How Novo Nordisk speeds up drug development with smarter AI

When I stepped into the role of Digital Transformation Officer (DTO) at Novo Nordisk three years ago, I had a clear mandate to transform our business by harnessing emerging technologies — without letting costs explode.

Our guiding principle is simple: “patients are waiting.” We use data and technologies to create groundbreaking science and to get it to people as quickly as possible. Every day we shave off the timeline is a day closer to helping someone manage diabetes, obesity, or a rare blood disease.

AI has massive potential to accelerate that mission. But in a hyper-regulated environment, speed alone isn't enough; we have to be certain our agents are operating exactly as designed.

Finding the “heat loss” in drug development

In the world of pharma, our work is incredibly complex and often manual. Developing new drugs is a tightly governed, decade-long journey where “time-on-patent” dictates a company's ability to reinvest in future R&D.

Having spent 30 years in this industry, I’ve seen first-hand how even the slightest hiccup in trial operations can delay the launch of innovative medicines to patients and compromise competitive advantage.

One of the fundamental challenges is fragmentation. Most people involved are domain experts focused on a microscopic fraction of the lifecycle and their operational data is not connected. This creates 'heat loss' — invisible silos where processes stall, handoffs fail, and data sits idle which slows down our speed to market. We needed a way to see the invisible.

Pioneering new technologies

To move faster than the market, we invested in high-impact, cutting-edge technologies early – whether it’s adopting technologies like Large Language Models, RAG, agents, or automation. We applied that same mindset by injecting Process Intelligence into the core of our business: drug development. (You can read our full story here). This was an industry first.

We started by targeting protocol deviations and issue management and then quickly moved to the clinical trial application process —the regulatory submission and review procedure used to obtain legal authorization to test an investigational drug on human subjects. It acts as a vital "green light" to transition a drug from preclinical development to clinical trials.

And Celonis delivered. Creating a digital twin of our end-to-end processes — enriched with our unique business context — the platform gave us critical insights to understand exactly where to intervene.

I call it the “digital brain” behind our operations, coordinating optimization across all layers of expertise. We can align every function to move at the same pace, ensuring that fixing one process area doesn’t just push the bottleneck further down the line.

The key to becoming an Agentic Enterprise

Our goal is to become an Agentic Enterprise, with agents driving some of our most complex workflows. However, many organizations make the mistake of deploying AI when they don’t fully understand how their business runs. They follow the hype — but don’t get any value out of it. My take:

With Celonis, we give our agents the context they need to succeed and operate at scale. We currently deploy over 100 AI agents in clinical development — ranging from off-the-shelf to customized solutions that we either built ourselves or we co-created with the Celonis team.

The majority of our agents run on the context layer provided by Celonis. We monitor their performance in real-time, allowing us to fix the agent or the underlying process the moment performance dips. And that’s not all.

Driving enterprise modernization on our terms

For too long, organizations (including ourselves) have taken a “technology-first” approach to enterprise modernization — forcing the business to adapt to new technologies. At Novo Nordisk, we’ve flipped the script. We can now isolate whether a business problem is caused by people, process, data or technology. If the process is broken, we fix the workflow before we ever touch the tech stack. If the tech is underperforming, we see the warning signs months earlier and solve issues before they become systemic.And we constantly need to ensure our people are trained on current and new processes and systems.

The proof: When we launched our new electronic data capture system, Celonis gave us the hard data to know exactly whether it worked or not. But it also helped us dig deeper — layer by layer — so we could start asking completely different questions about how our business runs.

Because a year is a lifetime

To understand the gravity of our transformation, you need to look at the industry standard. The time it takes to move a new drug from candidate selection to launch is roughly 12 years. This timeline hasn’t budged in a decade — despite massive tech leaps.

In this industry, saving a year or even two is a massive competitive advantage. But more importantly, it represents far earlier access for patients who depend on our science.

All this also has an impact on our culture. People in clinical development are striving for excellence. No one comes to work wanting to be ‘good enough’. We’ve already seen leadership teams scrap long-held assumptions and pivot strategic priorities the moment they saw the real friction in Celonis.

In short: Celonis has become a non-negotiable platform — ensuring we deliver on our promise that patients won’t have to wait any longer than necessary.

Stephanie Bova is R&D Digital Transformation Officer at Novo Nordisk. In her current role, she is re-engineering the speed at which science reaches patients and leading a massive modernization effort built on a single truth: Enterprise AI is only as powerful as the context behind it.