In a breakthrough that could revolutionize the field of genetic engineering, startup Tetsuwan Scientific is developing AI-powered robots that can automate the tedious and time-consuming process of pipetting, a crucial step in genetic engineering. Founded by Cristian Ponce and Théo Schäfer, the company is addressing a long-standing problem in lab work, where manual labor has been the norm due to the limitations of existing automation solutions.

The story of Tetsuwan Scientific began at a Halloween party in 2023, where Ponce and Schäfer met and bonded over their experiences as lab technicians. Ponce, who has a background in bioengineering from Cal Tech, and Schäfer, who studied at MIT and worked at NASA's Jet Propulsion Lab, shared stories about the drudgery of manual labor in genetic engineering. This sparked the idea to create a solution that would automate the process, making it faster, cheaper, and more precise.

The challenge lay in developing a system that could adapt to the ever-changing parameters of scientific experiments. Existing automation solutions required specialized programming skills and were often expensive and inflexible. Tetsuwan Scientific's solution was to modify lower-cost white label lab robots and integrate them with large language models (LLMs). The breakthrough came when Ponce demonstrated the capabilities of OpenAI's multi-model product launch, which showed people talking to the model. This sparked the idea to use LLMs to interpret scientific data and provide suggestions for improvement.

In a remarkable "light bulb moment," Ponce showed that GPT 4 could successfully interpret an image of a DNA gel, identify a problem, and offer a detailed scientific suggestion for improvement. This demonstrated the potential of LLMs to diagnose scientific outputs, but the missing link was the physical agency to perform the suggested actions. Tetsuwan Scientific's robots are designed to fill this gap, providing a physical interface that can execute the suggestions made by the LLMs.

The company's robots are not humanoid but rather a square glass structure, designed to evaluate results and make modifications on their own, just like a human would. This requires building software and sensors that can understand properties like calibration, liquid class characterization, and other physical qualities of the liquids being pipetted. Tetsuwan Scientific has already secured an alpha customer, La Jolla labs, a biotech company working on RNA therapeutic drugs, and has raised $2.7 million in an oversubscribed pre-seed round.

The ultimate goal of Tetsuwan Scientific is to create independent AI scientists that can automate the entire scientific method, from hypothesis to repeatable results. Ponce believes that this technology has the potential to catalyze hyperbolic growth in scientific discovery, and he's not alone in this vision. Other organizations, such as non-profit org FutureHouse and Seattle-based Potato, are also working on AI scientists.

As the field of AI-powered scientific discovery continues to evolve, Tetsuwan Scientific is poised to play a significant role in revolutionizing lab work and unlocking new possibilities for scientific growth. With their innovative approach to automating genetic engineering, they are paving the way for a future where scientists can focus on higher-level thinking, leaving the tedious manual labor to the robots.