teams. Business functions must take responsibility for the information they generate, while finance has a unique role to play in validating outputs and ensuring the integrity of financial reporting.

One of my biggest takeaways was the evolving role finance leaders see themselves playing in this new landscape. Rather than simply becoming consumers of AI, finance is increasingly acting as the organisation's checks and balances function, providing confidence that AI-generated outputs can be trusted, challenged and explained before they influence business decisions.

In many ways, AI doesn't hide weaknesses in data. It exposes them.

For me, this reinforced a simple but powerful message. Organisations looking to accelerate AI adoption should resist the temptation to treat technology as a shortcut. AI can amplify good processes and trusted data, but it rarely fixes poor ones. In many cases, the first AI project may not involve AI at all, it may simply involve getting the underlying data in order.

Although data accuracy was the theme of the day, there was another sense of progress over perfection and how the 80:20 rule is applied to making sure the data is good enough to get started but avoiding never starting.

Start with the business problem, not the technology

As the discussion evolved, another theme emerged time and time again.

The organisations making the greatest progress with AI are not starting with technology—they're starting with business problems.

Rather than asking "Which AI platform should we invest in?", finance leaders consistently asserted that organisations should first define the outcome they are trying to achieve. Whether the objective is improving forecasting accuracy, accelerating month-end reporting, strengthening cash flow management or providing richer commercial insight, the business challenge should always come before the technology.

I thought this was one of the most important shifts in mindset discussed during the morning.

Instead of beginning with tools, organisations begin with value. Instead of chasing the latest AI capability, they focus on solving specific operational problems with measurable outcomes.

Several attendees described the importance of working backwards from the desired business result: defining success, identifying the data required, understanding the existing workflow and only then considering where AI could genuinely create value. Individual pilots may generate useful learning, but every initiative should contribute towards a broader transformation agenda rather than becoming another disconnected experiment.

This approach closely mirrors Stanton House's wider AI Leadership research, which found that organisations making the greatest progress consistently start with real business problems rather than technology curiosity. Rather than pursuing large-scale AI programmes, they focus on solving tangible operational challenges through smaller, value-led initiatives that can be tested, refined and scaled over time.

As someone who works closely with interim CFOs and finance leaders, I also see the commercial benefits of this approach. AI projects become much easier to justify when success is measured against improved forecasting, faster reporting cycles, stronger cash flow visibility or better commercial decision-making, rather than the implementation of technology itself.

Escaping pilot paralysis

One of the most engaging discussions centred around a challenge many organisations are now beginning to experience: escaping pilot paralysis.

Experimentation is no longer the problem.

Most businesses already have employees using AI in some form. Individual teams are testing different tools, departments are developing their own prompting techniques, and employees are discovering new ways to automate repetitive tasks or improve productivity.

The real challenge is bringing all of that activity together.

How do organisations prevent knowledge becoming siloed? How do they share prompting best practice? How do they avoid multiple departments solving the same problem independently? And perhaps most importantly, how do they move from isolated success stories to consistent, enterprise-wide adoption?

Listening to the discussion, it became clear that scaling AI requires much more than approving additional licences or introducing new technology. It demands a structured approach to experimentation, creating environments where teams can safely test new ideas while ensuring successful approaches are captured, refined and shared across the wider organisation.

Prompting itself became an interesting point of discussion. Many organisations are already discovering that prompting improves dramatically through iteration and experience, yet without mechanisms for sharing those learnings, valuable knowledge can quickly become trapped within individual teams. Over time, this creates inconsistent ways of working and varying levels of capability across the business.

Rather than slowing innovation, attendees discussed the importance of creating common implementation frameworks, sandbox environments, prompt marketplaces and enterprise-wide standards that allow experimentation to flourish while building organisational capability at the same time.

For me, this was one of the clearest indicators that the conversation around AI is maturing. Moving beyond pilots isn't about introducing more AI. It's about creating the leadership, governance and knowledge-sharing structures that allow successful experiments to become repeatable business practice.

