Tetsuwan Scientific

Autonomous discovery (namely in materials and the life sciences) has been a quiet space. Ross King kicked off Adam & Eve in 1999, and it isn't until late 2023 that we see the next wave of autonomous systems for discovery (Coscientist/ChemCrow/Organa). We have been lucky enough to enjoy a small, friendly community with people who recognize that automating discovery is the most important thing we can do. This community will not be small for much longer. The back-to-back release of Evo 2 & Google's Coscientist paper this morning shows that we are approaching the inflection point we have been discussing since Halloween 2023. While both papers may focus on in silico models, the potential to use these models to drive autonomous systems is clear. However, models will be unable to meaningfully interact with the physical world unless we can solve the protocol-to-code problem. This problem is harder than it appears. How we can take a set of imperfect instructions as humans and use context and domain knowledge to subconsciously translate that set of instructions into a series of small, explicit actions (what we may call intuition or muscle memory) is a non-trivial thing to engineer. Simulating intuition to bridge models and lab robots also creates a bridge between scientists without automation engineering experience and lab robots. If you want to work in the most consequential field of our time, work in autonomous discovery. If you want to work in a small group at the wild frontiers of this field working to solve problems like emulating experimental intuition, come work with us.

Thank you to Nabil Laoudji for having us on for the inaugural episode of Discovery Engines! We can't wait to see what comes next in Nabil's exploration of the intersection of AI and science! https://lnkd.in/e2ZyBmS6

Pioneering autonomous AI researchers, Tetsuwan Scientific, is pushing the boundaries of lab automation to bring the next revolution in scientific research #RoboticScientist #TetsuwanScientific #AIinScience #TechFunding

AI scientists will catalyze a new era of scientific discovery. These systems combine agentic AI with lab robotics to address tasks previously unsolvable via traditional lab automation. This approach expands automation beyond the high-throughput, low-variety paradigm, unlocking new possibilities and use cases. Following delivery to our first customer last week, we are emerging from stealth to announce a $2.7M pre-seed round led by 2048 Ventures to pursue the development of these systems. This round has supported the rapid growth of our team, as we have grown to a group of seven from across Caltech, ETH Zürich, Massachusetts Institute of Technology, and more. This team represents experience from across bioengineering, lab automation, natural language processing, and autonomous robotics. More importantly, this funding has enabled us to rapidly build and iterate towards our grand vision. The AI scientist community is incredibly nascent- we are proud to build on the foundations laid late last year by papers like Coscientist, ChemCrow, Organa, Bioplanner, and many more. There are things that we are wrong about today and mistakes we will make tomorrow, but we will break, build, and work to advance this field forwards relentlessly to create tools that empower scientists. In the short-to-medium term, AI scientists/lab assistants will enable more reproducible experimentation to be done than ever before by unlocking lab automation for high-variety tasks. In the long term, AI scientists and the automation of the scientific method will catalyze an unprecedented rate of innovation and will be remembered as one of humanity’s most consequential technical breakthroughs. Read more about our work at tetsuwan.com (check out our out of stealth blog post!) or in our SynBioBeta announcement: https://lnkd.in/eFfEPDVX We’d like to give a special thank you to our lead investors at 2048, as well as Carbon Silicon Ventures and Everywhere Ventures, our incubator (Entrepreneur First) & program advisor Mickey Kertesz, our technical advisor Aaron Kimball, and the hundreds of people in the automation and life sciences community that have been so generous with their time. The process of scientific discovery will soon change in a monumental way, and it is in large part due to the generosity and patience of those willing to help people like us.

Stefan Golas and the labautomation.io community do incredible work in making lab automation more accessible, ultimately speeding up our rate of scientific discovery. We are happy to announce that we are now the sponsors of the community forum! We hope to help grow the lab automation community and provide community leaders like Stefan with the resources they need. In contributing to the grand goal of automating the scientific method, we must bridge the gap between hypotheses & experiments and execution on automation platforms. Today, automation engineers are the experts who perform this translation. Our ability to reach this final frontier of discovery is largely contingent on progress in solving this important problem- we look forward to achieving progress alongside the open-source community.