Most cosmetic formulation is iteration on formulas a lab has already made. The knowledge is there; it is just in old batch records, supplier PDFs and one chemist’s memory.
AI cosmetic formulation means using models to search and reason over formulation data: past formulas and their stability results, supplier technical data, INCI names and the restricted-substance annexes. It proposes starting points and screens candidates; the chemist designs, the bench tests, and a qualified safety assessor signs the safety report.
The first gain is usually retrieval, not generation. A lab with years of formulas can rarely answer which past emulsions were stable with this emollient at this pH, or why a fragrance was dropped. We make that history searchable, with every answer citing the batch record or test report it came from.
The second is screening. Every candidate is checked against the Annexes of Regulation (EC) 1223/2009 (prohibited substances in Annex II, restricted ones in Annex III, and the permitted colourants, preservatives and UV filters in Annexes IV to VI), and against your own internal restrictions and market lists. The system flags, with the entry and limit cited; it does not decide.
texture, claim, cost, market
similar past formulas
starting points, cited
Annex II to VI, internal lists
chemist decides what to make
held with entry and limit
It manages formulas, raw materials and their data: percentages, INCI lists, costs, regulatory limits and labelling. Many labs use a dedicated formulation system or spreadsheets for this. AI adds search across the history in those records and documents, proposes starting points, and screens candidates against regulatory limits, working alongside the system you already use rather than replacing it.
It can propose one, usually by adapting formulas that have worked before for a similar texture or claim. Whether it is stable, preserved, compatible with its packaging and safe is established by testing and by a qualified safety assessor. We treat every suggestion as a hypothesis that cites the data it is based on, so the chemist can judge it quickly.
Not in what we build. Formulas are among a beauty company’s most valuable assets, so models run on infrastructure you control, with no training on your data, and the formula library stays in your own storage. Where a hosted model is used at all, it sits behind terms that prohibit training and retention.
What we built, what stage it is at, and a link to the live thing wherever there is one.
yourOS is an operating system with an AI at its heart. You talk to it like a person, out loud or typed. It makes your calls, plans your days and works on your projects with a crew of agents, on your own private server. Our own product, built from the first commit; the waitlist is open.
Website, backend and app assets for Sonora, a free AI sound-therapy app on iOS and Android published by Cognitive AI.
Brand, website, prototype, apps on five platforms, the AI backend and the algorithm-intelligence pipeline for Podswap, a social growth engine that runs without a human in the loop.
In closed beta: Roxy, a private AI assistant with permanent memory running on infrastructure the user owns. Brand, site, product, backend and conversational design.
Live at coherence.com: brand identity generated with AI agents we directed, the motion system and website for the group behind Sonora, AboutAI, WakeUp and MudShadows. Agent work remains in build.
Live at domainbooth.com for DomainBooth, a premium domain brokerage: a new animated mark, every page redesigned and rebuilt with designed mobile layouts, and a custom CMS for its sales record, featured names, press and enquiries.
We start with a sample of your formula history and show what a cited search over it looks like.
30-minute technical call · no deck