Leading AI implementation

Despite its transformative potential, many businesses are still managing artificial intelligence (AI) like a conventional technology rollout – planned, governed and expected to deliver predictable returns. That approach is not just outdated, it’s one of the primary reasons AI initiatives stall.

 

AI breaks the assumptions that underpin most operating models. It does not deliver linear progress. It does not produce predictable outputs early. And it does not justify itself through traditional upfront return on investment (ROI).

 

Instead, value emerges through iteration, discovery and adaptation. Often in ways that cannot be fully anticipated at the outset. AI introduces ambiguity where leaders are used to certainty and rarely delivers predictable returns. The result? Leaders delay decisions waiting for clearer business cases, hesitate to propose ideas that are not fully formed, and evaluate initiatives too early against production-level expectations.

 

Arguably, therefore, the real challenge in AI adoption isn’t technical capability - something many firms have in abundance - it’s actually how organisations, and their leaders, respond to uncertainty.

 

 

The instinct for control can slow progress

Most leadership systems are designed to eliminate uncertainty through detailed business cases, fixed delivery milestones and predefined ROI thresholds. While these mechanisms work well for traditional technology, in AI they often become constraints.

 

This tension’s reflected in the 2026 State of the CIO survey. When asked about the biggest barriers to AI progress, 31% of CIOs cited a lack of clarity in corporate AI strategy, while a further 24% pointed to uncertainty over ownership and accountability of outcomes.

 

This challenge is not rooted in a technology gap. Instead, these are symptoms of operating models that expect certainty before action – something AI rarely provides. AI reverses that equation because progress is inherently non-linear.

 

As a result, in many organisations, many organisations struggle to calibrate expectations around AI adoption. Initiatives are either scaled too early, before they are robust enough, or abandoned too quickly because early outputs are imperfect. In both cases, the underlying issue is the same: leadership models designed for predictability are being applied to a technology that evolves through iteration.

 

Perhaps most important to note is that what can feel like imperfection in the early stages is, in reality, a necessary part of how meaningful outcomes are developed.

 

So in this context, traditional leadership instincts can unintentionally slow progress. When leaders expect fully defined business cases upfront, teams tend to become more cautious about putting ideas forward, and when early outputs are held to production-level standards, experimentation can quickly lose momentum.

 

The result is an all-too-familiar pattern: pilots that fail to scale, initiatives that stall, and growing frustration that “AI isn’t delivering”, despite the underlying technology being more than capable.

 

 

Leadership behaviour is the critical lever of success

If uncertainty is the defining challenge, then, arguably, the way leaders behave becomes the critical lever of success.

 

Organisations do not transform because leaders declare AI a priority; they transform because leaders consistently demonstrate how to operate when outcomes are still uncertain. The conditions for meaningful progress are created when leaders visibly support experimentation, reward learning velocity alongside commercial outcomes, and give teams space to refine imperfect early results rather than demanding full ROI upfront.

 

But behaviour alone isn’t enough; instead, it’s how that behaviour cascades into culture that determines whether AI efforts take hold or stall.

 

In organisations that are successfully scaling AI, leadership intent is translated into a culture of innovation that spreads across the business. When leaders approach AI with curiosity rather than caution, and with energy rather than obligation, that mindset filters through the organisation and shapes how employees engage with change.

 

This cultural dynamic is particularly critical when it comes to data, because becoming a genuinely data-driven organisation is not just a question of architecture or tooling. It requires a shared belief about the power of data, that it can drive more informed decisions and, ultimately, that it can create new forms of value. And without that collective mindset - and the enthusiasm to support it - even the most sophisticated data strategies can struggle to gather pace.

 

 

How organisations turn ambition into success

Across organisations successfully scaling AI, four operating shifts are consistently visible:

 

Firstly, they move from treating AI as a one-off project to building it as an evolving organisational capability. This shifts the focus away from fixed delivery milestones and towards continuous improvement, giving teams the space to build, test and refine over time.

 

Secondly, they replace functional silos with integrated teams. AI initiatives are not owned solely by technology departments; instead, business, data and technical expertise work as a single unit focused on delivering meaningful outcomes rather than exploring theoretical possibilities.

Thirdly, they move from technical exploration to value anchoring. Rather than asking, “What can the technology do?” they start with a measurable business objective and align AI efforts around improving it, even as the delivery path continues to evolve.

 

Finally, they shift from a mindset of perfection to one of iteration. Early outputs are expected to be incomplete, and progress is measured through learning and improvement rather than initial accuracy. These organisations are not necessarily more technologically advanced, but they operate far more effectively in the face of uncertainty.

 

 

Culture turns vision for data into a commercial reality

This isn’t just theory; we now have clear examples of what’s possible when the right culture underpins AI ambition.

 

One leading expert in global identity and location tech recognised the commercial potential of its data and moved quickly to transform fragmented capabilities into scalable, revenue-generating products. Instead of positioning AI as a long-term aspiration, the organisation prioritised rapid execution, launching a new revenue stream just nine months after the programme began.

 

The company consolidated more than 80 siloed products into a connected ecosystem capable of processing tens of billions of data points, enabling automated benchmarking and delivering actionable customer insights at scale. Crucially, this shifted data from an internal asset to a commercial one, which enabled entirely new forms of value creation.

 

Technology enabled the transformation, but culture and a genuine passion for engineering accelerated it.

 

 

A pragmatic path forward

For more traditional organisations, the shift required can feel so steep that it becomes paralysing. It’s important, therefore, to recognise that progress doesn’t require a complete organisational overhaul from day one. The most effective starting point is a focused and practical one.

 

Rather than attempting to define an enterprise-wide AI strategy upfront, organisations should start with two to three high-impact use cases tied to clear business metrics, whether that’s revenue growth, operational efficiency or customer experience improvement. These initiatives should be funded with explicit learning milestones, not just financial targets, and be supported by cross-functional teams with the autonomy to rapidly iterate.

 

This delivers two outcomes simultaneously. One, it creates tangible business value, and two, it builds organisational confidence in working with AI, including comfort with the uncertainty that comes with it.

 

Over time, what starts as a series of focused initiatives begins to shape broader ways of working; teams become more accustomed to experimentation, and leaders grow more comfortable making decisions with evolving information.

 

A matter of leading through ambiguity

AI is not simply a test of technological investment; it’s a test of how effectively leaders can operate in conditions where certainty is limited. Organisations that wait for complete clarity before acting are likely to move slowly, while those willing to learn, adapt and make decisions as understanding develops will be the ones that shape the next generation of competitive advantage.

 

---

 

John Berkin is Client Executive Director at Intellias

 

Main image courtesy of iStockPhoto.com and Krittamet Saehan