Winning with AI: What Leaders Are Gettin Right - And What's Still Holding Them Back

Nick Eaves

Chief Customer Officer

Winning with AI: What Leaders Are Gettin Right - And What's Still Holding Them Back

Posted on 22 December 2025

AI has moved beyond the hype cycle. For most organisations, it has become an unavoidable leadership challenge, not a technological one. Everyone feels the pressure to “do something with AI”, yet few feel confident they yet have the structure, clarity, or capability to move from ambition to execution.

This was the starting point for our recent AI Leadership Event Winning with AI: Your Pathway to a Smart, Affordable Strategy - bringing together more than 60 leaders from Finance, Transformation, Data and the wider C-suite.

Joining me on stage were three people who have lived AI transformation at the sharpest end of global industry:

Andy B. - former senior leader at Apple, Google & Nest, now Co-Founder at GT
John Harris - former Head of Global Architecture at GSK, CTO at Aimia, now Co-Founder at GT
Merxhan Bajrami - Head of AI & Data Science at GT and NASA award-winning engineer

Our goal was simple: To cut through hype and give leaders the clarity, tools and confidence to act – without big-consulting dependency or multi-year programmes.

We also shared early findings from Stanton House’s AI Leadership Survey, completed by close to 200 UK leaders. The headline was uncompromising:

Leaders believe deeply in AI’s potential — but most don’t currently feel equipped to lead it.

“AI isn’t a technology problem; it’s a leadership one. The barrier isn’t access to tools - it’s knowing how to use them well.” — UK CFO

The rest of this article blends those survey insights with the richest thinking from the event - the lessons, stories and practical guidance that leaders need right now.

Article content

High Ambition, Low Readiness: Why AI Efforts Stall

AI has become a strategic priority for nearly every leadership team. 60% of our survey respondents rated it a Level 4 or 5 (out of 5) board priority. The intent is there. The ambition is high.

But readiness? Far lower.

Only 12% say their organisation has a clear, enterprise-wide AI strategy. Almost half (48%) are still experimenting in pilots, and another 35% are implementing AI only within isolated functions.

This emerging pattern - ambition outpacing practical application - is driving what many leaders described as pilot paralysis: a widening gap between early proofs of concept and scaled, repeatable impact.

Underneath this sits a familiar strategic trap: starting with the technology, not the business problem.

Two survey respondents captured it plainly:

“AI is being treated as a solution before leaders even define the problem.” — Transformation Director

“AI is the opportunity to finally manage our data effectively and drive focused outcomes - the risk is jumping to shiny outputs before the groundwork is done.” — Data & AI Leader

One of the most common traps leaders fall into is spending months comparing tools, vendors, and platforms — trying to make the “right” technology choice before making any progress at all. John Harris was unequivocal on this point:

“You don’t need to understand every platform; you just need to understand what you want AI to do for your business.” — John Harris, Co-Founder, GT

He used a simple metaphor on the night: You don’t have to understand trains, planes, and bikes to know that transport is possible.

AI works the same way. Define the destination — the problem, opportunity, or workflow that needs rethinking — and your architects and technical partners can build the route.

The message to senior leaders was clear: Stop obsessing over systems. Start owning the strategy.

During the panel discussion, Andy Baynes reframed this gap powerfully. The last 18 months haven’t changed the fundamentals of AI; they’ve simply made the tools more visible and accessible.

“What’s changed is accessibility, not physics. The danger is we mistake ease of access for ease of impact.” — Andy Baynes, Co-Founder, GT

And the gap between intent and impact becomes even clearer when you look at the organisations that are making real progress. While most are stuck in pilots, a smaller group is quietly moving beyond experimentation and are delivering measurable business value — not because they have bigger budgets or more advanced tooling, but because they take a fundamentally different approach to execution.

What do the numbers tell us?

Article content

What the early movers do differently

The organisations breaking out of pilot mode share a common pattern in how they execute AI work. They:

Start with real business problems, not AI curiosity
Design small, value-led steps rather than sprawling programmes
Give clear ownership to the teams doing the work
Measure progress in weeks, not years

These are practical, repeatable ways of working — not big-budget transformations. And across the panel, the message was consistent: AI is not a monolith and it is not magic. It will not transform your organisation in one leap, but it can transform hundreds of workflows incrementally.

The organisations making real headway understand this. They know the difference between hype and practical value. And they build AI capability the same way any organisation builds a strategic muscle: one grounded, measurable step at a time.

