Why AI Impact Starts with Managers

Lindsay Pattison — Chief People Officer, WPP. Lindsay leads people strategy across WPP's 108,000-person global marketing services organization, overseeing the company's move from early AI experimentation through to enterprise-scale AI delivery — including building an internal LLM that analyzed capacity unlock across 6,000 roles.

Paula Landmann — Chief Talent and Development Officer, Novartis. Paula is driving AI adoption across Novartis, where nearly 40,000 employees are now active Copilot users and AI coaching has been piloted and is expanding to 5,000 managers. She leads the company's change management approach, including a network of over 1,000 change agents and executive-level role modeling to accelerate adoption.

Hein Knaapen — Former CHRO, ING. Hein brings a practitioner's perspective on frontline management, drawing on 40 years of organizational experience to argue that the skills of the manager are the single most overlooked driver of company performance.

Das Rush — VP of Marketing, Valence. Moderator.

WPP has 108,000 employees and a 23% AI capacity unlock mapped at the role level. Novartis has 40,000 active Copilot users and managers who say AI coaching is helping them lead their teams more effectively. Both companies are past the experimentation phase — and both point to the same conclusion: the highest-leverage place to invest AI capability is in the manager. In this panel, Lindsay Pattison (Chief People Officer, WPP), Paula Landmann (Chief Talent and Development Officer, Novartis), and Hein Knaapen (former CHRO, ING) join Valence COO Das to share what it actually took to get from excitement to enterprise adoption, what the data shows about where AI coaching matters most, and why the question is no longer what AI can do — but what it should do.

Key Takeaways

• AI adoption moves in stages — and most organizations are still in the middle ones. WPP describes a clear three-stage arc: AI-assisted work, AI-enabled workflows, and now AI-delivered outputs. Novartis moved from curated tool rollouts and immersion weeks to viral adoption driven by visible use cases and executive role modeling. Both companies emphasize that getting from experimentation to habitual use requires deliberate change management investment, not just access to tools.
• Managers are the highest-leverage population for AI coaching — and the most underserved. Across WPP's thousands of Nadia users, 55% are managers and 32% are senior managers. Managers are overwhelmed, sandwiched between ambitious junior colleagues and senior leaders, and the most diverse in what they need help with. The manager's skills are, in Hein's framing, the single most overlooked driver of company performance — and AI coaching is one of the few tools that can reach them at scale.
• Role modeling by senior leaders dramatically accelerates adoption. When Novartis executive committee members shared videos about how they personally use Copilot and Nadia for goal-setting and preparation, usage across the organization spiked immediately. Starting AI adoption at the top — and making that visible — is one of the most effective change management levers available.
• Specificity is what separates real AI strategy from vague aspiration. WPP built an internal LLM to analyze capacity unlock across 6,000 roles, arriving at a 23% average capacity unlock — ranging from 60% for payroll roles to under 1% for a clapboard operator on a film shoot. Novartis employees self-report saving four hours per week. Without role-level specificity, AI transformation talk remains noise. With it, it becomes a workforce planning tool.
• At a mentoring retreat, participants said Nadia was often the most effective coach in the room. Novartis recently ran an executive leadership development retreat where participants were coached by an executive coach, a business leader, and Nadia. Participant feedback consistently noted Nadia as the most effective of the three — a signal that AI coaching is not just a supplement to human development programs but can be the standout element within them.
• The right question is not what AI can do, but what it should do. As AI capabilities continue to expand, the companies that will navigate the transition most successfully are those anchoring decisions to their values — asking not "can AI do this?" but "should AI do this, given who we are and what we stand for?" The same applies to measuring success: organizations that tie AI investment to defined business performance metrics will separate real impact from the next fad.

Questions This Session Answers

What does enterprise AI adoption actually look like in practice — and what stages do organizations go through?

Both WPP and Novartis describe a multi-stage adoption journey. WPP moved from initial excitement and experimentation, to use-case-specific enablement by function, to AI-delivered outputs embedded in core workflows. Novartis began by intentionally seeding tools with heavy users, building a network of over 1,000 change agents, then investing heavily in executive onboarding and persona-based training. Both companies found that going from initial awareness to habitual daily use required deliberate investment in change management — not just tool access — and that viral adoption only kicked in once people could see visible, concrete examples from peers and leaders.

Why do managers benefit most from AI coaching — and what are they using it for?

Managers are simultaneously the most overwhelmed population in most organizations and the most critical lever for company performance. At WPP, 55% of all AI coaching usage comes from managers, with 32% from senior managers — and their usage is the most diverse, covering everything from navigating difficult conversations to career development to managing upward. Managers are caught between ambitious, tech-savvy junior employees and senior leaders who have already established themselves, making them uniquely hungry for in-the-moment support that they often cannot get from formal programs or their own leadership chain.

