Apoha raises $36M to teach machines how matter behaves using liquid wave form data

Apoha emerges from stealth with $36M to teach machines how matter behaves

The London deeptech wants to build a third data layer for molecular science, after sequence and structure: behaviour. Science can already tell you what a molecule is and what it looks like. What it has never been able to tell you, cheaply and at scale, is how the thing behaves once it meets the messy conditions of the real world. That gap is where drugs quietly fail in trials, where food products miss the palate they were built for, and where, increasingly, artificial intelligence runs out of road. Apoha, a London company spun out of 15 years of interfacial physics, says it has built the missing measurement. On 3 June it emerged from stealth with $36M in funding, announced at the Frontier Technologies Stage at SXSW London. The round is led by Singular, with participation from Tim Draper's Draper Associates and continued backing from seed investors Redalpine, Seedcamp, Wilbe and Nucleus, alongside grant funding from Innovate UK. The company calls its data layer Liquid State Intelligence, a new category it places alongside sequence and structure. Where genomics digitised the language of biology and structural biology digitised design, Apoha wants to digitise behaviour: what matter actually does under stress. The funding, it says, will go toward making that a foundational data class for biologics, food, materials and physical-world AI. The science traces to 2008, when founder and chief executive Shamit Shrivastava began working on a problem the Nobel-winning Hodgkin-Huxley model of nerve signalling had left open: the physics of the boundary where matter meets liquid. He went on to publish evidence for two-dimensional solitary sound waves at a lipid interface in 2014, work the company says was later named among Scientific American's discoveries that could change everything. In 2021 he co-founded Apoha with Anshika Srivastava, its chief operating officer and a former executive director at Goldman Sachs. The company now holds more than 60 patents across hardware, software, data and AI models. Its first product is VIBE, an empirical readout of how a sample behaves under controlled stress. The platform takes a quantity of material small enough to sit on a pinhead, suspends it in liquid, applies a sequence of perturbations, and records the wave patterns the molecule throws off in response. Those patterns resolve into more than 1,000 measured descriptors of behaviour in a single reading, where conventional assays capture one property at a time. Within minutes, the company says, a VIBE readout can flag whether an experimental drug will fail before it reaches a trial. The platform is already in commercial use, and the firmest evidence sits in a preprint. In joint research with Boehringer Ingelheim, a multi-year commercial partner, Apoha identified high-risk antibody candidates with greater than 90% precision from as little as 8 micrograms of material. A second version of the benchmarking work reports the platform outperforming 12 industry-standard developability tests across 236 clinical antibodies, and surfacing information the conventional measures miss rather than duplicating them. Other customers point to range. Apoha is working with German biotech Ethris on predicting how lipid nanoparticles carrying mRNA behave in animals, and with plant-based food company THIS on a protein replacement bound for supermarket shelves. It also lists Somru BioSciences and several Fortune 500 companies across pharma, food and materials. The wider bet is that physical-world AI will eventually need this. Models have learned to see and read, and a generation of physical AI systems is now being built to act on matter. None of them can yet feel how a drug dissolves or how a flavour holds, because that data has never been collected at scale. "It cannot be scraped from the internet, synthesised, or retrofitted from existing assays," Shrivastava said. "It has to be measured." Whether enough buyers agree to make a data class out of it is the question the next round will have to answer.

Apoha, a startup using building AI based on new liquid 'wave form' data, emerges from stealth with $36 million Series A round | Fortune

Apoha, a deep tech startup that is building AI models for creating new kinds of substances -- from proteins to food products to paints -- based on a new kind of data about how materials behave, is emerging from stealth today with $36 million in venture capital funding. The funding round, which is the London- and San Francisco-based startup's Series A, is being led by European venture capital firm Singular, with participation from Draper Associates and continued backing from existing seed investors Redalpine, Seedcamp, Wilbe, and Nucleus. The company also has a grant from Innovate UK, the U.K.'s national innovation agency. The company did not disclose its valuation following the funding. Apoha is betting that the key to unlocking many new kinds of materials rests in a kind of data that does not exist yet at scale: measurements of the wave forms these materials generate when suspended in a liquid and then acted on by outside forces. It turns out that these warm forms are unique to each material and also correlate to its properties, including qualities such as smell and taste, as well as things like reactivity. With enough of this wave data, Apoha's AI models will be able to suggest ways to modify or create a material in order to obtain the exact characteristics a user desires. Apoha calls this new AI method "liquid intelligence." "Machines have learned to see what matter looks like and to read what we say about it," Anshika Srivastava, Apopha's cofounder and chief operating officer, said. Many AI models are trained only on text or on image data. "They have not learned to taste, smell, or feel matter -- to perceive how a drug dissolves, how a flavour holds, how a material wears. That is the layer we are building." Srivastava, a former Goldman Sachs banker, cofounded Apoha in 2021 alongside Shamit Shrivastava, a mechanical engineer who did post-doctoral research at the University of Oxford after completing a PhD at Boston University. Shrivastava, who is now Apoha's CEO, pioneered the methods on which the company's technology is based. He holds the patent on the liquid wave form analysis the company uses to create the data for its AI models as well as on many of the specialized hardware devices the company has had to create to carry out its experiments. The company's name comes from a Sanskrit word that means "negation or exclusion" and is part of Buddhist philosophy that things are defined by what they are not more than by what they are. Apoha has built a piece of laboratory hardware that takes a sample of material so small it would fit on the head of a pin, suspends it in a liquid, and then applies a controlled series of tiny physical stresses to it. The device records the wave patterns that ripple back through the liquid in response. According to the company, those patterns yield more than 1,000 distinct numerical descriptors of how the material behaves, captured in a single run that takes minutes rather than the days or weeks conventional lab tests require. That readout -- which the company calls VIBE, short for Variations in Inter-facial Behaviour Under Excitation -- is its first commercial product. Apoha then turns the raw recordings into what Shrivastava calls a "behavioral embedding," a numerical fingerprint that AI models can be trained to recognize, compare and learn from. The VIBE measurement, Apoha's cofounders say, can predict whether a drug will hold together inside the body, whether a plant-based protein will tear apart on the tongue like chicken meat, or how a new material will wear over time. One Apoha's first customers was a food company that had to find a substitute for the key component in its plant-based vegan "chicken" within two weeks after a previous supplier went out of business. In pharma, the immediate use case is screening drug candidates before they enter expensive clinical trials. The company says a multi-year research partnership with German pharmaceutical firm Boehringer Ingelheim has shown Apoha identifying high-risk antibody candidates with greater than 90% precision from as little as 8 micrograms of material. In a separate benchmark on a dataset of 236 antibodies that had reached clinical trials, the company says its platform outperformed 12 industry-standard tests pharma firms currently use to predict whether a drug will fail in patients. Catching such failures earlier could save drugmakers hundreds of millions of dollars per failed candidate, Apoha says. Outside pharma, Apoha is working with German biotech Ethris on predicting how lipid nanoparticles carrying mRNA -- the same kind of delivery vehicle used in some COVID-19 vaccines -- will behave in animals. The startup also works with Somru BioSciences and what it describes as multiple Fortune 500 customers across pharma, food and beverage, and materials. Apoha says it has completed a total of about 40 customer projects to date. The company has about 25 employees. Srivastava said the Series A funds will go toward scaling Apoha's platform -- which includes custom hardware for carrying out the experiments needed to obtain the VIBE data, as well as the AI models built from the data -- to handle more sample types and more customers. Raffi Kamber, co-founder and general partner at Singular, said in a statement that Apoha represents "a new generation of European scientific companies where AI is not a future promise, but a practical tool already transforming how biology is done."