Profluent Announces Launch With $9 Million Seed Round

5 min read

Profluent announces its launch with a $9 million seed round led by global investor Insight Partners, with participation from Air Street Capital, AIX Ventures, and Phoenix Venture Partners. Profluent is using deep generative models to ‘learn the language of biology’ in order to design new, functional proteins.

The goal of protein design is to create novel or enhanced proteins with specific uses. These could include lasting cures for diseases that are free from patent restrictions or new enzymes that can break down unrecyclable plastics. Until now, however, the field has been reliant on two traditional methods: costly, time-consuming searches for existing proteins in nature, or attempting to introduce small edits to a protein in the hope that they achieve the desired outcome eventually.

Instead, Profluent is applying the power of generative artificial intelligence (AI) models to the challenge of designing new biomolecules. Like language models for text, Profluent’s models are trained on large-scale data and learn from elegant, creative objective functions, but instead of adjectives and nouns they learn the language of the genetic code.

“While companies are experimenting with exciting new biotechnology like CRISPR genome editing by repurposing what nature has given us, we’re doing something different. We use AI and large language models like the ones which power ChatGPT to learn the fundamental language of biology, and design new proteins which have the potential to cure diseases” said Profluent founder Ali Madani, “Due to our method’s wide-ranging versatility, the applications aren’t just limited to human health. We’re excited to pair the latest advances to build the solutions and tooling to engineer a new biocentury”.

The launch comes alongside a peer-reviewed paper published today in Nature Biotechnology, summarising work led by Profluent founder Ali Madani, who previously led research in large language models at Salesforce AI Research. Ali was the first in the world to apply large-scale generative models for controllable design of full-length artificial proteins. The paper details a demonstration of protein language models generating entire proteins which, despite being novel, function out-of-the-box as well as natural proteins that have been optimised by millions of years of evolution. The study is comprehensive across multiple protein families and includes experimental verification that extends to atomic-level characterization.

“I have first-hand knowledge in rigorously characterising the AI-designed proteins along multiple dimensions experimentally. Proteins designed using Profluent’s platform are very impressive. Consistently, Ali and his colleagues hold themselves to the highest possible scientific standards, ensuring that they back their claims with experimental data, and consistently iterating on their results.” said James Fraser, a Professor of Bioengineering and Vice Dean of Research in the School of Pharmacy at University of California, San Francisco (UCSF).

Richard Socher, former Chief Scientist at Salesforce, CEO of AI-based search engine you.com, and co-author on the Nature Biotechnology publication said: “Having worked closely alongside Ali at Salesforce, I have witnessed his unparalleled ability to possess a deeply technical knowledge of machine learning paired with an interdisciplinary vision for the application of AI for the biological sciences. I’m excited to continue supporting his journey at Profluent.”

“Ali and his team are doing groundbreaking work marrying deep generative models with the world of protein design. Even with all the recent AI-driven breakthroughs in protein structure prediction, Profluent’s work stands out. Their methods will transform how we design valuable proteins and enable solutions where blocking IP currently slows the advancement of life-saving new medicines” said Dylan Morris, Managing Director at Insight Partners.

“While the majority of the AI world focuses on generating spectacular artwork or amusing text, Profluent’s AI has the power to radically improve drug discovery. Their generative models are broadly-applicable across multiple industries and modalities including enzymes and antibodies” said Nathan Benaich, General Partner at Air Street Capital.

The proceeds from the seed round will be used to build out an integrated wet laboratory in Berkeley, CA, which allows Profluent to create a tight feedback loop between the data produced through experimental methods, and their AI systems, providing robust validation of any designed proteins and continual improvements to their AI.