Why AI transformation fails: The capability gaps holding organizations back

investing in generative AI, copilots, automation and AI agents to improve productivity, accelerate innovation and stay competitive.

Yet despite significant technology investments, many AI initiatives struggle to deliver lasting business value.

The reason isn't usually the technology itself. It's the workforce.

While companies are moving quickly to deploy AI tools, many employees still lack the skills, confidence and decision-making capabilities needed to use them effectively. As a result, organizations often see low adoption, inconsistent usage and limited impact.

During Lepaya's recent webinar on AI workforce transformation, learning solutions expert Sophie Baltus and Joppe Stins, co-founder of Maverx, explored what separates organizations that successfully scale AI from those that don't.

‍

The AI adoption paradox

According to global industry data, the gap between technology deployment and actual workforce adoption is stark:

• 88% of companies use AI in at least one business function. (McKinsey)  

• 95% have seen zero measurable impact on profits.(MIT)
• Only 12% of employees state that AI has changed how they actually work.(Gallup)

‍

"Many organizations are trying to move faster than their people are ready for,"

notes Sophie Baltus, Learning Solutions Expert at Lepaya. "They focus on deploying the AI technology without developing the capabilities that people actually need to use this new technology confidently and effectively".  

‍

Why tool rollouts stall before they scale

Regardless of sector. AI transformation breaks down when companies overwhelm employees with too many tools at once, skip foundational understanding before jumping to use cases, focus training entirely on tools rather than the thinking behind them, and expect organization-wide change to emerge from individual experimentation alone.

‍

That last point matters most for anyone building a business case:

‍

"If you want a transformation to happen on an organizational level, this will not automatically follow from individuals experimenting on an individual level. It really is key to start seeing AI transformation as a business impact topic," said Sophie Baltus.

What HR and L&D leaders say about employee readiness

A live poll during the webinar asked attendees how they'd describe employee sentiment toward AI in their own organization today. 

‍

‍

Perhaps most notably, almost nobody viewed outright resistance as the dominant challenge.

‍

Instead, organizations are facing something more nuanced: employees want to use AI but often don't know where to begin or how it fits into their work.

‍

Joppe Stins observed that strategic clarity directly correlates with workforce confidence:

"Employees at companies that take a clearer standpoint on AI feel a lot more comfortable with the tools that they're working with and navigating rapid changes".

The four-stage AI capability journey

One of the central ideas from the webinar is that successful AI transformation happens in stages.

Rather than expecting organization-wide transformation overnight, companies should build capabilities progressively.

Stage 1: AI understanding

The first priority is reducing anxiety. Employees need AI literacy, foundational knowledge, and confidence before they can meaningfully adopt new tools.

Stage 2: Individual productivity

Once employees understand AI, they can begin applying it to improve their own work through better prompting, automation, and workflow optimization.

At this stage, individuals begin experiencing tangible productivity gains.

Stage 3: Work redesign

Organizations can then redesign processes, teams, and decision-making to embed AI into daily operations rather than treating it as a standalone productivity tool.

Stage 4: Organizational transformation

Only after these foundations are established can organizations scale AI strategically across departments and business functions.

Most organizations try to skip straight to stage two or three. Stins explained why that backfires:

‍

"What I oftentimes see organizations do is they deploy a tool because they notice their people are already leveraging AI - maybe shadow AI, as I like to call it, meaning a tool that is not federated, not supported by the company itself. We strongly believe it starts with understanding. Only then can we work towards individual productivity."

‍

Stage four is where continuous improvement becomes a cultural expectation. As Stins put it, a company truly leveraging AI understands that the rapidly changing landscape means people have an obligation to keep learning - and only then can the organization move toward true transformation.

The three capability clusters your workforce needs

Getting to organizational transformation requires building across three interconnected skill domains simultaneously. Focusing on just one is where most programs fall short.

• ‍Data skills form the foundation.

Before anyone can use AI effectively, they need to read, question, and communicate with data. This isn't just about tools like Power BI or Excel - it's about developing the judgment to spot malinformation and critically assess what the numbers actually say. As Stins noted, working with AI means getting high-quality-looking output more easily, but people often forget to question it. Data literacy is what closes that gap.

