The missing operating model: Milica Sapic on why AI adoption stalls

Most companies have already moved past the decision of whether to use AI: the licenses are bought, the pilots are running, and the tools are in employees' hands. What's still missing for most is the organizational change needed to actually see returns from any of it.

To explore this challenge, we sat down with Milica Sapic, Talent and Organisational Development Manager at Publicis Sapient, and talked about why most AI adoption stalls at a tactical level, where HR and L&D fit into the shift, and how the function can support managers and the broader workforce in actually adopting AI well.

‍

How mature are organizations with AI adoption today, and what's holding them back?

The reality looks different from company to company, but I think a lot of organizations are still answering the question of how to deliver: which tools, which workflows. Far fewer are asking what the company itself needs to look like at this stage, including what talent and what setup the business actually needs. That higher-level mapping tends to fall behind, and we're already seeing some of the consequences in the form of layoffs rather than real structural change.

Leadership plays a part here, too. A lot of organizations get stuck operating in a short-term lane, rolling out tools quickly without working through what that means for the business model behind them. Teams end up with access to AI but no context for how to use it in their day-to-day work, so instead of becoming more productive, they end up more confused.

‍

What role should HR and L&D play in this AI transformation?

HR and L&D are sitting on more relevant knowledge here than people often give us credit for. We tend to operate as the connective tissue of the organization, meaning we already understand how work gets done and who's involved in delivering it. That puts us in a strong position to partner with the business, see where the bigger shifts are heading, and help shape what that means for the workforce.

A lot of attention currently goes into bringing technology to people, but I think our role here is to bring people to the technology as well. That means building awareness and skills to use new tools, communicating strategies, and managing the resistance that naturally comes with this kind of change. Training is one piece of that, but it's far from the only one.

When it comes to adoption, middle managers are often the ones having to translate strategy into daily work, but not all are equipped to do so. Where are the biggest gaps in how they're being prepared?

If the C-suite hasn't worked out what it wants, it's very hard for middle management to act on it. That's the first gap.

The second, and probably the bigger one, is time. Managers are expected to model AI use for their teams, but they rarely get protected time to experiment and build that experience for themselves, and you can't be the sense maker for your team without it.

What helps is creating that enablement at a higher level: protected time and shared forums where managers can come together to talk through what they're learning. Most teams already have a few people who naturally step up as AI champions, more curious or further along than the rest. They tend to lead the way and create a sense of community around these conversations on their own, which gives L&D something real to plug into rather than building enablement from scratch.

Third, design enablement around managers' real work. By embedding AI into the challenges they face every day, managers develop practical skills in use cases that matter most, making learning immediately applicable and accelerating adoption.

Without those resources in place, things can still get rolled out on paper, but they rarely lead to the kind of lasting shift in how a team works that actually shows a return.

What other skills or behaviors do you think managers need to develop to lead adoption?

Protected time and clarity matter, but a few other things make the difference too:

• Get clear on what success looks like: The bar keeps moving, so managers need clarity on what their team should achieve right now, then build the moments and conversations that move toward it.
• Be transparent, not falsely positive: Pretending to be an AI expert doesn't help the team. Admitting what you don't know and saying "we're figuring this out together" builds the psychological safety that makes experimentation possible.
• Network across functions: Partnering with people in other functions in the business and learning from how their work is changing is an underrated form of stakeholder management.
• Protect your own well-being: The pace of change doesn't let up on its own, so building in space to slow down has to be intentional.

‍

What about the broader workforce? How can HR and L&D help close the AI productivity gap, and where does upskilling fit in?

I see HR and L&D playing a critical role in helping organizations navigate the future of work and stay ahead of the curve. This is not just about teaching people how to use AI tools; it is about building the capabilities, mindset, and context needed for people to create better outcomes with AI.

That starts with creating a shared understanding of what good looks like: the quality standards, business context, customer expectations, and outcomes we are trying to achieve. Without this foundation, everyone starts creating their own way of working with AI, which one transformation leader described as “silos of one”, where outputs become inconsistent and do not scale across the organization.

L&D has an important role in helping employees build not only AI literacy but also the critical thinking and judgment to evaluate, refine, and take ownership of AI-generated work. AI can amplify productivity, but only when combined with strong business understanding and clarity on how each role creates value.

At the same time, staying ahead means looking beyond our own organization by bringing in external perspectives, connecting with other companies, and understanding how the industry is evolving so we can continuously challenge and improve our approach. We also need to help employees understand how their roles are changing, what skills will matter in the future, and how they can move towards that future state together with the organization.

Ultimately, the role of HR and L&D is to connect the outside-in perspective with internal capability building, helping the organization adapt while enabling people to grow with the change.

Beyond AI literacy, what other skills are necessary for an AI upskilling program?

A few more things matter:

• Critical thinking, in practice: Seeing your work from a different angle, checking sources, and doing proper research before reaching a conclusion. Research skills get taken for granted, even though a lot of work runs on assumptions that need real data.
• Ethics and data governance: Not accepting AI output without question, and understanding where the data behind it goes and what protections exist for employees and the organization.
• The ability to detach and reflect: AI can help gather information, but it's still up to each person to form their own thinking on top of it, through habits that help people hold on to their own point of view.
• A community of peers: Giving people internal forums, or pointing them to external ones, gives them a place to work through questions together instead of feeling like they're the only ones asking them.

‍

Is there anything about AI adoption you think is being overlooked that HR and L&D leaders need to hear?

The question I keep coming back to is what organizations, and society more broadly, will look like in five or ten years. The market is moving faster than society can adjust to, and most conversations right now stay focused on profit rather than the people whose roles change or disappear along the way.

I'd like to see organizations bring in more foresight and scenario planning: thinking through what different paths actually mean, not just for the business, but for the next couple of generations. IKEA is a useful example of getting this right. Instead of cutting customer service roles, they noticed from customer feedback that people wanted design advice, not just support, and reskilled those agents into interior design consultants. That's the kind of forward-looking workforce planning that's possible when a company looks at how its business model is actually evolving, rather than defaulting to layoffs.

HR and L&D are well placed to bring that kind of thinking into the room, but it takes some courage to raise it. I'd like to see more of us willing to do that.

‍