Tech

Artwork in PDF format is converted into images and undergoes a preprocessing stage. Our proprietary Layout Analysis model then precisely evaluates the location and hierarchical structure of components, including text, symbols, and tables. Through this process, we generate optimized input data tailored for regulatory text review, taking both visual arrangement and structural information into full account.

We perform Optical Character Recognition (OCR) based on the coordinates of the text areas extracted from the artwork. The text retrieved through this OCR process is then utilized as the direct input for our Large Language Model (LLM), which is specifically fine-tuned for the cosmetics domain.

CDRI's inference engine is an agentic AI powered by a Large Language Model specialized in cosmetics certification. It comprehensively analyzes complex information, combining both text and layout data. This enables the engine to accurately determine the regulatory compliance of the wording, effectively supporting the review and refinement of your certification documents.

CertiPT is a comprehensive solution that leverages LLM and multi-agent technologies to classify and analyze essential certification source documents within a highly secure private cloud. Multiple hyper-scale AI models collaborate to review these documents, proactively detecting typos and logical errors. These insights are reported to the compliance manager, enabling faster and more precise document processing. Furthermore, CertiPT suggests actionable next steps to accelerate decision-making and systematically manages task delegation and feedback histories for the team.

We have developed a systematic, centralized database for cosmetics ingredients. It consolidates ingredient data used globally, integrating sources such as the ICID, the Korean Cosmetic Ingredient Dictionary, and CosIng. This system allows you to easily manage vast amounts of data, including various ingredient names, CAS Numbers, and functions. Additionally, it intelligently interprets regulations across different countries and agencies, enabling the highly efficient management of regulatory information specific to ingredients, pigments, allergens, and product types.