Our Head of Technology, Josh Powley, has being working on leveraging AI to convert unstructured data into structured data. He has recently made significant progress in addressing a challenge related to processing customer sales orders received as natural language emails.
Typically, these orders pose difficulties for conventional intelligent document processing (IDP) tools, as comprehending and transforming them into actionable transactions requires sophisticated interpretation capabilities. However, when orders are received as email attachments generated by an Enterprise Resource Planning (ERP) system, IDP tools can effectively interpret, handle exceptions, and upload the order to the ERP. The challenge arises when dealing with natural language processing.
To overcome this hurdle, Josh has been actively testing a process that follows these steps:
1. The customer sends an email to the supplier.
2. Our tool extracts the text from the email and generates a set of prompts for AI.
3. The AI, based on the provided prompts, converts the text into structured tabular data.
4. The AI generates a PDF order acknowledgement and sends it to the customer.
5. Additionally, our process automation tool creates the corresponding transaction in the ERP system.
For instance, if the email content is in the following format:
“Please could we order x 26 brown cedar fencing boards @ £10.95.
On PO number ‘G00722212B00019’ for delivery to GAP Deeside on 15th Dec.”
Our AI processes the text and generates tabular data similar to the following:
I’m looking forward to seeing how this innovation will enhance our ability to streamline order processing.