AI adoption is widespread across life sciences laboratories, but it is still mostly in the experimental stage, with only 5 percent using AI agents in production, according to Cenevo’s second annual survey.

Cenevo, a specialist in enabling agentic, connected labs for life sciences through its agentic lab platform, recently surveyed more than 110 life sciences professionals across a variety of lab environments: R&D, discovery, chemistry, biology, clinical, and manufacturing, to reveal insights into AI adoption and how organizations are balancing innovation with the operational realities of modern lab environments.

While some researchers are wary of today’s versions of AI, with 58 percent having privacy or security concerns, AI is going to be a critical component in lab operations for the long term. Researchers are prioritizing the use of AI for data analysis and interpretation, workflow automation and orchestration, experiment design and planning, and sample and inventory management, rather than agentic-driven scientific discovery and decision making at this stage.

More than 60 percent of labs are exploring or piloting AI, with 57 percent using it for data analysis. 25 percent are already using generative AI in full production environments. Usage of agents to perform discrete, previously human tasks or more complex multi-agent workflows is also becoming more common, with 27 percent exploring or piloting the possibilities but only 5 percent already using it in production.

Budgets are reflecting labs’ quest to address connectivity, integration, and data challenges; instead of focusing on purchasing standalone tools, lab leaders say their investment priorities are on automation, AI-enabled software, systems integration, and data infrastructure and analytics. Connecting laboratory information management systems (LIMS), electronic lab notebooks (ELNs), and instruments is a priority, reported by 62 percent of small and medium-sized organizations and 50 percent of all organizations.

This year and last, scientists reported that connectivity is key to maximizing the benefits of AI overall. More than half lack integration among systems, and one-third still rely on manual operations. Overall, though, progress in automation has accelerated, as last year’s survey showed that more than half of labs relied on manual operations.

Data is still a core bottleneck, however. While 42 percent are reporting that data quality, overload, and management are issues blocking AI adoption, that is a drop from the 54 percent who cited it last year. Lab leaders still report significant challenges in making effective use of their data, with 55 percent reporting a lack of integration between systems as being the biggest problem, closely followed by managing unstructured or inconsistent data as well as data spread across instruments and other teams.

Visit Cenevo for the full report