How does role modeling by senior leaders accelerate AI adoption?

Novartis found that one of the most effective adoption accelerants was senior leaders publicly demonstrating their own AI use. When executive committee members shared short videos describing how they personally used Copilot to draft objectives and Nadia for coaching preparation, tool usage across the organization spiked immediately. The message it sent — that AI is not cheating, it is enabling — was more powerful coming from visible leaders than from any training program or communications campaign. Starting adoption efforts with the executive committee and making that visible to the broader organization is one of the most consistently effective change management moves available.

How should organizations measure the real impact of AI on workforce productivity?

Vague claims about AI productivity — "95% of marketing can be done by AI" — are not useful for workforce planning. WPP built an internal LLM to analyze 6,000 roles and arrived at a 23% average capacity unlock, ranging from 60% in payroll to under 1% for some on-set roles in production. Novartis employees self-report saving four hours per week with current tools — up from two hours a year prior. Both companies are now focused on directing that unlocked capacity toward higher-value activities, and on moving from self-reported productivity gains to measuring whether AI is helping managers lead their teams more effectively.

How does AI coaching compare to human coaching in a leadership development setting?

At a Novartis executive leadership mentoring retreat, participants were coached by three sources: an external executive coach, a business leader, and Nadia. Participant feedback consistently identified Nadia as the most effective coach of the three. Beyond that single data point, the broader finding is that AI coaching creates a psychologically safe space for managers to ask questions they might feel are too basic or too revealing to raise with a human colleague — and that the combination of availability, non-judgment, and personalization makes it a qualitatively different development experience than what human coaching programs can provide at scale.

What separates organizations that capture real AI value from those that don't?

Hein's framing is direct: organizations that fail to tie AI investment to well-defined business performance metrics will have experienced another fad — like the metaverse. The companies that will have genuinely benefited are those that moved beyond discovery and linked AI capability to measurable outcomes. Lindsay adds a values dimension: the question is not what AI can do — almost everything can be augmented by AI in some way — but what it should do, given the values and responsibilities of the business. Both dimensions — business metric linkage and values-anchored decision-making — are what separate durable transformation from temporary enthusiasm.

Full Session Transcript

Where are most organizations in their AI journey right now — and where are WPP and Novartis?

Das: Last 12 months — where are we right now? Where are most organizations, or the organizations you've been working with?

Lindsay: For the last 12 months, we've had three stages of our journey at WPP. WPP has about 108,000 colleagues around the world working in marketing services. The first stage was excitement, optimism, experimentation — all of which is good, because our industry is actually super optimistic about AI. We became very focused on mass adoption. We would shut offices for a whole day and really push AI training. So the first stage was excitement and experimentation — AI assisting people.

The second stage was moving it to more habitual use, because it's fine to experiment and be excited, but we need to be very specific about use cases by function. Creating AI-enabled workflows for our client work was one piece. Then thinking from a business function perspective — people, legal, finance — how do we think more specifically about how those functional roles can be augmented by AI? So we moved from assisted, to enabled. And now it's really about delivery by AI. Fundamentally, pieces of work — even org charts — being delivered by AI and baked into our absolute offering. Some people talk about a wave of AI, but I think it's a tsunami about to hit us all, and we are moving into the delivery phase.

Paula: At the start, we first understood which tools we wanted to put in everyone's hands. We created our own internal ChatGPT. We agreed Copilot would be one of our core tools. We also assessed which AI coach we wanted to proceed with. And then there were more specialized tools agreed for specific groups with particular needs — from GitHub to others that many of you probably use.

Our journey was then understanding how to get to real usage. We started by very intentionally giving licenses to people who already used Word documents or PowerPoint presentations heavily. Those people became our change agents very quickly — and that evolved into a network of over 1,000 change agents across the company. We also very intentionally onboarded our executive committee members and top leaders early, and started using AI in critical meetings — sessions around it, persona-based. It was really a heavy investment in change management from the start, with support from an external partner, because we believed that was what people needed to realize the benefits and actually start using the tools.

We ran a lot of Gen AI immersion weeks. From the start, people were intrigued and trying to understand it. And we slowly saw a huge shift to real adoption in the day to day. We track usage — we could see the number of hours people use Copilot per week climb until we became one of the companies with the heaviest active user counts. As people saw the benefits and the business cases, it became part of how we work. In development, in research, AI is now part of how people solve day-to-day tasks. And it became a bit viral. We're now at a point where there are great use cases in every single area of the company. It's a big shift.

