Technology

Tetsuwan Scientific Develops Autonomous Robotic Scientists Capable of Conducting Independent Experiments

The Evolution of Scientific Discovery: How Tetsuwan Scientific is Revolutionizing the Field with AI-Powered Robots

In 2023, a chance encounter between Cristian Ponce and Théo Schäfer at an Entrepreneur First Halloween party set in motion a chain of events that would change the face of scientific discovery forever. The two co-founders, clad in their Indiana Jones costumes, hit it off immediately, bonding over their shared experiences as lab technicians.

From Lab Technicians to AI-Powered Robots

Ponce and Schäfer’s conversation revolved around the drudgery of manual labor involved in genetic engineering. As they commiserated about the tedium of using scientific syringes (pipettes) to manually move liquids from tube to tube, an idea began to take shape.

Attempts to automate this process had been met with limited success due to the specialized, expensive robots required for the task. These robots demanded special programming skills, and every time scientists needed to adjust experiment parameters, they would have to wait for a programmer to reprogram the bot, debug it, and so on.

The lab techs’ discussions sparked an epiphany: what if there were a way to modify lower-cost white label lab robots to simplify this process?

Tetsuwan Scientific: A Solution in the Making

Thus, Tetsuwan Scientific was born. The company aimed to address the problem by adapting lab robots to make them more accessible and user-friendly for scientists.

However, their journey took an unexpected turn when they watched OpenAI’s multi-model product launch in May 2024. During the demo, Ponce had a lightbulb moment. He fired up GPT-4 and showed it an image of a DNA gel. The model not only successfully interpreted the image but also identified a problem – an unintended DNA fragment known as a ‘primer dimer.’

The Power of LLMs: Unlocking Scientific Potential

Ponce described this experience as a "light bulb moment," where Large Language Models (LLMs) demonstrated their capacity for diagnosing scientific outputs. However, he noted that these models lacked physical agency to perform the suggested actions.

This revelation marked a turning point for Tetsuwan Scientific. They realized that their initial approach – modifying lab robots – was only half the solution. To truly revolutionize scientific discovery, they needed to develop software that could translate scientific intent into robotic execution.

The Birth of Robotic AI Scientists

Tetsuwan’s research showed that no existing software existed to bridge this gap. The robot had no context to understand the physical qualities of liquids it was handling – whether they were viscous or prone to crystallization. To overcome this challenge, the team began building audio LLMs with reduced hallucinations (RAG) capabilities.

These AI-powered robots would be able to work with complex tasks that are difficult to code manually.

The Future of Scientific Discovery: AI-Powered Robots

Tetsuwan’s innovative approach has sparked interest in the scientific community. Other organizations, such as non-profit org FutureHouse and Seattle-based Potato AI, are also exploring the potential of AI scientists.

As we move forward into an era where AI-powered robots take center stage, one thing is clear: the future of scientific discovery will be faster, more efficient, and potentially groundbreaking.

Key Players in the AI-Powered Robot Revolution