The use of artificial intelligence in bioprocessing might still be more talk than action. “Lots of people are talking about it,” says William Moss, CEO and co-founder of U.K.-based Opvia, “but most of the industry uses very little machine learning or artificial intelligence.”
So far, Moss sees most bioprocess R&D teams—from small startups to big pharma—struggling with the same problem: “Scientists don’t have the tools to capture their data in a standardized way with all of the context.” Unless a company can solve that problem, there’s little they can do with the data. As Moss says, “That’s the bottleneck.”
Moss and his colleagues at Opvia are building software to solve this problem. “Before working with us, scientists had to manually manage hundreds of spreadsheets and loose files,” he says. “Now, they can leverage the power of a database—from instant visualizations of historic data and reports to deploying machine-learning models for better design of experiment.”
The data that companies have to handle varies widely and the tools need to be extremely flexible. “Processes and challenges vary widely,” Moss explains. “We aren’t the experts, the scientists are, so we’re here to enable them.” To overcome the challenges of flexibility and training, Opvia’s cloud-based software provides the familiarity of a spreadsheet, but the capabilities of a database.
Moss and his colleagues developed this system while working with biotech startups, then a contract development and manufacturing organization, and are now starting to work with large pharma. Very soon, Moss says, “we’ll be at the point where our software is the interface between scientists and their data across whole companies.”
That’s just a start for Moss. As a company collects more data, AI-driven design of experiment and eventually simulation will replace wet-lab experiments. “That’s where we’ll really see a change,” he says. “This is going to be crucial as products become more personalized.”