Employee Assist Agent: Enhancing Employee Engagement

Employee Assist Agent: Enhancing Employee Engagement

An Insight into the Development and Implementation of Employee Assist Agent.

Problem Statement

In large organizations with many departments, answering employee questions can be quite challenging. Departments like HR, IT, Finance, and Supply Chain each have their own rules and documents, often in different languages. The old ways of dealing with questions were slow and often led to mistakes. There was a strong need for a single system that could answer questions quickly and correctly while passing unresolved issues to the right team.

How We Planned It

To address the problem, we created an AI assistant called “Employee Assist Agent” using Microsoft Copilot Studio. The main goal was to develop a chatbot that could connect to various SharePoint sites containing documents and policies from different departments like HR, IT, Finance, Supply Chain, and more. These sites served as information sources for the chatbot to answer employee questions related to department-specific policies.

During the planning phase, we focused on:

• Identifying sources of information and linking them to Employee Assist Agent.
• Setting up automated workflows with Power Automate for tasks like IT tickets, HR queries, and ordering laptops or desktops.
• Adding a ticketing system using BMC Ticketing Software for questions the chatbot couldn’t solve.
• Using generative actions to understand employee needs and recommend suitable products based on their roles.

 

Implementation

The implementation phase was systematically divided into several key steps:

• Connecting SharePoint Sites : We began by linking numerous SharePoint sites to Employee Assist Agent, each containing department-specific policy documents in various languages. This extensive repository allowed Employee Assist Agent to access information quickly and provide precise answers to employee queries.

• Integration with BMC Ticketing Software : One of the standout features of Employee Assist Agent was its ability to raise tickets in the BMC Ticketing Software automatically. If Employee Assist Agent was unable to resolve a query, it would guide the user to submit a ticket, capturing the necessary input and raising it directly through Power Automate flows.

• Product Recommendations : Employee Assist Agent was designed to assist employees in selecting and ordering the appropriate laptop or desktop based on their role and requirements. By leveraging generative actions, Employee Assist Agent could understand the user’s needs and suggest the most suitable products.

• Power Automate Flows : We developed specific Power Automate flows for different types of requests, ensuring that tickets and purchase orders were raised accurately and efficiently. This automation streamlined the process, reducing the time and effort required for manual interventions.

• Development Structure : Below are the components utilized in developing this copilot:

• Instruction to Copilot : Instructions are essential for guiding a copilot on how to respond to user inquiries. They provide direction on the appropriate actions and considerations necessary when preparing a response to user queries. Here is the below screen shot of Instruction which we provided to this copilot.

 

Setup Knowledge Sources

Knowledge Sources is the section where we compile a knowledge base for the copilot. We have integrated all relevant SharePoint sites into this copilot knowledge base. Below is a screenshot illustrating a list of multiple sites that contain numerous documents within their libraries, representing various departments.

Setup Topics Triggers

Topics are essential for developing the copilot. We use Topics to define triggers or instructions for different query types, with separate topics for each department like HR, IT, Finance, Supply Chain, etc. Each topic has a generative action that understands user queries, filters the relevant SharePoint library and documents, and prepares an accurate response. If no data is available, the user is directed to raise a Ticket on BMC, as defined in Action and based on Power Automate flow.

Defined Actions (Power Automate Based)

Actions play a crucial role in the development of Copilot, particularly regarding external system connectivity. We have defined actions based on Power Automate flows, which include multiple processes for raising tickets on the BMC Ticketing system.

Setup Configurations and User Authentications

Authentication and configuration are essential for any application. We have set up Azure Active Directory V2 for user authentication and configured single sign-on with Microsoft Entra ID for seamless team authentication.

Challenges Faced

While the development and implementation of Employee Assist Agent were successful, we encountered several challenges:

• Data Integration : Ensuring seamless integration of data from various SharePoint sites, each with a different set of policies and documents, was a complex task. We had to ensure that the information was updated regularly and accurately fetched by Employee Assist Agent.
• Language Translation : Handling multiple languages required robust translation features. We had to ensure that Employee Assist Agent could either provide translations of the retrieved articles or offer a translation feature, making the information accessible to all employees.
• Resolutions : We overcame these challenges through innovative solutions and collaborative efforts:
• Enhanced Data Integration : We developed advanced data integration techniques to ensure that SharePoint sites were seamlessly connected, and information was accurately retrieved, providing reliable responses to employee queries.
• Robust Translation Features : By integrating powerful translation services, we ensured that Employee Assist Agent could handle multiple languages effectively, making the chatbot more inclusive and user-friendly.