Different vertical sectors come with their own unique requirements. In this post, we explore how large language models (LLMs) can play a transformative role in education settings by streamlining the Purchase-to-Pay process.
The Purchase-to-Pay journey often begins with a requisition, where users raise requests for purchases. Before these requests can be approved and converted into orders, an appropriate Nominal Code must be assigned. This task is frequently performed at the requisition stage by users who may have limited finance knowledge and are unfamiliar with the required nominal codes.
By leveraging previously coded orders, requisition line-item descriptions, and the power of LLMs, we can:
- Auto-code requisitions or, at the very least, present a reduced list of suitable codes for users to select from.
- Simplify the coding process, even when line descriptions don’t exactly match previously used data.
Let’s consider a simple example. A user raises a requisition for cleaning services:
Using a web service call, we send the line-level descriptions and historical coded requisition data to the LLM. The LLM processes this information and returns a list of matched nominal codes:
In this response, users can clearly see the match reasons and the accuracy of the suggestions. This approach not only simplifies the requisition process for users but also increases the accuracy of the nominal coding step, ensuring compliance and consistency.
This is a prime example of how technology—specifically large language models—can bridge the gap between finance and non-finance users, making complex processes easier and more efficient.