Global provider of laboratory automation and solutions, Tecan, recently integrated Agentic AI capabilities into its lab analytics platform Introspect. The toolkit leverages NVIDIA’s BioNeMo software allowing AI agents to access scientific AI capabilities directly via the Introspect platform, helping laboratories to streamline their operations.
Scientist Live asked Marco Ravot-Licheri, head of digital at the company’s life sciences business, to provide insights into the benefits of the collaboration.
What does agentic AI mean customers can do that they couldn’t before?
- Agentic AI enables proactive lab management. So instead of being informed after an error has happened, Agentic AI makes it possible to get notified before an error happens and have the chance to prevent it. This is an important step forward in lab operations.
What sort of proactive actions might they undertake? How will this affect their practices?
- Scientists can now easily set up an Error Diagnosis AI Agent to look for correlations between the scientific run they are about to execute and errors in the past. For example, this can uncover hidden patterns like a correlation between errors and temperature. Then, they can set up an Error Prevention AI Agent which constantly monitors the parameters linked with errors (e.g. temperature in this example) and informs the scientist when there is an increased risk of error, before the error actually happens. So the scientist can act in a timely manner and save the sample. Thanks to our work with NVIDIA, all of this can be now set up seamlessly using only natural language instructions on our analytics platform Introspect.
- In terms of impact on lab practices, this is meaningful because it finds the root causes of errors for their specific instruments (not generic specifications) and then gets the scientist to act only when needed. This allows scientists to actually focus on the science, with the peace of mind that they will be notified if their intervention is needed on the instrument.
Will it help with training?
- First of all this reduces the barriers to effectively use lab automation. And it can also be used to identify the protocols that are more prone to errors and tailor the training specifically on how to prevent the most frequent errors.
Do you have any case studies of projects on which agentic AI has been used?
- Labs in pharma, biotech and clinical diagnostics are already using Introspect’s Agentic AI capabilities with excellent initial results. In the following months we plan to make these Agentic AI capabilities available for all customers using Introspect.
Are there any more iterations of the product due for release and how do you expect Agentic AI it to change lab work, pharma and science over the next five or ten years time?
- Absolutely. If you think about the lab, it is a place where highly skilled scientists spend a lot of time on tedious repetitive tasks. I strongly believe in an AI-empowered human-centric approach, where AI enables the scientists to unlock their creativity and make new breakthrough discoveries.
- From the use of AI Agents for proactive lab analytics and for flexible experimental design closing the loop between the dry and the wet lab, all the way to Physical AI implementations making the instruments smarter and enabling effective error prevention and recovery, the lab is likely to become an even more exciting place in the next few years!
