What You Will Learn

• Why HR Copilot agents outperform traditional SharePoint/Wiki searches
• How to create and configure an HR Copilot agent in Microsoft 365
• How to link and manage a SharePoint knowledge source for the agent
• How to test, share, and iterate the agent's prompts and settings
• Limitations, privacy considerations, and future enhancement opportunities

Study Guide

Introduction

Welcome to the comprehensive guide on how to build an HR Copilot AI Agent. In this course, we will walk you through every step necessary to create a custom AI agent designed to enhance Human Resources operations using Microsoft 365 Co-pilot tools. This course is invaluable for organizations looking to streamline HR processes, improve employee experience, and leverage cutting-edge AI technology to provide intuitive and efficient solutions to common HR inquiries. By the end of this guide, you'll have a thorough understanding of the entire process, from conception to implementation, and be equipped with the knowledge to build and maintain a robust HR Copilot AI Agent.

The Value Proposition of HR Copilot Agents

Understanding the Why:
Before delving into the technical details, it's crucial to grasp why an AI agent is beneficial for HR functions compared to traditional resources like SharePoint sites or wikis. The primary advantages are:

• Natural Language Interaction: AI agents allow users to interact using plain language, making the experience more intuitive. Instead of relying on keyword searches, employees can simply type questions as they would naturally speak.
• Improved Search Mechanism: These agents employ advanced search algorithms, providing more accurate and relevant results than traditional search methods.
• Easy Accessibility: The AI agent can be integrated into Microsoft Teams, M365 Copilot.com, or the M365 Co-pilot application, ensuring that it's readily available to employees wherever they are.
• Simple Maintenance and Sharing: Once set up, the agent is easy to maintain and can be shared across the organization or with specific user groups.

These features collectively enhance the HR support experience, making it more efficient and user-friendly.

Functionality and Limitations of the Initial Agent

Capabilities of the HR Copilot Agent:
The initial version of the HR Copilot AI Agent is designed to handle basic HR inquiries by searching a defined knowledge source. Specifically, it:

• Searches a Defined Knowledge Source: The agent primarily searches across a specified SharePoint site containing HR-related documents, such as benefits, vacation policies, and training information.
• Answers First-Level HR Questions: It effectively handles common HR inquiries, providing quick and accurate responses based on the available data.

Current Limitations:
While the initial agent is a powerful tool, it has some limitations:

• Does Not Connect to Other Data Sources: The agent is not designed to integrate with systems that provide real-time, personalized data, such as remaining vacation time or initiating bookings.
• Not an Autonomous Agent: It functions as a "retrieval agent," requiring user initiation rather than proactively triggering actions based on events. Future enhancements could introduce autonomous capabilities using tools like Co-pilot Studio.

Technological Requirements and Building Process

Getting Started with Microsoft 365 Co-pilot:
Building the HR Copilot AI Agent requires access to Microsoft 365 Co-pilot and its agent builder capabilities. The recent introduction of Microsoft 365 Co-pilot Chat offers a cost-effective entry point for building and using agents.

Step-by-Step Guide to Building the Agent:
Here are the key steps involved in creating an HR "onboarding buddy" agent:

1. Accessing Co-pilot: Navigate to M365 copilot.com to begin the process.
2. Initiating Agent Creation: Select the "create an agent" option to start building your agent.
3. Describing the Agent: Use natural language to define the agent's purpose, such as "You're an onboarding agent for our new hires..."
4. Naming the Agent: Provide a clear and relevant name, such as "the onboarding buddy" or "HR Miracle Worker."
5. Refining Instructions: Specify what the agent should emphasize and avoid in its responses. Initially, concise answers are preferred, but this can be adjusted to allow more detailed replies.
6. Defining the Agent's Tone: Choose a suitable communication style, such as friendly and professional.
7. Configuring the Knowledge Source: Link the agent to the relevant SharePoint site and select the specific document libraries containing HR information.
8. Customizing Starter Prompts: Add pre-defined questions to guide users, like "What are my 401k benefits?" or "How many days do I get vacation?"
9. Testing the Agent: Interact with the agent in a separate chat dialogue to ensure it functions as expected.
10. Sharing the Agent: Make the agent accessible to relevant users or the entire company, while being mindful of SharePoint file access permissions.