AI is already delivering practical value

One of the things I enjoyed most about the discussion was how grounded it remained.

While much of the public conversation around AI focuses on transformational possibilities, the finance leaders around the table were far more interested in practical applications that are already creating value today.

Examples included supporting management reporting, accelerating data analysis, generating policy documentation, summarising large volumes of information and providing greater context behind financial performance. These may not be headline-grabbing use cases, but they are helping finance teams improve efficiency while also raising the quality and consistency of their outputs.

Forecasting, budgeting, planning and cash flow management also featured prominently throughout the discussion. AI's ability to process larger volumes of information, identify trends more quickly and support scenario modelling is enabling finance teams to spend less time gathering information and more time interpreting it.

Importantly, the conversation wasn't simply about saving time.

Throughout the morning, I repeatedly heard finance leaders return to the idea of improving the quality of decision-making. AI was viewed as a tool to strengthen analysis, surface deeper insights and support more informed commercial decisions, not simply as a way of completing existing tasks more quickly.

That distinction really resonated with me. The ambition isn't just to make finance more efficient. It's to make finance more valuable.

ERP capability, AI platforms and technology choice

One of the most thought-provoking discussions of the morning centred on a question that I suspect many finance leaders are currently wrestling with: where should AI capability sit?

For finance teams, the decision is becoming increasingly complex. ERP providers are rapidly embedding AI functionality into their platforms, offering new capabilities across forecasting, reporting, analytics and automation. At the same time, the market is evolving at pace, with specialist AI platforms and large language models (LLMs) offering increasingly sophisticated solutions that sit outside traditional finance systems.

This naturally prompted an interesting debate.

Should organisations maximise the AI capabilities already available within their existing technology stack? Should they build bespoke solutions internally? Or is the answer somewhere in between?

What became clear is that there isn't a universal solution. The right approach will inevitably depend on an organisation's technology landscape, internal capability and long-term strategic objectives.

For some organisations, leveraging AI functionality already embedded within their ERP environment may provide the quickest route to value while maintaining integration, governance and security. Others may require more specialist solutions or internally developed capability to solve more complex or organisation-specific challenges.

The discussion also broadened into the practical realities of managing AI at scale. As different teams begin experimenting with different LLMs, finance leaders are increasingly having to think about enterprise licensing, approved platforms, support models and ensuring employees are using AI consistently and securely across the organisation.

At the heart of this discussion sat the pace of change. Should you wait for the ERP offering? Will this evolve as quickly as proprietary prompts that are not dependent on third parties? What are the dangers of being tied into long term contracts with single suppliers in such a rapidly changing environment? This felt feels like a much more mature conversation than we were having even 12 months ago with many more considerations on the table.

Governance has become a strategic leadership issue

If data dominated the first half of the discussion, governance undoubtedly dominated the second.

As AI adoption accelerates, finance leaders are increasingly recognising that governance can no longer be treated as an afterthought. It has to evolve alongside the technology itself.

Topics including GDPR, authentication, access controls, vendor risk, data leakage and governance frameworks generated extensive discussion. As AI becomes embedded within forecasting, reporting and operational decision-making, finance leaders are increasingly asking questions that extend well beyond the technology.

·       How do we ensure employees are using approved tools?

·       Who has access to sensitive financial information?

·       Can AI-generated outputs be explained, challenged and audited?

·       How do we create guardrails that enable innovation without exposing the business to unnecessary risk?

These are no longer hypothetical considerations. They're practical leadership challenges that organisations are dealing with today.

One theme I found particularly interesting was the balance between governance and innovation.

Every organisation wants employees to experiment, discover new use cases and find better ways of working. Equally, every organisation needs confidence that experimentation is taking place within clearly defined boundaries. The consensus around the table was that good governance shouldn't restrict innovation; it should give people the confidence to innovate safely.

This reflects a broader challenge Stanton House identified in its recent AI Leadership research. While organisations continue to accelerate AI adoption, 77% still lack a fully implemented AI policy or governance framework, leaving many leaders trying to scale AI without the guardrails needed to do so confidently.

