Artificial intelligence in life sciences labs is moving beyond simple automation and isolated digital tools. In this Q&A, Jonathan Gross, Chief Product Officer at Cenevo, discusses what agentic AI means in a laboratory setting, why data and integration still come first, and how labs can balance autonomous workflows with scientific oversight. He also outlines where analytical scientists are likely to see the greatest impact as labs build the infrastructure needed to support it.
How do you define “agentic AI” in life sciences laboratories, and how does it differ from earlier automation or AI approaches?
Labs are essentially physical and that presents more of a challenge for the AI revolution in life sciences, than in other predominantly digital industries. However, earlier automation, device integrations, and the sample and experimental data collected gives us huge benefits for the future of AI and to drive results.
With this solid foundation, we are now beginning to layer analytics, MCP servers to open up data, and other agentic mechanisms on top. We are using these tools and capabilities to apply generative AI and agentic AI with our customers.
Working in analytical science?
Register for a FREE Separation Science account to subscribe to the Separation Science Newsletter.
Traditional automation follows predefined rules, while generative AI can summarize data and draft documentation. Agentic AI goes a step further by taking approved actions across laboratory systems. For example, it could assemble a run plan based on sample priority, instrument availability, and validated methods; check whether metadata is complete; and prepare results for scientist review. In that model, scientists spend less time on setup and administrative work, and more time reviewing exceptions, validating results, and making scientific decisions. Not everyone is there yet—our job is to help them get there faster. The game changer is agentic AI taking over processes that previously had to be carried out by people.
Why is the laboratory environment a distinct challenge for AI adoption?
Labs are in a unique position when it comes to the “AI revolution.” While the ultimate result of lab processes is massive volumes of data, that data is generated from “physical” activities. Therefore, labs still need to address the basics first, such as more effective instrument integration and automation. Without those, it’s not possible to maximize the value of AI, either generative or agentic.
How should laboratories balance autonomous AI workflows with human oversight to maintain data integrity, compliance, and scientific accountability?
Human scientists in the loop are critical to the success of AI. While AI can replace manual, repetitive tasks, it’s still critical to keep human validation. Scientific accountability needs to remain top of mind. Scientists need to feel confident that AI is acting correctly. Instead of reviewing every process performed by AI, evaluating a small representative subset based on the task, risk level, and compliance requirements should be sufficient.
Data integrity and compliance still remain integral aspects of lab activities, no matter the level of AI adoption. Data is the lifeblood of the lab. In many labs, that data is spread among a variety of systems and instruments. Without appropriate—and compliant—data management, AI adoption and the effectiveness of any adopted AI is going to be compromised.
Over the next three to five years, where will analytical scientists see the most meaningful impact from agentic AI in everyday laboratory workflows?
Agentic AI is going to replace manual tasks, once the infrastructure is in place to make it worthwhile. Last year, Cenevo surveyed more than 150 scientists across a variety of fields. 75 percent of those surveyed stated AI or machine learning will play a significant role in driving change this year. Data and sample management, device integrations, and automation were cited as the most critical tasks related to AI implementation. Once these are achieved, the benefits of AI can be realized.
