Across SLAS Europe 2026, the focus was on AI and automation, but the tone was notably pragmatic. While AI featured heavily across keynote sessions and vendor demonstrations, the emphasis was less on future promise and more on present-day constraints. Conversations moved quickly beyond model performance and algorithmic novelty to the realities of laboratory execution: data quality, system reliability, and the challenge of connecting digital tools with physical experiments.
That shift reflects a broader recalibration underway in drug discovery. The fundamental challenge facing pharma has not gone away. As Jonathan Wingfield, a Business Development Specialist at TTP, told DDN, “The cost of bringing new drugs to market keeps rising, while the effective window of intellectual property is shrinking. That combination makes it much harder to recover the investment made in early discovery.”
Rather than chasing ever more ambitious in silico predictions, many organizations are recognizing that the real bottlenecks sit elsewhere — in how experiments are executed, how data is generated, and how quickly teams can learn from results.
“We’re starting to come out of the AI hype cycle, at least in the life sciences,” Wingfield said. “AI can take you part of the way — particularly in early discovery — but once programs move into the clinic, the problems become much harder.”
Instead, the growing consensus at SLAS Europe 2026 was one that treats AI not as a standalone solution, but as something that depends entirely on the infrastructure beneath it. Automation, connectivity, and reliable data generation are becoming the true enablers of change.
From automation to orchestration
For years, automation in the lab was largely synonymous with throughput. Liquid handlers, robotic arms, and screening platforms were deployed to move faster and process more samples. But that narrow definition is starting to break down. Today, the conversation has shifted from how fast individual systems can run to how well entire laboratories can function as connected, coordinated environments.
“The biggest change we’re seeing is the move away from isolated, digitized labs toward fully automated, integrated environments,” Paolo Concio, Senior Director of Global Commercial for Digital Science Solutions at Thermo Fisher Scientific, told DDN. “This represents a major step change — from standalone work cells operating independently to connected systems functioning as part of a coordinated whole. It requires the ability to link instruments, digital platforms, AI tools, and robotics into a seamless, continuous loop.”
Crucially, this orchestration problem can’t be solved by a single vendor or platform. Laboratories are inherently heterogeneous environments. A single workflow may involve a dozen instruments from different suppliers, each generating its own data in its own format. “Customers are very clear about this,” Concio said. “They already have infrastructure in place. They’re not going to rip everything out and start again. The question is always: How does this new system fit into what I already have?”
That reality is driving a renewed emphasis on interoperability and open ecosystems. “Systems need to be designed to function within open ecosystems, capable of communicating and working across different platforms and technologies,” explained Concio. “That shift toward connectivity and integration is what is now clearly emerging across the field.”
Automation is no longer just about executing experiments. It’s about moving seamlessly from hypothesis to experiment to data analysis, and back again. That broader interpretation of automation is beginning to redefine what a modern discovery lab looks like, and it sets the stage for where AI can deliver its most practical value.
AI in closed-loop labs
For all the attention AI receives in drug discovery, its real impact is tightly bound to the physical laboratory. Models don’t learn in a vacuum. They learn through repetition — hypothesis, test, learn, repeat — and that cycle only works when digital intelligence is anchored to wet-lab reality.
“In discovery, you can generate huge numbers of candidates with AI,” Marco Ravot-Licheri, Head of Digital for the Life Sciences Business at Tecan, told DDN. “But even the best models still contain assumptions. You have to test those hypotheses in the lab, bring the results back, and refine them. That feedback loop is non-negotiable.”
Making that loop work requires a well-connected lab, where AI can also be effectively used in areas such as resource planning, data analytics, and workflow optimization. However, its performance drops significantly when data is fragmented or inconsistent.
“We need to bridge wet and dry lab environments, and we need the right application programming interfaces (APIs) in place — not just for instrument control, but also for analytics tools,” said Ravot-Licheri. “You need those APIs to enable the closed loop.”
A useful example of how this shift is already playing out can be seen in the evolution of lab analytics platforms. These systems were initially designed to monitor instrument performance and provide retrospective insight — essentially helping teams understand what went wrong after an experiment had already failed or deviated.
More recently, they have started to incorporate agentic AI capabilities that move them from reactive reporting to predictive intervention. “A scientist can now say, 'I’m about to start this run on this instrument,' and the system can look back across historical data, identify patterns, and flag potential risks before the run even begins,” explained Ravot-Licheri.
This shifts the role of the system from post-hoc analysis to continuous monitoring and early warning, where emerging issues can be detected and acted on before they affect experimental outcomes.
Closed loops and human-centred labs
These developments point to a more significant trend in how discovery labs are designed and operated. As orchestration improves and AI becomes embedded in day-to-day workflows, the question is no longer whether laboratories can be automated — but how that automation should be structured, and who remains in control.
Rather than converging on a single model, SLAS Europe 2026 revealed two complementary paths emerging in parallel: closed-loop laboratories, where AI systems actively guide experimental cycles end to end, and human-centric labs, where automation and intelligence enable scientists to focus on higher-order interpretation, experimental design, and decision-making.
Within that framework, the future of the lab is not a simple choice between full autonomy and human control. “It depends on the application,” Ravot-Licheri said. There are clear use cases where self-driving labs make sense — environments where execution can be largely automated and scientists focus on interpreting results and designing the next question. But there are equally important contexts where labs remain fundamentally human-centric, with AI acting as an enabling layer rather than a replacement.”
A connected foundation
Across SLAS Europe 2026, the message that emerged most clearly was that AI is no longer being viewed as a standalone driver of transformation, but as something that is only as powerful as the laboratory it sits on top of. The most advanced models and algorithms still ultimately depend on whether experiments can be run reliably, whether data is captured in a structured way, and whether results can flow back into the next cycle of design and decision-making.
Overall, the impact of AI depends heavily on how automation is used to generate high-quality, real-world data. Without that foundation, AI on its own is unlikely to fundamentally change how discovery workflows operate.