Importance of Knowledge Source

Ensuring Quality and Accessibility:
The effectiveness of the HR Copilot AI Agent relies heavily on the quality and accessibility of the information within the designated SharePoint site. It's crucial to ensure that intended users have the necessary permissions to access the underlying documents. For example, if certain users lack access to specific SharePoint files or folders, the agent will not use those files when answering questions. Therefore, it's important to share files and folders with all intended agent users to ensure consistent performance.

Iterative Improvement

Continuous Refinement:
The configuration of the HR Copilot AI Agent can be edited and updated after creation, allowing for ongoing refinement and adjustments to instructions, knowledge sources, and other settings. This iterative improvement process ensures that the agent remains effective and relevant as organizational needs evolve.

Potential Opportunities and Future Enhancements

Exploring Future Possibilities:
The HR Copilot AI Agent offers numerous opportunities for future enhancements:

• Integration with Other Data Sources: Future versions could enable the agent to access and utilize information from various HR systems, such as payroll and time-off booking.
• Development of Autonomous Capabilities: Creating agents that can proactively trigger actions or notifications based on specific events is a promising area for future development.
• Building Other Types of Agents: The Microsoft overview guide includes templates for agents like resume reviewers, contract reviewers, and research assistants, suggesting a wide range of potential applications.

Conclusion

Building an HR Copilot AI Agent using Microsoft 365 Co-pilot tools is a practical and accessible way to enhance HR support and improve employee experience. The ease of natural language interaction, improved search capabilities, and straightforward building process present a compelling case for organizations to explore this technology. While the current iteration is primarily a retrieval agent, future development promises more advanced capabilities through integration with other systems and the potential for autonomous actions. By following this comprehensive guide, you are now equipped to create and maintain an HR Copilot AI Agent, ensuring thoughtful application of these skills to meet your organization's unique needs.

Podcast

There'll soon be a podcast available for this course.

Frequently Asked Questions

Introduction

Welcome to the FAQ section for the 'Video Course: How to Build an HR Copilot AI Agent.' This resource is designed to provide clear and concise answers to common questions about creating and implementing an AI agent for Human Resources. Whether you're new to AI or an experienced practitioner, this guide aims to offer valuable insights and practical advice.

Why should we use an AI agent for HR questions instead of our existing resources like SharePoint or a Wiki?

An AI agent offers several advantages over traditional HR resources. Firstly, it allows employees to ask questions using natural language, making the interaction much more intuitive and user-friendly than keyword searches. Secondly, these agents typically employ more sophisticated search mechanisms, leading to more relevant and accurate results compared to standard website searches. Finally, accessibility is greatly improved, as the agent can be integrated into platforms like Microsoft Teams, M365 Copilot.com, and the M365 Copilot application, making it readily available to employees wherever they are. Additionally, once set up, the agent is designed to be easy to maintain and share across the organisation or specific user groups.

What kind of HR questions can this AI agent actually answer?

The primary function of this AI agent is to answer first-level Human Resources queries by searching across a designated knowledge source, which in this example is a SharePoint site containing various HR-related documents. This repository can include information on topics such as employee benefits, vacation policies, and training programmes. Therefore, the agent can effectively respond to common questions related to the information contained within these documents.

What are the current limitations of this HR AI agent?

Currently, this specific AI agent has a few key limitations. It is not designed to connect to other data sources beyond the specified knowledge base. For instance, it cannot directly access an employee's vacation balance or initiate the booking of leave. Additionally, this is a retrieval-based agent, not an autonomous one. It doesn't proactively monitor systems and trigger actions or notifications based on events like upcoming training deadlines or performance reviews that are yet to be started. While the potential for such autonomous functionality exists using other tools like Copilot Studio (which may have different licensing implications), it is not a feature of this basic implementation.

What are the prerequisites for building this type of HR Copilot AI agent?

