Startups doing autonomous scientific experimentation (self-driving labs)

1. Arctoris — Operates a robotic, cloud-based drug-discovery laboratory that runs cellular and molecular experiments continuously and captures the resulting data in a structured platform. Why it fits: Founded in 2016 and based in Oxford; it has raised venture funding and is one of the clearest early examples of an end-to-end autonomous drug-discovery lab.
2. Strateos — Provides remotely operated robotic laboratories for biology and chemistry, allowing customers to specify workflows that are executed and monitored automatically. Why it fits: A real, venture-backed successor to Transcriptic; its cloud-lab infrastructure is already used for repeatable, highly automated experimentation rather than merely instrument control.
3. Emerald Cloud Lab — Offers a large centralized laboratory in which users program experiments across hundreds of instruments and receive standardized data without physically operating the lab. Why it fits: Commercially deployed and backed by substantial institutional capital; it is more mature than a typical seed startup, but is a foundational platform for self-driving laboratory workflows.
4. Kebotix — Combines machine learning, materials databases, and automated laboratory execution to discover and optimize new materials. Why it fits: Founded in 2018 and venture-backed; its explicit “closed-loop” materials-discovery model connects prediction, synthesis, testing, and model updating.
5. Chemify — Develops AI-guided chemistry software and automated chemical laboratories intended to plan, execute, and iterate synthesis autonomously. Why it fits: UK-based and founded in 2019; it has attracted significant backing and is pursuing a broad “autonomous chemistry” platform rather than a single therapeutic program.
6. CuspAI — Uses generative AI and automated experimentation to discover materials with specified properties, including materials relevant to sustainability and industrial applications. Why it fits: A young Cambridge-based company that announced major early funding and partnerships; its model is explicitly oriented toward AI-directed, experimentally validated materials discovery.
7. Lila Sciences — Builds “physical intelligence” systems that combine AI scientists, robotics, and laboratory infrastructure to run scientific experiments autonomously. Why it fits: An unusually early and well-capitalized entrant, launched with Flagship Pioneering backing; its stated goal is broad autonomous experimentation across biology, chemistry, and materials science.
8. LabGenius — Uses robotic experimentation, machine learning, and evolutionary design loops to engineer therapeutic proteins such as antibodies. Why it fits: Founded in 2012 and backed by major life-science investors; although more advanced than a seed-stage company, its core platform is a genuine closed-loop autonomous experimentation system.
9. Culture Biosciences — Provides automated, remotely managed bioreactor experimentation for optimizing microbial and cell-culture processes, with software controlling experiments and analyzing process data. Why it fits: Founded in 2018 and venture-backed; it is particularly relevant to the self-driving-lab thesis in bioprocess development, where experiment execution and feedback can be tightly integrated.
10. Atinary Technologies — Provides AI software for experiment design, Bayesian optimization, and laboratory workflow orchestration across chemistry, materials, and process development. Why it fits: A relatively young, venture-backed company founded in 2019; while it is more software-led than the companies operating full robotic labs, it supplies the decision-making layer required to turn automated equipment into a self-driving laboratory.

White-space gaps noticed

• Autonomous scale-up and manufacturing transfer: Most platforms optimize discovery experiments at small scale; there is still limited coverage of the handoff from optimized lab recipes to pilot-scale synthesis, fermentation, formulation, and process control.
• Affordable self-driving labs for ordinary research groups: Many systems require centralized facilities, expensive robotics, or enterprise-scale deployments. A modular, lower-cost platform that could serve university labs and small industrial R&D teams remains relatively underdeveloped.