All of which points to a deeper truth…

Article content

Why Behaviour, Not Technology, Is Holding Organisations Back

If ambition isn’t the problem, and access to tools is no longer the barrier, then why are so many organisations still struggling to turn AI excitement into real outcomes?

Our survey and the evening’s discussion landed on the same conclusion: the biggest blocker to AI progress today isn’t technical capability - it’s leadership behaviour.

For many leaders, AI represents the first major transformation where they’re expected not only to sponsor change, but to shape it. This is uncharted territory. AI forces leaders to rethink how they frame problems, how they make decisions, how they empower teams, and how they measure progress. Across industries, three behavioural gaps are emerging most clearly:

1. Leaders are trying to apply old habits to a new kind of problem

AI doesn’t behave like traditional technology. It’s fast-moving and deeply dependent on context and data quality. Yet many organisations still approach it with the same mindset they apply to multi-year IT programmes.

As Andy put it on the night:

“If you try to run AI like a traditional programme, you will slow it down to the point of irrelevance. AI inside the enterprise must be treated as a product. Not a project. Not an initiative. A product that evolves, iterates, and improves with feedback.” — Andy Baynes, Co-Founder, GT

Leaders who succeed treat AI more like an evolving capability rather than a project - something that matures through short cycles, experimentation, and rapid learning.

This distinction matters. Projects end. Products mature. AI needs the latter.

2. The biggest capability gap is in problem framing, not model building

One of the most striking insights from both our survey and the panel was that technical skill is rarely the limiting factor. Most teams can access models, tools, and engineering talent when they need it.

What’s missing is the leadership ability to ask the right questions.

“If you can’t articulate the problem clearly, no model in the world will solve it.” — Merxhan Bajrami, Head of AI & Data Science at GT

Leaders who unlock value from AI excel at problem clarity: defining the issue precisely, being explicit about the outcome they want, and knowing when not to use AI.

This is why the organisations making the fastest progress aren’t necessarily the most advanced technically - they’re the ones whose leaders think with greater clarity.

3. Teams lack the guardrails to experiment safely and at pace

A powerful theme from the evening was that most teams want to explore AI, but don’t yet feel they can do so safely. This isn’t because of a lack of interest or capability, it’s because the organisational environment hasn’t caught up with the speed of the technology.

Only 23% of our survey respondents said that they feel their organisation is well-equipped to manage the emerging risks that AI introduces - from data exposure to AI-enabled phishing, deepfake fraud, and ethical bias. And with 77% lacking a fully implemented AI policy, uncertainty becomes the default.

In this vacuum, many leaders’ instinct has been to restrict or even ban public AI tools. But as the panel made clear, that approach is both unrealistic and counterproductive.

“Guardrails enable innovation. Bans just drive shadow AI.” — Andy Baynes, Co-Founder, GT

Real-world incidents, such as Microsoft CoPilot inadvertently surfacing confidential SharePoint data due to weak permissions, demonstrate why guardrails matter. But they also highlight why bans don’t solve the underlying issue: people will still experiment, only with more risk and less oversight.

AI progress depends on leaders building an environment where experimentation is safe, not stifled. John put it plainly on the night:

“Guardrails clarify what’s safe to test, prevent data missteps, and remove friction between risk and reward.” — John Harris, Co-Founder, GT

Governance, in this sense, isn’t bureaucracy – it’s forward motion. It’s the bridge between innovation and protection. The companies making the most progress build minimum viable AI policies: lightweight guardrails, fast decision paths, and clear rules that remove uncertainty – freeing teams to explore without fear. As Andy warned:

“If people are afraid to try things and fail fast to learn - you’re dead before you even begin.” — Andy Baynes, Co-Founder, GT

Guardrails give teams the confidence to test. Clarity gives them the confidence to learn. And psychological safety gives them the confidence to share what didn’t work. Together, they create the cultural conditions that make meaningful progress from AI possible.

