HR went first: How Merck made its people agent the blueprint for enterprise AI

At Merck Group, HR settled the debate over who owned AI by going first.

Alexis Saussinan, Merck’s Group Head of People Insights & Effectiveness, tells UNLEASH that rather than asking permission, the HR function “rallied behind the fact that it’s a people topic” and created a strategic plan.

At the heart of the strategy is V.E.R.A., Merck’s enterprise agentic People agent. Already live enterprise-wide, it has become the blueprint for how Merck – a pharma, life sciences and semiconductor business with 63,000 employees – builds AI agents across the enterprise.

Why HR was ready to lead AI transformation at scale

While most areas at Merck were still in pilots with AI, HR’s developed a strategy with clear outcomes from the start.

The crux of the strategy is “AI only creates value when it changes how people work end-to-end, not when it sits as isolated tools.” This means that AI has “measurable business impact,” rather than purely driving “isolated productivity wins,” notes Saussinan.

The plan was important, but alone it wouldn’t have convinced the rest of the organization that HR should lead the way on AI at scale. What set HR apart was its data readiness.

Saussinan argues HR is one of the most data-mature function at Merck, thanks to several years of investment in HR data governance, quality and privacy. That groundwork enabled HR to be chosen to serve as Merck’s blueprint.

That strategic positioning and ownership is the company’s “differentiator,” in Saussinan’s view. “I don’t think we are the only one in the world, but I can count very very few.”

V.E.R.A.: Merck’s AI strategy in practice

Those strategic and data foundations were non-negotiable when building HR’s agent, V.E.R.A.

However, three core decisions shaped the plan:

• Anchoring in business outcomes
• Redesigning work, not tasks
• Building enterprise-grade agents, not experiments

Saussinan is clear: “We don’t have endless pockets; we really want to be super focused on the business and people outcome we’re going to get.” Saussinan is clear: “We don’t have endless pockets; we really want to be super focused on the business and people outcome we’re going to get.”

This requires starting not with the AI tool but focusing on the problem that needs solving.

For V.E.R.A., that problem was clear. HR processes were too complex, fragmented and costly – this meant managers were spending more time navigating the processes than leading people.

HR was also overloaded with repetitive queries. “We had a lot of costs in shared services answering very basic questions,” Saussinan adds.

Rather than “sprinkling AI features across HR”, it used V.E.R.A. to reinvent and orchestrate HR workflows end-to-end. “V.E.R.A. will be the only employee-facing People agent,” notes Saussinan.

Several months since roll-out, the impact that is being delivered is a significant reduction in HR operational workload, improved consistency and quality of HR decisions, and measurable time freed up for HR and managers to focus on value-creating work.

Hard return on investment was also central to the business case. Savings with V.E.R.A. is mainly coming from cutting spending on consulting services and IT tools, rather than headcount.

As Saussinan puts it, his argument to the board was “if I need to ask you for €100, I am telling you that already in year one, you’re getting sizeable euro back in hard return, and over three to five years you will break even.”

Lessons from 18 months in: What Merck would do differently

HR being the AI pioneer at Merck came with “all the good and the bad.” If Merck were starting afresh, Saussinan would change three things to move faster.

The first is technology. Saussinan wishes the agentic AI market tech stack had been more mature.

Because HR went first, it had to solve several very difficult problems that belonged to the whole company, such as employee authentication. “You talk to V.E.R.A., V.E.R.A. knows you. The access rights are managed properly” – they’re safe and secure. “The truth is it took a few months to crack,” and the partnership with our AI and IT teams, was critical and there are many more examples like this which were enterprise, not HR specific, challenges, Saussinan notes.

The vendor market has also not caught up; “I don’t see the providers having really fully understood how to position themselves to truly help companies as we are all learning together.” This is particularly the case around integration.

The second is change management. Saussinan wishes there had been more external examples of HR functions rallying behind AI transformation. The solution is to show rather than tell, letting people see the impact of AI first-hand before scaling.

The third is design. With hindsight, Merck would have been clearer on the operating model upfront: who owns what, how decisions are made, and how freed-up capacity is redeployed.

Even so, Saussinan is pragmatic. “I wish we could have been faster and earlier, but I don’t know if it’s realistic since everyone was and is still learning,” he says.

His advice to HR leaders early in their AI journey comes down to one choice.

“Decide to redesign the work, not the tools,” he says. “If you redesign the work, AI becomes a lever for transformation. If you only deploy tools, AI becomes a hobby.”“Decide to redesign the work, not the tools,” he says. “If you redesign the work, AI becomes a lever for transformation. If you only deploy tools, AI becomes a hobby.”

What’s next: AI beyond HR at Merck

Proving AI in its own function, HR now wants to shape how the rest of Merck reinvents its organization and workforce itself with AI.

To be credible AI leaders, HR had to role model – “now I want to make sure that the entire enterprise AI workforce transformation is being driven with intention,” Saussinan states.

This means every AI initiative across Merck now must pass four tests:

• A clear business problem with measurable value
• End-to-end workflow redesign, not isolated automation
• Enterprise readiness across data, governance, adoption and scalability
• Putting people at the center and transparency about the impact on them

Saussinan is clear that “if an agent cannot deliver impact at scale, it should not move forward.”

Shortcuts are not how Merck operates – rather than hiring and firing at the whim of the AI market, “we stay true to our culture, which means we will bring our people along the journey.”

Aligned with this, an open question that Saussinan and Merck are exploring is workforce and agent integration. “It would be a lie to say we’ve cracked it,” Saussinan admits. “We are on it.” It starts with defining the lifecycle of an agent.

“How do you look for, hire, interview, onboard, contract and retire an agent? Nobody knows that today.” Only once that is answered can HR have a holistic view of how to move to the next stage: integrating the human and agentic workforce.