‍

Training modules in this cluster include: Data Literacy Essentials, Problem Solving, Storytelling with Data, Excel with Copilot (levels 1–5), Power BI (levels 1–4), Power Automate Essentials.

‍‍

• ‍AI skills build on that foundation. 

Once people understand data, they can engage with AI tools deliberately rather than experimentally. This cluster spans the full range from foundational literacy through to building agents - but the connective tissue is always judgment alongside capability.

‍

"Developing the capability to critically question the outputs of an AI model is equally important,  not just the focus on the AI tool itself." 

Joppe Stins

‍

Training modules in this cluster include: AI Essentials, AI Literacy & Ethics, Prompt Engineering, Building AI Agents with Copilot Studio, Building AI Assistants, Taking the Lead with AI.

‍

• ‍Business skills are what most AI training programs omit entirely. 

Technical fluency and AI literacy only produce impact when people can translate them into decisions, priorities, and change. This cluster covers the skills that let individuals move from using AI to leading with it - and from redesigned work to organizational transformation.

‍

Training modules in this cluster include: Strategy & Prioritization, Critical Decision-Making, Project Management Essentials and Advanced, Leading Change Processes.

‍

The design principle across all three clusters: don't make AI adoption the goal itself. As Stins put it, have it supported by a business strategy that's underlying - design around where AI will actually make a business outcome more efficient.

Three real-world use cases from the field

In practice, organizations tend to approach the Data & AI training through one of three doors.

‍

Embedding AI into existing talent programs. 

‍

"Organizations don't always want a totally new academy - they just want to integrate AI into what already exists. So we had an Excel program; now we have Excel with Copilot incorporated. The outcome is that AI becomes part of a capability rather than a separate initiative." - Joppe Stins

‍

In practice: leadership programs gain AI-powered decision-making modules, graduate programs fold in AI productivity, traineeships add data-driven problem solving.

‍

Building analysts and consultants who influence decisions, not just produce reports.

‍

The pain point is familiar to most L&D leaders: talented analysts already work with data but struggle to move from reporting to influencing. The fix isn't a standalone Power BI session - it's pairing tools with critical decision-making, data literacy essentials, and storytelling with data. The outcome moves teams from reporting information to driving decisions.

‍

Organization-wide AI upskilling, anchored in policy. 

‍

For companies aiming at scale, Stins cuts against the instinct to lead with training: it's not just learning. He strongly suggests creating a clear policy first on what AI is and what the organization can and cannot do with it - then supporting that with a learning program, not the other way around.

‍

A fourth pattern surfaced in live discussion: infusing AI into existing manager training. Baltus made the case for why middle managers are the lever that determines whether transformation sticks. They sit between the strategy the leadership team sets and the individual experiments happening in their teams - making them the link that enables or blocks adoption at scale. By intertwining AI into manager programs rather than separating it, it becomes a red thread rather than a standalone topic.

How to sustain momentum past the initial training

The hardest part isn't the first session - it's month four, when initial energy fades. Stins drew on a past engagement at Philips:

‍

"After the session I facilitated, they created a group of individuals in the organization that kept the Data Academy up and running internally. So I always suggest: create a peer group internally that keeps these topics top of mind. The obligation to do continuous learning doesn't stop after the formal training ends."

‍

His second recommendation: at the end of every AI Fundamentals and AI Literacy session, he links participants to relevant podcasts and ongoing sources. Not to extend the program, but to keep people engaged with the pace of change after it ends.

What this means for your next learning roadmap

The data is consistent across three independent sources: tool deployment is high, business impact is low, and the gap is capability - not access. The four-stage model gives L&D and HR leaders a sequence to test their current program against, rather than a checklist of tools to buy.

‍

If your organization is somewhere between "rolled out Copilot licenses" and "redesigned how teams actually work," that's the gap worth diagnosing before adding another tool to the stack.

‍

Get a complimentary Data & AI Capability Scan. 

In a short session, Lepaya will help you assess your current maturity stage, identify priority capability gaps, determine which employee groups to prioritize first, and design a practical learning roadmap. Book your scan →

‍