Apoha surfaces from stealth mode with $36 million in funding to instruct machines on the behavior of matter.

      Science is capable of explaining what a molecule is and its appearance. However, it has not been able to provide information about how these molecules behave in the intricate conditions of the real world in a cost-effective manner and at scale. This deficiency is where drugs often fail during trials, where food products fail to meet taste expectations, and where artificial intelligence starts to struggle increasingly.

      Apoha, a company out of London that emerged from 15 years of research in interfacial physics, claims to have developed the crucial measurement needed. On June 3, it came out of stealth mode with $36 million in funding, which was announced at the Frontier Technologies Stage at SXSW London.

      The investment round is led by Singular and includes participation from Tim Draper’s Draper Associates, as well as continued support from initial investors Redalpine, Seedcamp, Wilbe, and Nucleus, along with grant funding from Innovate UK.

      The company has introduced a data layer called Liquid State Intelligence, which it defines as a new category complementary to sequence and structure. While genomics has digitized the language of biology and structural biology has digitized design, Apoha aims to digitize behavior: what matter actually does when under stress. According to the company, the funding will support the establishment of this as a key data category for biologics, food, materials, and physical-world artificial intelligence.

      The scientific foundation dates back to 2008 when founder and CEO Shamit Shrivastava started addressing a gap left by the Nobel-winning Hodgkin-Huxley model concerning the physics at the intersection of matter and liquid. He later published findings on two-dimensional solitary sound waves at a lipid interface in 2014, which the company notes were recognized among Scientific American’s discoveries that could potentially transform various fields. In 2021, he co-founded Apoha with Anshika Srivastava, its COO and a former executive director at Goldman Sachs.

      Currently, the company holds over 60 patents related to hardware, software, data, and AI models.

      Its first product, VIBE, provides an empirical assessment of how a sample reacts under controlled stress. This platform uses a small quantity of material, enough to fit on a pinhead, suspends it in liquid, applies a series of perturbations, and captures the wave patterns emitted by the molecule in response.

      These patterns yield over 1,000 measured behavioral descriptors in a single reading, in contrast to traditional assays that evaluate one attribute at a time. The company claims that within minutes, a VIBE readout can indicate if an experimental drug is likely to fail before entering trials.

      The platform is already being utilized commercially, with strong evidence found in a preprint. In collaboration with Boehringer Ingelheim, a long-term commercial partner, Apoha identified high-risk antibody candidates with over 90% accuracy using only 8 micrograms of material.

      A second phase of benchmarking indicates that the platform surpasses 12 industry-standard developability tests involving 236 clinical antibodies, revealing information that conventional methods tend to overlook rather than simply replicating them.

      Other clients emphasize the platform's versatility. Apoha is collaborating with German biotech company Ethris to predict the behavior of lipid nanoparticles carrying mRNA in animals, and with plant-based food company THIS for a protein alternative aimed at supermarket shelves. The company also works with Somru BioSciences and several Fortune 500 companies across pharmaceuticals, food, and materials.

      The broader expectation is that physical-world AI will eventually require this type of data. Current models can see and interpret, and a new generation of physical AI systems is being developed to interact with matter. However, these systems are not yet capable of understanding how a drug dissolves or how a flavor is sustained, as such data has not been collected at scale.

      “It cannot be scraped from the internet, synthesized, or retrofitted from existing assays,” said Shrivastava. “It has to be measured.” The key question for the next funding round will be whether enough buyers recognize this as a valuable data class.