"Along with Dribia, we have developed a project that uses AI to improve processes in the quality department. This will not only allow us to improve deviation management but also open the door to further optimizing our processes using AI and data science".
FEATURES
Generative AI assistant to support the quality team during deviation management.
IMPACT
Improve deviations management efficiency, quality of documentation information, and reduce the number of deviations, especially recurring ones.
ACHIEVEMENTS
Creation of a tool that allows expanding the knowledge of deviations and their associated investigations and CAPAs, between the quality team and the operators.
BUSINESS PROBLEM
At Cloverty, detecting deviations in the production process involves manually creating detailed documents and action plans without structured access to information from previous plans. Lacking tools to analyze historical data—currently scattered across PDFs with heterogeneous formats and TrackWise—considerable time is spent searching for and understanding the background information. This makes the process critically dependent on the expert knowledge and individual experience of each employee, leading to disparities in resolution depending on who manages the incident. This lack of access to information hinders deviation tracking and could lead to recurring incidents and a limited understanding of the frequency, type, and correct implementation of action plans.
DRIBIA’S SOLUTION
Together with the Cloverty quality team, a tool based on generative artificial intelligence has been designed to support the analysis of past deviations and the generation of documentation for new ones. On the one hand, a preprocessing process has been developed that extracts information from existing documents, structures it into relevant and standardized fields, classifies the type of deviation, and stores this data in a centralized knowledge base. On the other hand, a conversational assistant has been created that leverages all this structured information to answer specific questions about historical data and assist in writing new deviations and CAPAs, based on previous solutions, past descriptions, and factory protocols.