What I found encouraging was that governance wasn't viewed as an obstacle. Quite the opposite. Throughout the discussion it was consistently positioned as one of the key enablers of successful AI adoption. When employees understand what tools they can use, what information can be shared and how AI-generated outputs should be validated, organisations are able to innovate with greater confidence and far greater consistency.

The biggest challenge may be cultural, not technical

Although the event focused on AI adoption, one of the strongest themes to emerge was that technology alone will never determine success. People will.

Many organisations have already invested in AI platforms. Far fewer have invested in building the confidence, capability and behaviours needed to use them effectively.

Listening to the discussion, I was reminded that every significant finance transformation I've seen has ultimately been driven by people rather than technology. AI is proving no different.

Finance leaders spoke about the importance of creating environments where people feel comfortable experimenting, sharing ideas and learning from one another. Upskilling was viewed as an ongoing leadership responsibility rather than a one-off training exercise, with organisations recognising that AI capability will continue to evolve at pace.

The conversation also reinforced the growing importance of cross-functional collaboration. As AI becomes increasingly embedded within finance, success will depend on finance leaders working much more closely with technology, data and operational teams than ever before.

Several attendees also reflected on the emergence of a new type of capability within organisations—individuals who naturally bridge finance, technology and process improvement. These "builders" are helping organisations translate business challenges into practical AI solutions, often acting as the link between technical specialists and business stakeholders.

As someone who recruits interim finance leaders, I couldn't help thinking about what this means for the future of finance leadership. Increasingly, organisations aren't simply looking for technically strong CFOs or Finance Directors. They're looking for leaders who can navigate transformation, ask the right questions, challenge assumptions and bring people with them through change.

Rather than replacing finance professionals, AI is changing what organisations expect from them.

As routine activities become increasingly automated, the value of finance shifts further towards interpretation, judgement, commercial insight and strategic leadership. The finance professionals who thrive will be those who combine strong technical expertise with curiosity, adaptability and a genuine willingness to embrace new ways of working.

Beyond the pilot

Reflecting on the discussion afterwards, one thought stayed with me more than any other.

The conversation wasn't really about AI.

It was about leadership.

When organisations first began exploring AI, much of the conversation centred on the technology—what it could do, how quickly it would evolve and what impact it might have.

That conversation has matured.

Today's challenge isn't introducing AI into the business. It's embedding it in a way that is secure, scalable, governed and capable of delivering measurable value.

Throughout the morning, finance leaders repeatedly returned to the same principles. Start with the business problem, not the technology. Build strong data foundations before pursuing sophisticated AI solutions. Create governance that enables safe experimentation rather than restricting it. Invest in people as much as platforms. And view every AI initiative as part of a broader transformation journey rather than as an isolated project.

As someone who works closely with interim CFOs and senior finance leaders every day, I can already see these priorities influencing the conversations organisations are having. Increasingly, businesses are looking for finance leaders who can bridge finance, technology and transformation—leaders who can combine commercial judgement with digital curiosity, challenge AI-generated outputs with confidence and help organisations navigate change responsibly.

Perhaps the strongest message of all was the evolving role of the finance function itself.

As organisations accelerate AI adoption, finance's responsibility extends well beyond producing reports or measuring financial performance. Finance becomes the function that validates the data, challenges the outputs, governs the process and provides confidence in the decisions being made.

Technology may generate answers, but finance provides the confidence that those answers can be trusted.

The organisations that move successfully beyond the pilot won't necessarily be those investing in the most sophisticated AI tools. They'll be the organisations that build the strongest foundations—combining high-quality data, clear governance, capable people and disciplined leadership to turn technological potential into lasting business value.

For me, that's the biggest takeaway from the discussion.

In the age of AI, finance doesn't simply measure trust. Increasingly, it becomes the function that helps organisations earn it.

Sources

_{Stanton House: Future of Finance RoundTable
Stanton House: Winning with AI - Leadership Insights
McKinsey: How finance teams are putting AI to work today}