The London-based start-up Apoha has launched with technology designed to enable machines to simulate feel, taste, and smell and to understand how matter behaves, using data generated in its own wet lab. The company also announced that it has raised $36 million from private investors.
Apoha describes its technology as being able to provide a “missing layer” of structured data relating to the behavior of matter, formulations, and materials. The firm was cofounded in 2021 by Shamit Shrivastava, a former University of Oxford postdoc and mechanical engineer, and his wife Anshika Srivastava, a former Goldman Sachs executive. Apoha is coming out of stealth status after the fundraising with a staff of about 25 staff.
To generate data, Apoha’s roboticized lab system takes a sample of a material—such as lipid, protein, or even soil—that is small enough to fit on a pinhead and suspends it in a liquid. The company then applies a series of controlled stresses to the sample and captures the resulting wave pattern data.
Within minutes its machine learning–based artificial intelligence system is able to convert the wave pattern data into more than 1,000 descriptors of behavior, Shrivastava says. Apoha uses a second AI system to convert this descriptor data to information about a material’s suitability for certain applications. Conventional approaches, in contrast, measure one property at a time.
As a concept “it’s like a new kind of Raman,” Shrivastava says, referring to Raman spectroscopy, which involves illuminating a substance with a laser and analyzing the light that is scattered.
The technology “took 15 years of science and 5 years of company-building to bring to life,” he says in an emailed announcement of the launch of the firm. “There is no shortcut to this data class—it cannot be scraped from the internet, synthesised, or retrofitted from existing assays.”
Apoha’s technology, called Liquid State Intelligence (LSI), is protected by more than 60 patents covering hardware, software, data, and AI models, the company says.
The firm is already applying LSI to pharmaceuticals, foods, beverages, and other materials by partnering with companies that include the drug firms Boehringer Ingelheim and Somru BioScience. Boehringer has used Apoha’s platform to identify antibody candidates from just 8 mcg of material, Apoha says.
Data generated from Apoha’s wave pattern data test, published in a preprint research paper that has not been peer-reviewed, were used to identify antibody-drug candidates carrying multiple biophysical liabilities. The paper was coauthored by Apoha scientists and Boehringer principal scientist Maureen Crames. The study shows that LSI outperformed 12 tests used by pharmaceutical companies to determine how a drug candidate will behave in real-world conditions, the company says.
