Tetsuwan Scientific

Tetsuwan Scientific develops software and robotic systems intended to let researchers define and run wet-lab biology experiments without physical lab access. Its platform, ResearchOS, accepts experimental procedures expressed as uploaded files (research papers, kit manuals, informal notes) or natural-language descriptions and renders them into a visual protocol format. Users can parameterize any element of a procedure — reagents, sample numbers, dyes, treatments, instrument parameters, volumes and concentrations — either by uploading a CSV of existing plates or by querying the system in natural language. The platform then compiles the procedure and its variables into an automation script executable on a robot, and offers simulation so users can visually verify a protocol before running it. Low-level automation decisions such as liquid classes are handled automatically but remain viewable and editable.

Earlier public coverage describes the company as building "autonomous AI scientists": systems combining large language model reasoning with laboratory robotics that can evaluate experimental results and modify experiments on their own. Rather than building proprietary hardware, the company modifies lower-cost white-label lab robots and adds software and sensors to handle calibration, liquid class characterization and other physical properties. Reported robot form factor is a square glass cube rather than a humanoid. The company positions itself as adding intelligence to lab automation rather than dexterity, targeting the interface between scientists, existing lab automation tools and AI models.

The company's stated long-term aim is automating the scientific method end to end, from hypothesis generation through repeatable results, and building infrastructure described by one third-party profile as a virtual research laboratory.

Founding story

Cristian Ponce and Théo Schäfer met at a Halloween party in 2023 hosted by Entrepreneur First's U.S. program, where Ponce attended in an Indiana Jones costume. Ponce began college at 16 studying computer science at UNC Charlotte before moving into bioengineering at Caltech, working on E. coli genomics in the Wang Lab and on full-genome engineering; Schäfer holds a master's in mechanical engineering from ETH Zürich with thesis work at MIT on underwater autonomous robots, and previously built autonomous robots for extreme environments at NASA's Jet Propulsion Laboratory, including work related to exploring Jupiter's moons. The pair bonded over the drudgery of manual pipetting and lab technician work. The founding thesis initially centered on modifying lower-cost white-label lab robots; in May 2024, after watching OpenAI's multimodal product launch, Ponce tested GPT-4 on an image of a DNA gel and found it correctly identified a primer dimer and proposed conditions to prevent it, which the founders describe as the moment they concluded LLMs had scientific reasoning but no physical agency.

Business model

Tetsuwan offers a cloud lab service and automation software: customers submit experimental questions and protocols through the ResearchOS browser platform, and experiments are executed on Tetsuwan-operated robotic instrumentation, with results returned to the researcher. The company modifies lower-cost white-label lab robots rather than manufacturing bespoke hardware, which is described as reducing capital requirements and speeding deployment. Sources do not disclose pricing or contract structure.

Not disclosed in the available sources.

Traction

As of late 2024 the company reported an alpha customer (La Jolla Labs) and its first autonomous AI system operational in a rare disease gene therapeutics lab, with a team that had grown to seven people drawn from Caltech, ETH Zurich, MIT and UC Davis. Reported cumulative funding is $5.7M across two rounds per a third-party database, of which a $2.7M pre-seed is corroborated by multiple sources. No revenue, customer count or usage metrics are disclosed.

Latest developments

The company website, current as of 2026, presents ResearchOS as a browser-accessible cloud lab where researchers configure, simulate and run experiments without entering a lab. Blog activity in early 2026 covers cross-platform integration with PyLabRobot and UniteLabs, use of Claude in the wet lab, and a recap of an Autonomous Science Night event.