12-06-2024

Case AWVN

AI Assistant Implementation at AWVN

AWVN is a leading employers' organization in the Netherlands, involved in over half of all collective labor agreements (CAOs) in the country. A significant part of their role includes providing job descriptions using their advanced, self-developed ORBA analysis method. This process is time-consuming and involves extensive writing. An efficient AI application could offer a solution.

Challenge

AWVN had the necessary knowledge and data to implement an intelligent chatbot but needed a bridge between the available data and its effective utilization. Student Consultant was tasked with: 1) conducting an in-depth analysis of the possibilities for data use and chatbot implementation, and 2) executing the necessary steps to achieve this implementation.

Our Approach

Our student consultants began by creating a research report, exploring the potential for implementing a GPT model as a chatbot and understanding the needs of the staff. They also assessed the availability, quality, and accessibility of relevant training data. Additionally, they investigated the technical setup required for a chatbot, identifying the steps needed and the most promising options.

The implementation process involved:

• Building the architecture within Microsoft Azure.
• Creating system instructions for the GPT model.
• Developing a PowerAutomate flow to translate the AI assistant's output into a standardized Word template.
• Writing operating instructions for AWVN advisors.

Outcome

The evaluation revealed that the AI assistant reduced the time required to draft a job description proposal by 30%, and the created descriptions received high ratings. Given the nature of the GPT model, its performance is expected to improve with increased use. AWVN advisors now have an additional tool for generating job descriptions efficiently.