Onepot has emerged from stealth with a $13 million funding round and an ambition to redesign how small molecules are created. The company has built an automated synthesis lab powered by Phil, an AI organic chemist that plans, runs, and learns from real experiments. Early commercial partners are already receiving new molecules six to ten times faster, addressing one of the most stubborn chokepoints in drug and materials discovery.
Addressing the Hidden Bottleneck in Chemistry
Every small molecule based drug, material, or agrochemical begins with the same challenge, which is making a compound that has never existed before. Today many pharmaceutical companies still wait six to twelve weeks for a small batch of candidate molecules, with little transfer of know-how between projects. Synthetic routes are often based on decades old literature behind paywalls, limited reaction scopes, and almost no visibility into failed experiments, which together drive high failure rates and wasted effort.
Building an Automated Synthesis Facility
To break this pattern, onepot rethought the entire compound synthesis cycle, from reagent supply to storage, reaction setup, workup, purification, and quality control. The company uses liquid handlers, plate sealers, and plate based reactions, and has even deployed robotics inside a glovebox to handle sensitive chemistries. This hardware infrastructure is designed to run many reactions in parallel, but the team recognized that true impact required automating the decisions that govern those experiments.
Phil, the AI Chemist in the Loop
Phil, onepot's AI chemist, is given direct control over the lab environment instead of being limited to passive recommendations. The system can trigger tools like a plate sealer, write liquid handling protocols, interpret LC/MS data, identify byproducts, generate hypotheses, and immediately validate them through new experiments. In just a few weeks, Phil has onboarded dozens of reaction types, optimized conditions, and made five core transformations - including key couplings and acylations - available to paying customers, with more to follow.
Turning Experimental Data into Chemical Insight
Phil has already run more reactions in a single month than many graduate students complete across an entire PhD, and each run is captured as structured data. Onepot combines these in-house datasets, spanning both large scale pretraining style information and higher quality analytical traces, with models trained to predict which protocols will work on new substrates. Internal benchmarks show that for difficult reactions, the system's predictive power can rival or exceed seasoned organic chemists, and an AI driven method construction engine has already raised the typical purity levels delivered to customers.
Toward Instantaneous Molecular Discovery
The company believes that waiting weeks for synthesis results will soon look as outdated as batch computing queues from the early days of software. In the vision promoted by onepot, every reaction feeds into a continuously improving model, routes are adapted in real time, and chemical intuition becomes a resource that can be stored, scaled, and improved by machines. This could enable drug discovery teams, materials scientists, and chemical innovators to iterate on ideas in days rather than months, and potentially cut both timelines and costs for new therapies and products.
Backing, Hiring Plans, and Commercial Onboarding
The $13 million round is led by Fifty Years, Khosla Ventures, Speedinvest, and Script Capital, with participation from angel backers such as Agata and Wojciech Zaremba, Jeff Dean, NAVEC, and other supporters focused on deep tech and AI. Onepot is now recruiting chemists, software engineers, and robotics specialists who want to treat molecules as programmable objects and help scale the platform. At the same time, the company is opening a waitlist for organizations in pharmaceuticals, advanced materials, and related fields that want to compress molecular discovery timelines by an order of magnitude.
By combining a highly automated lab with an AI chemist that can act, observe, and learn from every experiment, onepot is attempting to redefine what is possible in small molecule synthesis. The company positions itself at the intersection of robotics, data, and chemistry, targeting a bottleneck that has quietly constrained innovation for decades. If its model scales as promised, the path from molecular idea to physical compound could become dramatically shorter, reshaping how new drugs and materials are brought into the world.