To build an HR Copilot AI agent like the one demonstrated, you will need access to a Microsoft 365 environment. Specifically, the demonstration utilises Microsoft 365 Copilot and its agent builder capabilities. However, it was also highlighted that Microsoft has announced Microsoft 365 Copilot Chat, which is available at no additional cost to users with an Entra ID account and includes an agent mechanism with a pay-as-you-go model. This offers a more cost-effective starting point for building and using agents. A well-organised document repository, ideally within a SharePoint site, containing the relevant HR information is also crucial, as this serves as the agent's knowledge source.

How is the knowledge source for the AI agent defined and managed?

The knowledge source for the HR AI agent is defined by linking it to a specific location, such as a SharePoint site and the document libraries within it. When configuring the agent, you can browse and select the relevant SharePoint site and then specify the document folders or individual files that the agent should draw information from. It's important to ensure that these documents contain the up-to-date and accurate HR information you want the agent to use when answering questions.

How is the HR AI agent created using Microsoft 365 Copilot?

The creation process involves using the agent builder functionality within Microsoft 365 Copilot (or potentially the agent features in Microsoft 365 Copilot Chat). You can initiate the process by describing the purpose of the agent, in this case, an onboarding assistant for new hires focused on HR processes. The platform then guides you through steps such as naming the agent, refining its instructions (what to emphasise and avoid in responses), defining its tone (e.g., friendly and professional), and, most importantly, specifying the knowledge source (the SharePoint document repository). Once configured, the agent can be created and is then ready for use.

How can the HR AI agent be accessed and shared with employees?

Once the HR AI agent is created, it can be accessed from within the Microsoft 365 Copilot interface. Additionally, a direct link to the agent can be generated and shared with employees. During the sharing setup, you can control the access level, choosing to share it only with yourself, specific users, or the entire company. It's crucial to remember that for the agent to function correctly for all intended users, the underlying SharePoint files and folders that serve as the knowledge source must be appropriately shared with those users.

Can the HR AI agent be modified or updated after its initial creation?

Yes, the HR AI agent can be modified and updated after it has been created. The creator of the agent can access its configuration settings through the Microsoft 365 Copilot interface. From there, it's possible to make changes to various aspects, including the agent's name, description, instructions, the specified knowledge sources, and even the starter prompts that are suggested to users. After making any necessary modifications, the changes can be saved and the agent will operate with the updated configuration.

How does Natural Language Processing (NLP) enhance the HR AI agent's functionality?

Natural Language Processing (NLP) enables the AI agent to understand and process human language, allowing users to interact with it using everyday language instead of specific commands or keywords. This makes the interaction more intuitive and user-friendly, as employees can ask questions in their own words and receive relevant responses. NLP helps bridge the gap between human communication and machine understanding, ensuring that the AI agent can accurately interpret and respond to a wide range of HR-related queries.

How does the HR AI agent integrate with Microsoft 365 applications?

The HR AI agent can be seamlessly integrated into Microsoft 365 applications such as Teams and SharePoint. This integration allows employees to access the agent directly from the platforms they already use for daily tasks. For example, within Teams, employees can interact with the AI agent in a chat format, asking HR-related questions and receiving instant responses. This seamless integration enhances accessibility and ensures that the AI agent is a convenient tool for employees, reducing the need to switch between different applications or websites.

How is data privacy ensured when using an HR AI agent?

Data privacy is a critical consideration when implementing an AI agent for HR purposes. The AI agent operates within the secure environment of Microsoft 365, which adheres to strict data protection and privacy standards. Additionally, organisations can control access to the AI agent and the underlying knowledge source by setting appropriate permissions on SharePoint and other integrated platforms. It's essential to regularly review and update these permissions to ensure that only authorised personnel can access sensitive HR information.

Can you provide a real-world example of an HR AI agent in action?

Consider a large organisation with employees spread across multiple locations. The HR department receives numerous inquiries daily about leave policies, benefits, and training schedules. By implementing an HR AI agent, employees can quickly get answers to these common questions by simply typing them into a chat interface in Microsoft Teams. The AI agent searches the designated SharePoint site for relevant documents and provides accurate responses. This reduces the workload on HR staff, allowing them to focus on more complex tasks while ensuring employees receive timely and accurate information.

What challenges might an organisation face when implementing an HR AI agent?

