Grounding

Google Bard: What does grounding mean in M365 Copilot?

Grounding in M365 Copilot refers to the process of connecting the AI model to the user's business data. This is done through the Microsoft Graph, which is a real-time index of all the user's content and context in Microsoft 365.

Grounding allows Copilot to understand the user's specific needs and context, and to provide more relevant and accurate suggestions. For example, if the user is writing a document about a specific product, Copilot can access the user's product documentation and provide suggestions that are relevant to that product.

Grounding is an important part of what makes Copilot so powerful. It allows Copilot to be more than just a general-purpose AI assistant. It allows Copilot to be a truly helpful tool for users in the workplace.

Here are some of the benefits of grounding in M365 Copilot:

• More relevant suggestions: Copilot can provide more relevant suggestions because it understands the user's specific needs and context.
• More accurate suggestions: Copilot can provide more accurate suggestions because it has access to the user's business data.
• Improved productivity: Grounding can help users to be more productive by providing them with the information they need to get their work done faster and easier.
• Increased collaboration: Grounding can help users to collaborate more effectively by providing them with a shared understanding of the work that needs to be done.

Overall, grounding is a key feature of M365 Copilot that makes it a powerful and valuable tool for users in the workplace.

Microsoft Edge: What does grounding mean in M365 Copilot?

Grounding is an approach that Copilot uses to improve the specificity of the input prompt, so the user gets answers that are relevant and actionable to their specific task. Grounding involves using the information from the Microsoft Graph, such as the user's calendar, emails, chats, documents, meetings, and contacts, to enrich the prompt with additional context and details. For example, if the user asks Copilot to “write a summary of the project status”, grounding will add information such as the project name, deadline, progress, and stakeholders to the prompt before sending it to the large language model (LLM) for generation. This way, Copilot can produce a more accurate and useful response for the user based on their data in Microsoft 365 .