Founded by Deepmind Alumnus, Latet Labs launches with $ 50 million to make the biology programmable

A new startup founded by an old Google Deepmind The scientist comes out of stealth with $ 50 million in funding.

Latent laboratories Built AI foundation models to “make programmable biology”, and it plans to associate with biotechnological and pharmaceutical companies to generate and optimize proteins.

It is impossible to understand what Deepmind and his fellows do without first understanding the role that proteins play in human biology. Proteins lead everything into living cells, enzymes and antibodies. They are made up of around 20 distinct amino acids, which connect together in strings that bend to create a 3D structure, the shape of which determines the functioning of the protein.

But to determine the shape of each protein was historically a very slow process and with a high intensity of labor. It was the great breakthrough that Deepmind made with Alphafold: He marked automatic learning with real biological data to predict the form of some 200 million protein structures.

Armed with such data, scientists can better understand diseases, design new drugs and even Create synthetic proteins For completely new use cases. This is where latent laboratories enter the fray with its ambition to allow researchers to “create therapeutic molecules on the computer level from zero.

Latent potential

Simon Kohl (photo above) started as a researcher in Deepmind, working with the nucleus Alphafold2 team before co-directing the protein design team and Deepmind wet laboratory configuration At the Francis Crick Institute in London. At that time, Deepmind also generated a sister -based business in the form of isomorphic laboratorieswhich focuses on the application of research on Deepmind AI to transform the discovery of drugs.

It was a combination of these developments which convinced Kohl that the time had come to go rider alone with a leaks leaks specifically on the construction of frontier models (i.e. fucking point) for the design of proteins. Thus, at the end of 2022, Kohl left Deepmind to throw the foundations of latent laboratories and incorporated the company into London in mid-2023.

“I had a fantastic and impactful moment [at DeepMind]And has become convinced of the impact that generative modeling was going to have in biology and protein design in particular, “Kohl told Techcrunch in an interview this week. “At the same time, I saw this with the launch of isomorphic laboratories and their Plans based on alphafold2that they started a lot at the same time. I felt like the opportunity was really to walk in laser on protein design. The design of proteins, in itself, is such a vast field and has so much unexplored white space that I thought that a truly agile and targeted outfit would be able to translate this impact. »»

Translating this impact as a startup supported by a company involved hiring around fifteen employees, including two from Deepmind, a main engineer from Microsoft, and doctorates from the University of Cambridge. Today, the latent workforce is divided on two sites – one in London, where the magic of the frontier model occurs, and another in San Francisco, with its own wet laboratory and the computer protein design team.

“This allows us to test our models in the real world and get the comments we need to understand if our models progress as we wish,” Kohl said.

While wet laboratories appear in the agenda in the short term in terms of validation of latent technology predictions, the ultimate objective is to cancel the need for wet laboratories.

“Our mission is to make programmable biology, really bringing biology in the computer field, where dependence on biological and humid laboratory experiences will be reduced over time,” said Kohl.

This highlights one of the main advantages to “make programmable biology” – overwhelming a process of discovery of drugs which is currently based on countless experiences and iteration that can take years.

“This allows us to make really personalized molecules without counting on the wet laboratory – at least, it is the vision,” continued Kohl. “Imagine a world where someone comes with a hypothesis on the target of the drug to follow for a particular disease, and our models could, a” push button “, make a protein medication that comes with all the desired properties. “”

Biology activity

In terms of commercial model, latent laboratories do not consider themselves “centered on assets” – which means that it will not develop its own internal therapeutic candidates. Instead, he wants to work with third -party partners to accelerate and risk the previous R&D stages.

“We think that the biggest impact that we can have as a business is to allow other biopharmatic companies, biotechnologies and life sciences – either by giving them direct access to our models, or by supporting their programs as Discovery via project -based partnerships, “said Kohl.

The cash injection of $ 50 million from the company includes a seed branch of $ 10 million previously unexpected, and a new series of $ 40 million per round led by Radical Ventures – in particular, partner Aaron Rosenbergwho was previously head of strategy and operations at Deepmind.

The other co-directed investor is Sofinnova Partners, a French captain company with a long track record in the space of life sciences. The other participants in the round include Flying Fish, ISOMER, 8VC, Kincred Capital, Pillar VC and notable angels such as the chief scientist of Google Jeff Dean, the founder of Cohere Aidan Gomez and the founder of Elevenlabs Mati Staniszewski.

Although a piece of money will go to wages, including those of the new automatic learning hires, a significant sum of money will be necessary to cover the infrastructure.

“Calculation is also a significant cost for us – we build fairly important models, I think it is just to say, and that requires a lot of GPU calculation,” said Kohl. “This funding really puts us to double on everything – to acquire calculation to continue to scale our model, to put the teams on the scale and to start building the bandwidth and the ability to have these partnerships and The commercial traction that we are looking for now. “”

Aside from Deepmind, several startups and scales supported by companies seek to bring the worlds of calculation and biology closer, like the cradle And Bioptimus. Kohl, for its part, thinks that we are still at a sufficiently early stage, where we still do not know what will be the best approach in terms of decoding and design of biological systems.

“There have been very interesting seeds planted, [for example] With Alphafold and a few other early generative models from other groups, ”said Kohl. “But this area has not converged in terms of what is the best model approach, nor in terms of what the business model will work here. I think we have the capacity to really innovate. »»