One of the primary challenges is ensuring that the knowledge source is comprehensive, up-to-date, and accurately reflects the organisation's HR policies and procedures. Additionally, organisations may need to invest time in training employees to use the AI agent effectively and addressing any initial resistance to adopting new technology. Another challenge is maintaining data privacy and ensuring that the AI agent only accesses and shares authorised information. Regular updates and maintenance of the AI agent and its knowledge source are also crucial to ensure its continued effectiveness and reliability.

What is the potential future evolution of the HR AI agent?

As AI technology advances, HR AI agents could evolve from retrieval-based systems to more autonomous agents. This means they could proactively monitor HR systems, send reminders for upcoming deadlines, and even initiate certain actions based on predefined triggers. For example, an autonomous HR agent could automatically remind employees of upcoming performance reviews or training sessions. To achieve this level of functionality, organisations may need to explore additional tools and licensing options, such as those offered by Copilot Studio.

What strategic approach should organisations take to deploy an HR AI agent successfully?

Organisations should start by clearly communicating the benefits and capabilities of the HR AI agent to employees, highlighting how it can streamline HR processes and provide quick access to information. Training sessions can help employees understand how to interact effectively with the AI agent. Encouraging feedback and using it to make continuous improvements to the agent's functionality is essential. Regularly updating the knowledge source to reflect any changes in HR policies and procedures is also crucial to ensure the AI agent remains a reliable resource for employees.

How does low-code development facilitate the creation of HR AI agents?

Low-code development allows individuals without extensive programming skills to create and configure AI agents using visual interfaces and pre-built components. This approach makes it easier for HR professionals and other non-technical users to build and customise AI agents according to their specific needs. With tools like the Agent Builder in Microsoft 365 Copilot, users can define the AI agent's purpose, instructions, and knowledge source without writing complex code, significantly reducing the time and effort required to deploy an AI solution.

Why is choosing the appropriate knowledge source critical for the AI agent's effectiveness?

The effectiveness of an HR AI agent largely depends on the quality and relevance of its knowledge source. A well-curated SharePoint site with up-to-date HR documents ensures that the AI agent can provide accurate and relevant answers to employee queries. Organisations should regularly review and update the knowledge source to reflect any changes in HR policies, procedures, or benefits. Ensuring that the documents are well-organised and easily accessible will also enhance the AI agent's ability to retrieve and deliver the right information efficiently.

How can organisations encourage the adoption of an HR AI agent among employees?

To encourage adoption, organisations should highlight the benefits of using the AI agent, such as quick access to HR information and reduced wait times for responses. Providing training sessions and resources can help employees become familiar with the AI agent's capabilities and how to use it effectively. Creating a feedback loop where employees can share their experiences and suggestions for improvement can also foster a sense of involvement and ownership. Regularly communicating updates and enhancements to the AI agent will keep employees engaged and informed about its evolving capabilities.

What are some common misconceptions about AI agents in HR?

A common misconception is that AI agents can completely replace human HR staff. In reality, AI agents are designed to complement human efforts by handling routine queries and tasks, freeing up HR personnel to focus on more complex and strategic responsibilities. Another misconception is that AI agents can access and process all types of data. However, their functionality is limited to the defined knowledge source, and they cannot autonomously access or modify sensitive data without appropriate configurations and permissions.

What additional capabilities might an autonomous HR AI agent possess?

An autonomous HR AI agent could monitor HR systems and workflows, sending notifications for upcoming deadlines, such as performance reviews or mandatory training sessions. It could also initiate certain actions, like scheduling meetings or reminding managers of pending approvals. Achieving this level of autonomy would require integrating the AI agent with other HR systems and databases, as well as ensuring compliance with data privacy regulations. Organisations may need to explore advanced AI tools and licensing options to develop these capabilities.

What are the cost considerations for implementing an HR AI agent?

The cost of implementing an HR AI agent can vary based on the tools and licensing options chosen. Using Microsoft 365 Copilot Chat with a pay-as-you-go model can be a cost-effective option for organisations looking to pilot the technology without committing to a full Microsoft 365 Copilot license. However, organisations should also consider the potential costs of training, maintenance, and updates to the AI agent and its knowledge source. Evaluating the overall return on investment, including time savings and improved employee satisfaction, can help justify the costs associated with deploying an AI agent.