We’re at an exciting moment in life sciences, according to Alan Gallagher, life sciences market unit sales director at Expleo Ireland. Artificial intelligence (AI) is showing real potential to revolutionise human health not just in how we develop and manufacture therapies, but in how we might one day prevent – or even cure disease – he said.
However, the technology needs to be deployed within a complex regulatory framework – and this has slowed pharma companies’ ability to grasp the AI nettle.
“Across the sector, there’s growing momentum to move beyond experiments and start delivering AI-driven impact at scale,” Gallagher said. “The tools are here. The ambition is real. But there’s a structural issue that needs attention: the regulatory playbook hasn’t caught up.”
In practical terms, this means pharmaceutical companies face uncertainty in validating AI-powered tools, potentially slowing down the adoption of innovative solutions and impacting the speed at which new therapies can reach patients.
“Key guidelines like 21 CFR Part 11 in the US and Annex 11 in Europe were written long before AI – and certainly before generative AI (GenAI). They were built around static, deterministic systems where validation followed a defined, unchanging path. But with GenAI, we’re dealing with something entirely different: systems that learn from diverse datasets and evolve autonomously. The core question we now face is: how do you validate a system that is, by design, constantly changing?”
This fundamental difference necessitates a re-evaluation of traditional validation methodologies and a move towards more adaptive and risk-based approaches that can accommodate the evolving nature of AI systems.
The core question we now face is: how do you validate a system that is, by design, constantly changing?
“At Expleo, we’re helping the world’s top life sciences companies navigate exactly this challenge. We don’t deploy technology for technology’s sake, we take an engineering-led, outcomes-focused approach to ensure that digital transformation is safe, structured, and compliant, without losing the speed and innovation that AI can deliver.”
Despite these challenges, Gallagher sees tremendous potential for AI in specific areas of pharmaceutical operations. Quality control is one area where there is an immediate impact, he said.
“AI can analyse thousands of data points from production lines to identify defects with greater accuracy than human inspectors, while maintaining full compliance with quality standards.”
Expleo offers more than 50 AI applications across industries, meaning it is able to bring techniques honed in, for example, automotive or aerospace manufacturing to life sciences.
“Everybody is talking about AI, predictive analytics and big data, but all of that is predicated on having good data. The real strategy is in acquiring and contextualising data,” he said.
“If you take a manufacturing facility, they have reams of data coming off the machines, so the skill is finding it and contextualising it so that you can send it to the AI. Then you have predictive maintenance: if you are looking at the machines and OEE – overall equipment effectiveness – you want the machines up and running,” he said.
However, despite integrating learnings from other sectors, Gallagher said that the specificities of life sciences remained.
“The difference between life sciences is that it is very heavily regulated,” he said.
Consequently, Expleo’s approach, Gallagher said, focuses on understanding both the intricacies of AI and the nuances of life sciences regulations, allowing them to guide companies towards compliant and impactful implementations.
This requires a new mindset, one where regulatory considerations are integrated from the outset of AI development, and where validation strategies are adaptable to the evolving nature of these systems.
“There’s also a broader shift underway. Feedback from across the industry tells us that many AI case studies stall at ‘compliance readiness’ but never reach full GxP (goods, variables and practices regulation) implementation. That’s not good enough, especially when patient outcomes are at stake. We believe this is a pivotal time to reframe how we define, prioritise, and regulate digital innovation in pharma.”