We immediately knew generative AI could achieve the speed and scale required for this Herculean task.
Our two main goals were to:
- Consolidate the data: We wanted to compile the data from all document types into one accessible and understandable format
- Cleanse and enrich the data using AI: To ensure the accuracy of information, we had to guide the AI algorithms and eliminate any false data
The first step was to use advanced machine learning algorithms and natural language processing to extract data from various source documents. Next, we employed optical character recognition (OCR) and machine learning models to extract information from image-based documents. And finally, to enrich the data, we trained a large language model (LLM) using PaLM 2, a next-generation LLM from Google.
The implemented generative AI tool began swiftly collecting and cross-referencing data like never before. What's more, to maintain the integrity of the information, we conducted thorough manual checks using a human-in-the-loop (HIL) approach. These checks ensured the LLM models worked as expected and the data was always reliable.
By harnessing the power of generative AI, we improved the part-ordering process. The combination of advanced algorithms, OCR, machine learning models, and manual checks allowed us to achieve the necessary speed and accuracy in processing data, ultimately leading to a successful outcome for our client.