Cristian Ponce met his co-founder Théo Schäfer at a Halloween party in 2023, where Ponce was dressed as Indiana Jones. The event, hosted by Entrepreneur First, a startup incubator, connects aspiring founders to launch innovative ideas. It was here that Ponce first encountered Schäfer, whose background at MIT and NASA’s Jet Propulsion Lab, where he explored Jupiter’s moons for alien life, fascinated Ponce. For his part, Ponce had studied bioengineering at Caltech, working with E. coli. Despite their distinct scientific backgrounds, the two bonded over shared frustrations with the tediousness of lab work.
As lab technicians, both had spent countless hours performing repetitive tasks, particularly in genetic engineering. One such task involves using a syringe-like tool called a “pipette” to transfer liquids between test tubes. While automation has attempted to ease this burden, the robots capable of doing so are expensive, specialized, and require extensive programming. If a scientist needs to adjust an experiment, they often face delays waiting for the bot to be reprogrammed. As a result, humans are still often more efficient at these tasks, despite the manual labor involved.
Motivated by this challenge, Ponce and Schäfer co-founded Tetsuwan Scientific to develop cost-effective lab robots that could ease these burdens. However, the breakthrough came in May 2024 when the co-founders watched OpenAI’s multi-modal product launch, showcasing AI’s advancements in scientific reasoning. Ponce had an epiphany. He immediately tested GPT-4 by showing it an image of a DNA gel. The AI not only identified the image but also detected an error—a primer dimer—and proposed a detailed scientific explanation and solution.
This moment illuminated a gap in the current technology: while large language models (LLMs) could analyze and interpret scientific data, they lacked the capability to physically perform the necessary tasks. Ponce and Schäfer realized that the key to revolutionizing lab automation lay in combining AI’s reasoning power with robots capable of physically executing the suggestions made by the AI.
Tetsuwan’s research soon revealed another significant hurdle: no software existed to convert scientific intent into robotic execution. For example, robots couldn’t understand the properties of the liquids they were handling, such as whether a substance was viscous or likely to crystallize. This gap in understanding highlighted the need for more advanced robots that could evaluate results and make real-time modifications, much like a human scientist would.
To address these challenges, Tetsuwan Scientific has focused on developing robots equipped with sophisticated software and sensors that can understand properties like liquid viscosity, calibration, and other key variables. These robots are not humanoid but rather function as square glass structures capable of autonomously making adjustments based on experimental data.
Currently, Tetsuwan Scientific is working with an alpha customer, La Jolla Labs, a biotech firm specializing in RNA therapeutics. The robots are assisting in determining the effectiveness of drug dosages. Additionally, the company raised $2.7 million in an oversubscribed pre-seed round, attracting investment from 2048 Ventures, Carbon Silicon Ventures, Everywhere Ventures, and prominent biotech angel investors.
Ponce is particularly enthusiastic about the future of Tetsuwan Scientific. He envisions a future where AI-driven robots can fully automate the scientific method—from forming hypotheses to conducting experiments and generating repeatable results. “It’s the craziest thing we could possibly work on. Any technology that automates the scientific method is the catalyst for hyperbolic growth,” he says.
Tetsuwan Scientific is not alone in its pursuit of AI-powered scientific discovery. Other groups, like the non-profit FutureHouse and Seattle-based Potato AI, are also working on similar technologies to transform the field of science. With the combination of AI and robotics, the possibilities for accelerating scientific innovation are limitless.
By Impact Lab
