Problem

Drug discovery is constrained by slow, expensive chemical synthesis. b12 Labs’ YC materials describe new-drug discovery as taking more than 10 years and costing over $2 billion, with only about 2 in 10,000 compounds succeeding. The bottleneck is not just discovering promising molecules: chemists must design routes, run experiments, and iterate through trial and error, while software remains fragmented and expensive laboratory robots are often difficult for scientists to use. The central use case is helping pharmaceutical and biotech teams make complex, high-value molecules that are otherwise hard to synthesize, accelerating early-stage discovery from years toward months.

Product / Service

b12 Labs provides an AI-powered chemical-synthesis platform combining software tools with optional execution services. Its products include Selenium for retrosynthesis, which turns a target molecule into ranked, reviewable synthesis strategies; Palladium for designing high-throughput experimentation plates and optimizing reaction conditions; and Cobalt for converting natural-language intent and optimized conditions into execution-ready workflows for systems such as Chemspeed and Opentrons. The platform is designed to let scientists describe what they want in natural language, review the proposed protocol and reasoning, and run experiments without coding.

The delivery model is flexible: customers can use the software on a self-serve basis, run workflows in their own laboratories, or ask b12 to execute chemistry through its synthesis service. b12 says it can handle complex, multi-step targets and deliver results from screening plates through gram-scale synthesis. The intended benefit is to close the loop from synthesis planning to robotic experimentation, reducing manual programming, fragile handoffs, and the number of experimental iterations required to identify workable chemistry.

Market

b12 competes at the intersection of AI-powered drug discovery, chemical-synthesis planning, laboratory-automation software, and synthesis services. Its software overlaps with retrosynthesis tools such as IBM RXN for Chemistry, SYNTHIA, and ASKCOS, while its automation layer is adjacent to platforms and hardware ecosystems including Chemspeed, Opentrons, and ChemOS. b12’s differentiation is its attempt to integrate route planning, experiment design, and physical execution into one natural-language, vendor-agnostic workflow rather than offering only route prediction or only robotics.

The company is an early commercial entrant rather than a purely pre-revenue project: YC identifies it as an active Summer 2025 company and says it was running paid pilots with large pharmaceutical companies. Those pilots reportedly included an experiment plan that achieved full conversion on the first attempt, and b12 later reported up to 99% yield on a first optimization plate with some customers and partners. The evidence supports early enterprise validation and product availability, but does not establish recurring revenue, a large customer base, or mature scale; its primary market opportunity is the expanding laboratory-automation and AI-for-drug-discovery spend of pharmaceutical and biotech R&D teams.