Automating regulatory and supplier diligence for European cosmetic brands — 75% saved per audit

The regulatory burden on European brand and ingredient manufacturers has grown faster than the teams that handle it. A single product launch can require evidence packages spanning REACH, CLP, food contact regulations, allergen disclosures, supplier provenance, and dozens of customer-specific compliance forms. Each one is assembled by hand from documents scattered across email, supplier portals, and ERP systems.

Our partner — an emerging European platform serving brands, chemicals manufacturers, and ingredient suppliers — saw this work bottlenecking a whole sector. Sales teams were spending whole afternoons answering customer regulatory inquiries instead of selling. R&D was waiting weeks on supplier data to finalise formulations. Procurement was chasing the same vendor information quarter after quarter. The cost wasn't only inefficiency. It was deals lost to faster competitors and product launches delayed by paperwork.

They came to us with the platform concept and a small founding team, and asked us to build it. The pitch to their customers was direct: automate the parts of the regulatory workflow that don't need a human, and give the human team back its afternoons.

What we learned

We built the platform from day one as an operating system for the brand-and-ingredient supply chain, not as a tool that bolts onto an existing one. That meant designing the data model first — a single source of truth covering raw materials, suppliers, formulations, regulatory documents, and customer-specific compliance requirements. Once that foundation existed, the AI agents had something coherent to reason over.

The agent suite ships in production today. A regulatory inquiry agent reads inbound customer questions and assembles the answer from the platform's document store and the supplier data attached to each ingredient. It cites sources and flags anything that needs a human. A supplier diligence agent automatically collects, verifies, and refreshes vendor documentation, so procurement stops chasing the same forms every quarter. A formulation agent helps R&D build to spec by surfacing compliant raw material combinations from the catalogue. A procurement agent handles routine purchase order workflows with suppliers.

Each agent runs against the same shared platform model, which is why they compose. When a customer asks a regulatory question, the inquiry agent can pull supplier data the diligence agent maintains and reference formulations the R&D team is working on. The agents are independently useful and dramatically more useful together — which is the architecture decision that turned a feature roadmap into a product.

The stack is OpenAI for the language layer, PostgreSQL for the platform model, AWS for hosting and orchestration, and Next.js for the customer-facing application. Models sit behind a provider abstraction so we can swap them as the LLM market shifts. It has shifted twice already during this engagement.

What this shaped

Customer regulatory inquiries that used to take a sales or compliance rep three to four hours now resolve in under an hour, and a meaningful share never touch a human at all. The work that used to anchor a sales rep's afternoon now happens before they read their first email of the day.

Adoption tells the same story. The platform is in production at three of the top ten European chemical brands, with several more in pilot. Revenue has grown roughly 3x in the year following launch, and the customer success motion has shifted from "selling the platform" to "onboarding the next agent."

What this proved