Building Data-Driven Labs With AI

As AI tools become more widely used in life sciences research, including areas such as drug discovery, attention is shifting toward how efficiently laboratories can test and validate the growing number of computational outputs. Faster hypothesis generation is only part of the equation—how those insights translate into reliable experimental data is becoming a defining challenge.

Hal Wehrenberg, vice president of global services and digital innovation at Tecan, has spent more than two decades working across laboratory automation, from early roles as a service engineer through to leading global product and digital initiatives. His experience spans both US and European lab environments and includes work on high-throughput platforms and workflow integration, providing a practical perspective on how automation, data, and AI are converging in day-to-day laboratory operations.

Technology Networks spoke with Wehrenberg to discuss how expectations around automation are changing, where data integration still falls short, and why connecting computational and experimental workflows is becoming increasingly important.

Closing the loop between the dry lab and the wet lab

From your perspective, what are the most significant themes emerging, and how do they reflect where life science laboratories are heading over the next few years?

AI is already accelerating hypothesis generation and experimental design, but that progress is beginning to shift pressure onto the lab.

“One of the strongest themes is the growing importance of closing the loop between the dry lab and the wet lab. AI-driven approaches are rapidly increasing the speed and quality of hypothesis generation, candidate selection, and experimental design. As a result, the bottleneck is increasingly shifting from generating ideas to testing and validating them in the laboratory,” Wehrenberg said.

Rather than treating computational and experimental work as separate stages, laboratories are starting to connect them more tightly. Predictions need to move quickly into experiments, and experimental results need to feed back into models.

At the same time, AI is becoming more embedded in lab environments. “The conversation is moving beyond standalone AI tools toward systems that can help guide workflows, improve usability, support decision-making, and make laboratory operations more adaptive and resilient,” Wehrenberg added.

What this shift means for lab operations:

• Validation capacity is becoming a primary constraint
• Iterative feedback between models and experiments is increasingly expected
• AI is moving into core infrastructure rather than sitting alongside it

How expectations around lab automation are changing

How are customer expectations around lab automation evolving, and what capabilities are laboratories prioritizing?

Speed and throughput still matter, but they are no longer the only benchmarks. Laboratories are looking more closely at how systems perform over time, and how usable their data is.

“Today, laboratories are placing greater emphasis on robustness, reproducibility, connectivity, and flexibility,” Wehrenberg said.

That shift is tied directly to how automation feeds into downstream analysis. If data is inconsistent or difficult to access, it limits what teams can do with it later.

Interest in interoperability is also growing. “We are also seeing increased interest in open architectures and API-enabled systems that allow laboratories to integrate instruments, software platforms, and analytical tools more easily,” Wehrenberg explained.

Rather than optimizing individual steps, labs are focusing on how workflows connect end to end.

What laboratories are now prioritizing in automation:

• Consistent, reproducible data that supports analysis
• Systems that integrate with existing tools, platforms, and environments
• Flexibility to adapt workflows without major reconfiguration

Why laboratory systems need to work as a connected ecosystem

As automation platforms become increasingly sophisticated, how important is it that instruments, software, and data systems work together as part of a unified ecosystem?

Laboratories are generating large volumes of experimental and operational data, but much of it remains spread across different systems.

“The real value emerges when these systems can work together,” Wehrenberg said.

Connecting those datasets makes it easier to understand experimental outcomes in context—linking results with workflow conditions, instrument performance, and other variables.

Wehrenberg emphasized that this is not about replacing infrastructure. “At Tecan, we see our role not as creating a closed ecosystem, but as helping laboratories connect the ecosystems they already have,” he added.

Efforts such as supporting SiLA 2 standards and enabling APIs are aimed at making that connectivity more practical across mixed environments.

Where connected systems add the most value:

• Experimental results can be analyzed alongside contextual metadata
• It becomes easier to identify sources of variability, and inconsistency
• Labs can build on existing systems rather than replace them

Tackling scheduling and resource challenges in complex workflows

What are the biggest scheduling and resource-management challenges laboratories face today, and how can intelligent workflow management help?

In many environments, workflows now rely on coordinating multiple instruments and processes within a single setup.

“Some laboratory workflows require coordinating an entire room full of instruments as part of a single end-to-end process,” Wehrenberg said.

That introduces challenges around timing, dependencies, and resource allocation. Managing those effectively requires a broader view than individual instruments.

“Rather than starting with a liquid handler and building around it, we start with the scientific workflow and the problem the customer is trying to solve,” he explained.

Tools such as FlowPilot focus on that level, helping coordinate workflows to improve throughput, efficiency, and utilization without adding complexity.

Where workflow management can have the biggest impact:

• Reduces bottlenecks caused by scheduling conflicts
• Improves utilization of existing instruments and systems
• Keeps workflows aligned with experimental goals

What still holds back truly data-driven laboratories

What barriers still exist to creating truly data-driven labs, and how are they being addressed?

Despite progress, many laboratories are still working with fragmented data environments.

“Experimental data, operational telemetry, and environmental information are often stored in separate systems, making it difficult to generate the holistic insights needed to improve scientific outcomes,” Wehrenberg said.

Even when the tools are available, adoption can be uneven.

“Building a data-driven laboratory also requires trust in data, confidence in digital tools, and a willingness to evolve established ways of working,” he added.

In practice, teams that make progress tend to focus on specific use cases first, demonstrate value, and build from there.

Key barriers laboratories are still working through:

• Data fragmentation across instruments, systems, and platforms
• Limited confidence in new digital tools, and workflows
• Challenges in scaling beyond initial, proven use cases

What advanced AI capabilities now enable in the lab

Tecan recently announced a collaboration with NVIDIA to create data-driven labs. What does this partnership enable that was previously difficult or impossible to achieve?

Advances in AI are starting to move beyond data analysis and into how laboratories operate day to day. More recently, Tecan announced the integration of agentic AI capabilities into its Introspect platform, using the NVIDIA BioNeMo Agent Toolkit—a step that reflects how these systems are moving closer to practical deployment.

“Our collaboration with NVIDIA has helped us unlock grounded, guardrailed agentic AI capabilities that can operate reliably in laboratory environments,” Wehrenberg said.

The update allows AI agents to interact directly with laboratory data and workflows inside Introspect, supporting more continuous analysis of system performance, experimental conditions, and operational patterns.

“It is also helping us move from reactive to proactive laboratory operations,” he added.

Rather than identifying issues after they occur, these systems can begin to surface risks earlier, flag inefficiencies, and suggest actions based on patterns in the data. That shift—from monitoring to intervention—is central to how agentic AI is being positioned within lab environments.

What these AI capabilities enable:

• Earlier identification of risks and inefficiencies
• More accessible interaction with laboratory data
• A shift from reactive to proactive operations

What to prepare for over the next five years

What developments in digital laboratory technologies will have the greatest impact over the next five years, and what should researchers and lab leaders be preparing for today?

The pace of iteration between hypothesis and validation is expected to become a defining factor in laboratory performance.

“The greatest impact will come from reducing the time required to move from scientific concept to experimental validation while improving the quality of decisions made along the way,” Wehrenberg said.

That will depend on how well computational and experimental systems are connected, and how effectively data is reused.

Even as automation and AI advance, scientific expertise remains central to interpreting results and guiding research.

What laboratories should focus on now:

• Building strong data foundations, and interoperability
• Connecting computational and experimental workflows
• Improving decision-making alongside speed

 

This content includes text that has been created with the assistance of generative AI and has undergone editorial review before publishing. Technology Networks' AI policy can be found here.