Company claims AI model armed with physics, chemistry expertise can cut inefficiencies in mass production.
Toronto-based Basetwo has closed a $11.5 million USD ($16.5 million CAD) Series A round for its artificial intelligence (AI)-powered copilot for manufacturing engineers.
The software-as-a-service (SaaS) startup is trying to solve tech inefficiencies in large-scale chemical manufacturing. Basetwo says its AI platform can cut energy consumption and costs for the mass production of pharmaceuticals, consumer goods, and gas processing.
According to Gaffoor, AI models don’t understand foundational physics or chemistry, hindering AI deployment in manufacturing.
The all-equity, all-primary round was led by Paris-based firm Axa Venture Partners (AVP), with participation from Glasswing Ventures, Deloitte Ventures, Global Brain Ventures, Shimadzu Corporation, Chiyoda Corporation, and United Arab Emirates (UAE)-based angel investors. Investors from AVP and Glasswing are now on Basetwo’s board of directors.
The financing, which closed in September, brings the company’s total amount raised to $17.5 million USD, following its 2022 seed round led by Glasswing Ventures and Argon Ventures.
Basetwo’s AI copilot seeks to help chemical manufacturers make production more efficient. For pharmaceutical companies, this could mean identifying and setting the optimal reactor temperatures and mixing speeds when scaling up the production of new drugs.
Thouheed Abdul Gaffoor, CEO and co-founder of Basetwo, said that most generative AI applications have focused on consumer use cases, but optimizing manufacturing through AI requires a different approach that incorporates expertise in physics and engineering.
The biggest problem with deploying AI in manufacturing, according to Gaffoor, is that AI models don’t understand the foundations of physics and chemistry. Instead, they generate outputs based only on historical manufacturing data.
“This is okay for problems such as classifying cats or dogs in images,” Gaffoor told BetaKit. “Still, this would be highly problematic when trying to simulate engineering problems such as chemical reactions in tanks, combustion in engines, or distillation in columns.”
Gaffoor said that Basetwo’s model combines this data with physics and chemistry principles found in engineering handbooks, chemical data tables, and equipment specification files. The approach is handy for pharmaceutical manufacturers, who may be testing new compounds and have limited data available about how they behave in mass production.
Source: BetaKit