Streamline supplier verification with Document Intelligence

The Enterprise Data & Analytics team at a major US utilities company faced a daunting task. They needed to streamline the manual verification process of Certificates of Insurance (COI) for over 2,500 suppliers and vendors. These COIs arrived in multiple file formats, including PDF, JPG, and PNG, creating a complex and time-consuming workflow.

The client required a solution to automate the verification process, ensuring that each vendor had a valid COI or needed to update their COI. Additionally, the system had to record which forms had been checked by humans versus those verified through automation for audit purposes.

The team also saw this as an opportunity to test Microsoft Azure Cognitive Services and OpenAI, aiming to leverage these technologies to support their operations in the future.

To address these challenges, Fractal developed a comprehensive landing zone architecture to consolidate all COIs into a single location. This centralized repository allowed for automatic analysis and easy retrieval by staff for human verification.

Leveraging one of the Microsoft Azure Cognitive Services known as Document Intelligence (formerly Form Recognizer), the solution could extract visual data from the COIs and analyze the information efficiently. This approach not only streamlined the verification process but also paved the way for further automation and enhanced data management capabilities.

The implementation of this solution led to the creation of a Minimum Viable Product (MVP) that successfully extracts and analyzes data from COI forms. This initial success marks a significant milestone, demonstrating the potential for further advancements.

The MVP serves as a foundational stage before introducing a ChatGPT interface, which will enable users to search and interact with forms more intuitively. The major utilities company has thus taken a crucial step towards more efficient supplier verification, leveraging cutting-edge technology to improve their operations.