At a Halloween party in 2023, Cristian Ponce donned an Indiana Jones costume and met Théo Schäfer, a fellow entrepreneur with a background in underwater autonomous robotics from MIT. The event was organised by Entrepreneur First, a startup programme designed to connect aspiring founders. Their shared experiences of academic life led to the creation of Tetsuwan Scientific, a company determined to streamline laboratory processes that currently rely heavily on human intervention.
Ponce, formerly of CalTech, recalls the laborious tasks faced by lab technicians, particularly in the realm of genetic engineering. Traditional methods often require technicians to manually manipulate liquids with precision instruments like pipettes. Despite various attempts to automate these processes, existing solutions are frequently hindered by high costs, specialised programming skills, and a lack of flexibility, which hampers scientists who frequently change experiment parameters.
The breakthrough for Tetsuwan Scientific came in May 2024, during OpenAI’s multi-model product launch. During the demonstration, Ponce was inspired by the capabilities of large language models (LLMs) in scientific reasoning. He noted their potential when he tested GPT-4 with an image of a DNA gel, resulting in not only an accurate identification of an unintended DNA fragment but also detailed suggestions on mitigating the observed issue, indicating an ability to diagnose scientific outcomes.
However, Ponce observed a significant gap in the automation process: “That robot doesn’t have the context to know," he explained. Robots lack the understanding required to interpret the physical characteristics of liquids and adjust their actions accordingly, which is fundamental in laboratory settings where nuances matter greatly.
The company has been actively addressing these challenges by developing lower-cost, white-label lab robots equipped with new software to interpret scientific intent, enabling automated and intelligent execution of lab tasks. Current iterations of their robotic systems are designed in a minimalist cube shape, focusing on integrating a range of sensors to evaluate results and make modifications autonomously, akin to human scientists.
Tetsuwan Scientific has already made strides in the industry with an alpha customer, La Jolla Labs, a biotech firm invested in RNA therapeutic drugs, where their robots are playing a key role in dosage measurement and efficacy evaluation. The business has successfully raised $2.7 million in an oversubscribed pre-seed funding round led by 2048 Ventures, welcoming contributions from Carbon Silicon, Everywhere Ventures, and notable biotech angel investors.
Ponce envisions a future where scientific inquiry is revolutionised through these technologies: “It is the craziest thing that we could possibly work on. Any technology that automates the scientific method will serve as a catalyst to hyperbolic growth,” he remarked, illustrating the substantial potential impact of AI on scientific discovery. The rise of independent AI scientists, capable of managing the entire scientific method from problem formulation to repeated testing, remains a key aspiration for Ponce and Schäfer.
Their exploration of AI in scientific research aligns with broader industry trends seen in various other endeavours, including those from non-profit organisations and tech start-ups focused on similar objectives. As the field matures, Tetsuwan Scientific may play a pivotal role in shaping the future of laboratory automation and scientific methodology.
Source: Noah Wire Services