Hein: Paula and Lindsay are probably a little ahead of the curve — and I can understand how their type of work drives that. What I observe more broadly is that excitement about new tools is great, because losing curiosity is downward. But excitement doesn't always make it easy to keep company performance as a North Star. You don't think your way into new acting; you act your way into new thinking. But how are you sure — or how are you evolving — to a point where you can clearly say: here are the parts of our processes where it works and has real value, and here are the parts where it's only nice to have?

Das: What did you hold as a North Star through your initiatives?

Lindsay: The North Star is performance. Paula and I both work in very competitive categories. More simplistically: we'll come to managers, who are the majority of our workforce. We need a competitive advantage. Getting ahead in adopting and using AI is going to help us win and have the business succeed. Simple as that.

Why did Novartis embed AI coaching specifically within its manager initiative — and what has the impact been?

Das: Paula, you've embedded Nadia within your Align initiative, which is explicitly an initiative for managers. Can you talk about why that initiative, and why an AI coach within it?

Paula: Align and Nadia are actually separate tools, but we have both. Align is a tool for team effectiveness — a super simple diagnostic that allows teams to rate themselves on habits shared by high-performing teams, and then triggers the right conversations in the areas the team needs. We're using it across the enterprise for all sorts of team conversations. Both are Valence tools, and very helpful for us.

Coach Nadia, which we originally piloted with a few hundred people and are now expanding to 5,000 at Novartis, is really a game changer for us. We've had experience with it for almost a year across the organization. Our North Star here is individual development. In essence, it helps any person get support in the moment they need it. They don't have to wait for their next coaching conversation. It's accessible — it creates a safe space where people don't have to worry about being judged for what they ask.

The feedback has been excellent. We've surveyed managers and tracked their development over time. We recently embedded Nadia in a mentoring retreat for executive leadership minus one — participants were coached by an executive coach, Nadia, and a business leader. The three coaching sources coached individuals across the retreat. And the feedback from participants was that Nadia was many times the most effective coach.

People who go through it see the benefit and then want their teams to have it. They talk about it. You see the effect it creates. It can be as much an opportunity for people to stop, reflect, learn, and get advice, as it is a tool that nudges you: "Have you thought about that today?" — sending you a prompt. It has been one of the most impactful tools we are currently using.

Lindsay: Building on that — as we think about adoption at scale, it's really FOMO. Fear of missing out. And I think it was very clever to start with the executive committee using the tools, because for everyone else, you need to understand that AI isn't cheating. AI is enabling, assisting, helping you. And everyone in a high-performing organization wants to be better at their job. Whether that's Coach Nadia helping you think through how to have challenging conversations or develop your career, or Copilot helping you curate documents more efficiently — it's a way of being better at your job. Who doesn't want that?

Paula: And Lindsay, just on that point — we very intentionally put out videos of our executive committee members talking about where they use the tools. One of our exec-co members shared that he used Copilot to create his own objectives and got hints from Nadia along the way. Usage of the tools after that video went out just went up drastically. It really shows how role modeling plays a role in day-to-day adoption.

Why do managers specifically represent the biggest opportunity for AI coaching investment?

Hein: I want to bring in a perspective I'm a bit obsessed with: the role of the middle manager. 85% of our people report to frontline managers. There's a great book from last year by Bob Sutton of Stanford called "The Friction Project." He says leadership should be the guardian of its people's time — protecting them to spend that time being relevant for the customer. But in actual fact, we often become robbers of that time, burdening people with pet projects. And our frontline managers — dignified, respectable people, often not much more advanced than the people they lead — are frequently left alone to grapple with the practical reality of driving performance. What I've experienced over 40 years is that the most overlooked, single most important driver of company performance is the skills of the manager.

From that perspective, I look at Nadia and AI with real interest. I'm an apprentice and a starter here. But I can see how powerful it is, because it creates a psychological safety for the manager to ask questions they might be afraid are too basic to ask other people. And that builds their skills — and as a result, their confidence to steer performance.

Lindsay: Managers are overwhelmed. We looked at our Nadia usage across WPP, where thousands of people use the tool. We segmented by seniority: 55% of usage was by managers, 32% by senior managers, and a small minority by junior colleagues. And the range of topics managers bring to Nadia was the most diverse across any cohort, because they are grappling with the most — trying to move up the corporate ladder, managing super tech-savvy, ambitious Gen Z colleagues below them, while Gen X leaders above them have earned their position and aren't moving. There is a big burden on managers, and it's our job to support them.

Paula: And when we look at the topics managers bring to Nadia in aggregate — without any individualized data — we get really useful insights into what we can do to support managers better. We can see that managers are really trying to understand how to influence at scale. That signals what capabilities we need to build. And in our Gen AI weeks, the sessions with the highest uptake are always the persona-based ones focused on tools and tips specifically for managers. Fifteen thousand managers across the company joined those sessions in total. Because they're overwhelmed, and they want pragmatic, practical help. Tools that take a lot of time to learn create friction they don't have time for. The value is in helping them get what they need in the flow of work.

