Potion

Market position

Positioned as an AI-native software provider for beauty and personal care R&D, addressing workflows where ingredient data sits in PDF documents, regulatory requirements are complex and changing, and formulation is largely trial-and-error. The company describes its ingredient library as first-of-its-kind and its formulation workspace as containing the industry's first co-pilot.

Domain-specific AI models trained on beauty and personal care data rather than general-purpose tooling; combination of ingredient discovery, INCI reverse engineering, document-based ingredient management, formulation generation and compliance automation in one platform; private-cloud deployment with SOC 2 Type II certification, customer data ownership and the option to train customer-specific private models.

Technology

Potion trains proprietary AI models specialised for beauty and personal care. Components include a natural-language search layer over a raw-material database of more than 80,000 ingredients; a reverse-engineering model that infers likely raw materials, including specialty blends, and predicts concentration ranges from a published INCI list; document-ingestion capabilities that extract and structure information from ingredient PDFs into an Ingredient Vault; a generative-AI formulation workspace with a co-pilot; and agentic AI that executes routine compliance checks. Customers can train custom private models on their own data. The company reports SOC 2 Type II certification and data encryption. The founders are described as having developed a patented AI system for converting development briefs into starter formulations.

Go-to-market

Self-serve signup for the free Discovery Platform funnels users toward demo requests and premium private-cloud plans; the company also uses a bi-weekly newsletter on beauty product launches, customer testimonials, trade-press coverage in beauty industry media, and LinkedIn as its listed social channel.

Cosmetic chemists, formulators and product development teams at beauty and personal care brands and contract manufacturers, plus procurement professionals and ingredient/chemical suppliers who use the tools to track product launches, market trends and use of specialty chemicals. Named users in testimonials come from organisations including Goodier Cosmetics, Olaplex, THG Labs, Herrco Cosmetics, Rebrand Skincare, Much Love Cosmetic Labs, RNA Corporation, Evolv R&D and Luluble.

Geography

Headquartered in San Francisco, California, per its Y Combinator profile; users quoted in testimonials are based at companies in multiple markets, and the product is delivered as web software.

History

Potion was founded in 2019 and participated in Y Combinator's Summer 2021 batch. The company says it has been training proprietary AI models focused on the beauty and personal care sector for several years, and by the time of a 2024-era industry write-up had released free AI-powered ingredient search and INCI benchmarking tools. It has since added paid modules for ingredient document management, generative-AI formulation and agentic compliance.

Risks & controversies

Potion operates in a market with several similarly named unrelated companies, which can complicate brand identification. Its compliance and formulation outputs are AI-generated: the company describes reverse-engineered formulas as starting formulations and predicted concentration ranges, indicating outputs require professional validation.