From Big Consulting to Bite-Sized Progress

A core message from the evening was that meaningful AI progress doesn’t need to begin with a Big4 roadmap or a multi-year transformation plan. In reality, many organisations can unlock value far faster — and far more affordably — by starting smaller and focusing on the problems that matter most. John Harris encouraged leaders to rethink where they look for opportunity:

“Find your uneconomic oil wells.” — John Harris, Co-Founder, GT

Every organisation has problems that have sat on the “too hard, too costly, or too slow” list for years. AI changes the economics of these challenges. Suddenly, tasks that once required teams of analysts or months of modelling become feasible in weeks. Examples highlighted on the night included:

Mining decades of scientific journals to accelerate drug discovery
Automating repetitive underwriting decisions
Retailers forecasting operational bottlenecks previously handled manually

These weren’t large-scale AI programmes. They were focused, contextual efforts led by people who understood their workflow, their constraints, and their definition of “good”.

Merxhan Bajrami, Head of AI & Data Science at GT, illustrated this shift through his earlier NASA award-winning work mapping wildfire patterns in North Macedonia. Traditionally, modelling and predicting fire behaviour would have required months of manual analysis. With the right AI techniques, he achieved it in 36 hours — fundamentally transforming how emergency responders prepared for and managed real-time risk.

The lesson wasn’t about scale or complexity. It was about feasibility. AI changed what was possible within a tiny team and a tight time window. Work that was previously considered too slow, too costly, or too complex suddenly became viable. Problems once deemed “uneconomic” moved firmly into reach.

This shift also explains why the traditional big-consulting playbook struggles to deliver early value. Large programmes tend to move slowly, sit too far from the day-to-day, and focus on designing the ideal future state rather than delivering immediate impact. AI doesn’t reward that approach. It rewards momentum, proximity to the workflow, and tight loops between engineers and the business.

The organisations gaining traction are the ones that start by solving something real, work in short cycles, and learn their way forward. They don’t wait for transformation; they build capability through a series of small, contextual wins. Their success isn’t accidental — it’s cumulative.

This mindset — prioritising focused progress over sweeping programmes — is one of the biggest opportunities for leadership teams today. It offers a practical route out of pilot paralysis and creates the conditions for AI capability to take root inside the organisation.

And that leads to the next question: If AI progress depends on small, contextual steps rather than large, consulting-led programmes, then who inside the organisation makes this possible?

Article content

The New Capabilities Every Organisation Now Needs

Across the evening’s discussion, three critical roles emerged — not job titles, but capability archetypes that every organisation will need to embed if they want to move beyond pilots.

1. Translational leaders: bridging intent and implementation

As John Harris put it: “The missing link isn’t technology, it’s translation.”

Translational leaders are the connective tissue of AI progress. They don’t need to code, but they understand enough to connect:

business intent with technical capability
board priorities with data workflows
outcomes with constraints
experimentation with execution

They turn abstract ambition into concrete, solvable problems.

In our survey, only 33% felt confident that their organisation’s leadership was equipped to guide AI-driven change (average confidence rating: 2.9/5). That gap isn’t about technical knowledge — it’s about fluency, vocabulary, and confidence. Translational leaders close that gap.

They give senior teams the clarity to make decisions and give technical teams the direction to deliver against real business problems.

And the need for this capability is only growing. Our survey found that 29% of leaders see lack of internal skills and knowledge as the top barrier to AI progress — more than budget, data quality, or vendor overload. In other words, the limiting factor isn’t the technology itself, but the confidence and capability to use it well.

2. Process redesigners: re-engineering the work, not just the tasks

AI rarely succeeds when it is layered onto existing workflows. It succeeds when those workflows are rethought entirely. Process redesigners are people who:

understand how work actually happens
can map workflows end-to-end
know where the friction, waste, and failure points sit
can spot where AI creates new pathways, not just optimisation
collaborate across Finance, Data, HR, Operations, and Tech

They aren’t traditional business analysts. They combine operations, product thinking, and systems awareness. They understand constraints and know how to deliver “just enough” redesign to make AI effective.

These are the people who prevent organisations from “automating the inefficiency” — and instead help them reimagine the flow of work around AI.

3. Intrapreneurs: the internal experimenters

These are the individuals already experimenting with AI tools in their day-to-day work — often quietly, and often without permission. They:

spot opportunities others overlook
build rough prototypes that spark real solutions
join the dots between pain points and AI capability
learn quickly and share generously
operate with curiosity, urgency, and ingenuity

In many organisations, these are the people closest to the problem and most motivated to solve it. They don’t wait for permission; they start somewhere real. And when supported, they become the internal engine of AI progress.

Merxhan’s engineering perspective reinforced this:

“AI development needs fast feedback loops, simple evaluation criteria, and business partners who can articulate constraints.” — Merxhan Bajrami, Head of AI & Data Science at GT

Intrapreneurs thrive in exactly those environments.

