New M365 Copilot Tuning

Microsoft 365 Copilot Tuning is a new feature currently in preview. Microsoft 365 Copilot's large language model (LLM) is a general model used by all Microsoft 365 tenants. This LLM is not specific to any customer, and it is not a learning model. When users interact with Microsoft 365 Copilot, the general LLM along with Microsoft 365 data is used to generate responses. While this setup functions adequately, the responses are not entirely tailored to an organization's unique knowledge or approach.

With Microsoft 365 Copilot Tuning, users can now create fine-tuned models that perform domain-specific tasks related to their organization's knowledge. These models can be trained on a subset of data, such as authoritative documents in a library. Once created using domain-specific data from your tenant, these models can be utilized by a Copilot Studio agent. The tailored model remains within your tenant and does not revert to the generic Microsoft 365 Copilot LLM. Therefore, results provided by an agent leveraging a fine-tuned model created with your organization’s data will be more accurate and comprehensive. These models are low-code solutions that are straightforward to create using Copilot Studio.

Reference: 

https://learn.microsoft.com/en-us/copilot/microsoft-365/copilot-tuning-overview

https://techcommunity.microsoft.com/blog/microsoftmechanicsblog/new-microsoft-365-copilot-tuning--create-fine-tuned-models-to-write-like-you-do/4416100

https://www.microsoft.com/en-us/microsoft-365/blog/2025/05/19/introducing-microsoft-365-copilot-tuning-multi-agent-orchestration-and-more-from-microsoft-build-2025/