Where AI adds value in device development

Across the drug delivery industry, artificial intelligence (AI) is moving from experimentation into everyday work. As the technology becomes more accessible, however, access alone is becoming less of a differentiator. The harder question is knowing where AI can genuinely improve the way work gets done and delivers measurable business value.

For companies developing and manufacturing drug delivery devices, that question comes with an additional layer. New capabilities must work alongside engineering judgment, established development processes, quality requirements, and the controls expected of a regulated industry. 

SHL Medical has been exploring where these opportunities meet. One recent step was the company’s AI hackathon, where six cross-functional teams came together in a focused two-day innovation sprint to apply AI to real business challenges. 

Alex Stoeckli, Head of AI and Data Competence Center at SHL Medical, discusses what the exercise revealed, where AI could add value across device development and manufacturing, and why triaging AI-driven business cases matters even more than the technology itself.  

How is SHL Medical approaching AI? 

Alex Stoeckli: We treat AI as an enterprise capability, not simply as a collection of tools. A tool can make a task faster. A capability can change how an organization explores ideas, analyzes information, automates workflows, and gets from a question to a first workable direction. Beyond individual productivity gains, we see AI creating opportunities to improve how work gets done across functions and business processes. The long-term value is not only in faster execution, but also in enabling new ways of collaborating, making decisions, and operating at scale. Ultimately, we want that value to reach the pharma companies for whom we develop and manufacture devices.  

The hackathon gave colleagues a dedicated environment to see what that could mean in practice. We brought together employees from different functions with different levels of AI experience and gave them two days to take an idea and develop working AI-driven solutions, with AI itself supporting the build.

That concentrated format changes how people work together. In our day-to-day roles, we naturally focus on our own responsibilities and timelines. A hackathon creates a different dynamic, with a room full of people exchanging perspectives, challenging assumptions, and building on each other’s ideas. And because AI could support much of the technical implementation, teams could spend more time focusing on the problem they were trying to solve and whether the solution made sense. Interestingly, the winning ideas in recent hackathons came from non-technical domain experts rather than implementation specialists. 

It was as much about experiencing a different way of working as it was about what each team produced at the end. As more colleagues develop an instinct for where AI fits, we get better at applying it to the problems where it can make a real difference to how we work and how we support our customers and partners.  

What did the hackathon reveal about where AI can create value? 

Alex Stoeckli: One of the clearest lessons was that the strongest ideas did not start with AI. They started with a well-defined problem, and a clear understanding of the people involved and how they already work. From there, AI could help automate part of the process or make it easier. This was often the difference between an interesting demo and something people could realistically use in their day-to-day work. 

We see several recurring patterns where AI currently can be useful. It can help people work through large amounts of information more efficiently, support repetitive analysis or documentation tasks, and shorten the path from a question to a first workable direction. In engineering work, that includes drafting code, tests, or implementation options faster. But the human-in-the-loop principle remains essential. Engineers still need to review, verify, and validate the results within the established development process. 

AI can accelerate parts of the work, but faster output is not the same as a validated answer. The expertise to assess what comes back, understand its limitations, and decide what happens next remains essential.   

Why does this matter in device development and manufacturing? 

Alex Stoeckli: An autoinjector or a pen injector has to perform consistently across potentially millions of units. Achieving that requires a development and manufacturing system capable of controlling many variables, repeatedly and at scale. 

Automation has long been central to how we manage that complexity. AI adds another layer of capability, giving us additional ways to keep the line efficient and to catch potential issues earlier. It also helps us improve the processes behind a device, not just the device itself. 

The objective is not simply to make individual tasks faster. If a capability helps us identify an issue earlier, avoid unnecessary iterations, or make a better-informed decision sooner, that can reduce uncertainty further downstream. This is where the relevance extends to our pharma partners. Improvements within these processes can ultimately help create a more efficient and predictable development path for the combination product as a whole. 

How does SHL Medical make AI accessible without losing control? 

Alex Stoeckli: Different people need very different things from AI. Someone using it to speed up an everyday task needs something very different from a specialist building a new tool. That is why we, at SHL Medical, are building AI capability at several levels. 

The majority of employees have access to approved AI tools that help with everyday work with no technical background needed. Where the need is more specialized, we develop solutions tailored to particular domains and workflows. And behind all of it, our AI and data teams provide the technical expertise, governance, security, and platforms needed to develop and operate these technologies responsibly. 

At the same time, we are also developing training materials for different levels of experience so that people understand how AI relates to their role, how to use it effectively, and where its limitations and risks lie. We realize people develop much better judgment once they have actually tried to solve a problem with these AI tools, which is one reason initiatives such as the hackathon are so valuable.

For a company operating in a regulated industry, wider access has to sit on top of clear governance. People need room to explore, but they also need to understand that additional controls, validation, security, or human review are required before a solution touches our daily work.  

Where does SHL Medical’s AI journey go next? 

Alex Stoeckli: The technology itself will keep changing. Our task is to make sure SHL Medical has the strategy, scalable platforms, and operating discipline to deploy and govern AI capabilities consistently across the organization, so that we can absorb each new wave of technology without starting over. 

In the near term, that means strengthening that foundation by increasing AI literacy and judgment across the company while continuing to explore and challenge practical applications on our processes. That means looking at how AI can support more complex workflows, rather than just individual tasks within defined governance and human oversight.  

After all, for pharma partners and eventually the patients using the devices we develop, being consistent and responsible matters more than being first to every new technology.