These capability archetypes are what enable organisations to move from bite-sized progress to scalable outcomes — without losing speed, context, or ownership.

And together, they set the foundation for the next shift:

If these capabilities are essential, how is AI reshaping the work people do every day?

How AI Is Reshaping Work — and Redefining Value

If translational leaders, redesigners, and intrapreneurs are the people who make progress possible, then what does AI mean for the work itself? And how does it change the value organisations can create? Across the evening, two realities stood out.

AI lifts the floor of work — so people can raise the ceiling

The panel were clear: AI is removing the administrative, repetitive, and manual parts of work — not the human judgement, interpretation, or decision-making.

In Finance, AI can now automate reconciliations, compile board papers, run variance analysis, and model scenarios in minutes. But the decisions still come from people.

In customer operations, AI can analyse thousands of calls to identify the behaviours of top-performing agents, giving managers new coaching pathways that raise performance across the entire team — not replace it.

In transformation and planning, AI accelerates the slowest parts of the process: gathering requirements, mapping constraints, validating problems, and surfacing insights.

Across all of these examples, the pattern is consistent: AI takes work away from people, so it can give better work back to them — freeing teams from low-value activity so they can focus on solving, interpreting, advising, and improving.

AI creates new value — not just new efficiency

A dominant theme from both the panel and the survey was that many organisations still view AI through a cost-cutting lens. But the leaders getting ahead are reframing AI as a value engine, not a workforce reduction tool.

Examples shared on the night included:

A consumer bank using voice-AI to route customers more accurately, improving both speed and experience.
Retailers predicting operational constraints earlier, enabling better planning and fewer last-minute failures.
Contact centres analysing behavioural patterns of top agents to improve coaching and raise the performance baseline.
Finance teams shifting from manual reporting to higher-quality insight generation.

In all of these cases, AI isn’t about doing the same work with fewer people. It’s about doing better work with more clarity.

As John put it:

“AI’s biggest value isn’t in reducing headcount — it’s in augmenting human capability.” — John Harris, Co-Founder, GT

AI changes what teams can see, what they can solve, and how fast they can act. It collapses the distance between information and decision. It amplifies expertise rather than replacing it.

This is the real prize: performance uplift, capability expansion, and improved customer experience.

Article content

From Hype to Habit: What Leaders Must Do Now

If the past two years were defined by AI hype, the next phase will be defined by something far harder to fake: leadership habits. Not pilots. Not proofs of concept. Not vendor showcases. But the everyday behaviours that turn AI from an experiment into a capability.

And that starts at the top. Get in the “sandpit”

Leaders need their own space to experiment — to get hands-on with AI so they can lead from experience, not theory

John Harris was clear on the night: you cannot guide AI transformation from the sidelines. You don’t need to code, and you don’t need to understand every platform — but you do need to understand how these tools behave. Where they add value. Where they break. What “good” looks like. Leaders who don’t experiment default to noise, fear, or vendor dependency. Leaders who do build fluency, confidence, and credibility.

This experimentation doesn't have to happen in the workplace, try out AI tools to address a problem at home or within your community and you will start to get personal insight into the scope for making transformational impact, for making even previously unachievable tasks highly attainable.

Focus on progress you can repeat — not programmes you can announce

The organisations moving fastest aren’t the ones spending the most; they’re the ones building momentum through small, meaningful, repeatable steps. They solve real problems. They measure real outcomes. They scale what works — and stop what doesn’t. AI maturity isn’t delivered through presentations or roadmaps. It’s built through compounding progress.

Ask the right questions

And for boards, this comes down to asking better questions. Four matter most right now:

1. Have we defined the problems clearly enough for AI to solve? Nothing works without this. Clarity beats complexity every time.

2. Do we have the internal capability — translational leaders, redesigners, intrapreneurs — to turn early wins into sustained progress? The bottleneck isn’t tooling. It’s capability.

3. Do we have the minimum viable guardrails to enable safe, confident experimentation — and does every team know them? Without this, progress stalls or risks escalate.

4. Where is AI already enhancing human judgement or customer experience — and how do we scale that value? The organisations winning with AI see it as a value engine, not a cost lever.

The strongest message from the evening was simple: The companies that win with AI won’t be the ones spending the most — but the ones learning the fastest, governing wisely, and scaling what works.