Hein: And those are relatively micro interventions. You don't need a three-day course. You can take 10 minutes. That is wonderful.

Paula: Exactly — in the flow of work. And what we hear from managers is that having tools at their fingertips that increase their productivity and unlock hours for them to be more innovative is genuinely helpful. I was talking to a manager recently who said: "I used to have to think about where to find a tool to help me have a difficult conversation or host a development conversation. Now I have not only Nadia who supports me but also Copilot. All the prompts I use are automated and ready when I need them." That's what makes the difference.

Lindsay: The core reason we use Nadia specifically as a tool for managers is democratization of coaching. Everyone talks about the value of the safe space, the testing, the role-play. The reason senior people often reach the top is simply experience — they've had the experiences that build judgment. Nadia allows you to shortcut that, to role-play experiences before they happen. It's democratizing something that has historically only been available to very senior people. That's why we love it: speed and democratization.

How do you move from vague AI transformation language to specific, measurable business impact?

Das: What does it take to go from "AI is going to transform your workforce" to "this is how we did it, and this is the measurable impact"?

Lindsay: We paid a lot of attention to strategic workforce planning — thinking about the shape of the organization and trying to be very specific about what AI will unlock. We created our own LLM and put in 6,000 different roles, then broke down every role to ask: based on AI tools available now, what would be the capacity unlock? Otherwise people talk in very vague terms — some will say 95% of marketing can be done by OpenAI. I would say that's nonsense. But through that detailed work, we arrived at a 23% average capacity unlock from AI across our workforce. And that varies wildly — from around 60% for someone in payroll to below 1% for a clapboard operator on a shoot, because we still need a human for that.

Being specific has helped us. We've since run a follow-up study showing roughly 20 to 21%. That's great knowledge to have. But now we need to move to directing and guiding targets against that capacity unlock — and thinking about what we actually do with the time saved. What are the high-value activities we're going to enable our colleagues to do now that we've taken away some of the drudgery? We have the data. We now need to be more prescriptive about gaining that unlock back, turning it into a commercial model, and then asking: what can we now do that we couldn't do before? What's uniquely human? What's creative?

Paula: For us, impact looks different across different parts of the business. Operations and technical use AI quite differently than sales or research and development. What we get reported on average is that people say they save at least four hours a week with the tools we provide — up from two hours a year ago. We want to see that trend continue. And to Lindsay's point, we're also focusing on helping people understand that the time saved should go toward other work, toward innovation. When we surveyed our nearly 40,000 Copilot users, 89% said they feel more productive, and 76% said they feel more creative. That creativity component matters to us — it's part of our culture aspiration. We want people to feel inspired at work. And the shift we're beginning to hear in surveys — from "AI saved me time" to "AI helped me lead my team more effectively" — that is the journey we're on.

What are the early signals pointing toward where AI in the workplace is headed in the next 12 months?

Das: What are some of the things we're observing now that hint at where we'll be 12 months from now — and where do you hope to be?

Paula: Twelve months from now, I think people will stop calling it AI and just call it the work. The best companies won't treat AI as a separate workstream. They'll embed it into everything — onboarding, performance coaching, leadership development, day-to-day business operations in every function. And the companies that get it right will be the ones where managers don't just use the tools but genuinely feel — and help everyone around them feel — that AI is part of how they work in the flow of every day.

Lindsay: What companies need to really consider — not to be a downer — is not what can be done by AI, because almost everything can be augmented by AI in some way, but what should be done by AI. The larger question is the ethics of AI: thinking about the values of the company you work for, and thinking responsibly about how you use AI in the goods and services you offer. Tools like Nadia are genuinely wonderful and effective, and we really hope they'll help people be better managers, better leaders, and create a happier, more productive, more effective workforce. But there are other aspects of AI that we should be thoughtful about — holding to the values we've always had as a business, and asking not what can it do, but what should it do for our business.

Hein: I'm not sure 12 months is quite long enough a horizon. In the slightly longer run, the companies that will have done better are those that have tightly linked AI to business value metrics. We're in discovery right now, and that's a necessary and useful step. But in the longer run, if AI investment isn't linked to well-defined company performance metrics, it will have been another fad. That's what I'm a bit afraid of.

Lindsay: It'll be another metaverse if we remember that. I think this one is here to stay.

Das: I think so too. We've traced a nice arc here — maybe along the Gartner hype cycle for where we are. We'll see where we are 12 months from now. Lindsay, Paula, Hein — thank you so much for joining.