Microsoft’s recent Build 2025 announcements have brought some massive updates to the whole GenAI Copilot platform. One of the most exciting features for me (and perhaps a revealing feature in terms of Microsoft’s GenAI strategy) is the introduction of Microsoft 365 Copilot Tuning. This new feature is set to revolutionise how organisations tailor AI to their specific needs, moving powerful model fine-tuning capabilities from the often complex of technical staff to the more accessible world of business power users.
What is it?
So, what exactly is “model fine-tuning,” and why is this announcement such a big deal?
Essentially, fine-tuning allows you to take a powerful, general-purpose AI model and customise it with your own organisation’s data and unique language. Think of it like tailoring a bespoke suit perfectly fitted to your measurements, rather than buying off-the-rack.
Previously, achieving this level of GenAI model customisation was often a significant technical hurdle, demanding technical expertise and considerable effort. The great news from Build 2025 is that Microsoft is democratising this capability. With Copilot Tuning in Microsoft Copilot Studio, fine-tuning is now accessible through a low-code, self-service interface. This means subject matter experts within your business, the people who truly understand your operations and data can now directly participate in training AI models for specific tasks like document generation, summation, and expert Q&A. It’s a significant step, bringing sophisticated AI customisation closer to the users who benefit most, without them needing to become AI programming experts.
What does it mean from a business perspective?
The ability for organisations to easily fine-tune AI models with their own data has some profound implications:
- Hyper-Relevant AI: Imagine AI that truly “speaks” your company’s language, understands its uniqueness, and can generate documents or answer questions with the precision of an internal expert. Fine-tuning makes this much more of a reality.
- Increased Productivity & Efficiency: By training models on specific tasks like creating complex contracts or summarising lengthy reports, you can automate and accelerate workflows, freeing up your team for more strategic work.
- Democratised GenAI Model Development: You no longer necessarily need a large technical team for every GenAI model customisation project. Business users and subject matter experts can now take the lead, fostering innovation from the ground up using intuitive tools.
- Enhanced Data Security and Governance: Copilot Tuning is built with enterprise-grade security. It uses “Eyes-Off Processing” so no one sees file contents during ingestion, and the fine-tuned models inherit the access permissions of the training data, ensuring your sensitive information remains protected.
- Seamless Integration: These fine-tuned models can be more easily incorporated into agents built with Copilot Studio and deployed across Microsoft 365 applications, embedding specialised AI directly into your daily workflows.
What do I do with it?
Ready to explore how this can benefit your organisation? Here are a few starting points:
- Check Eligibility & Rollout: At the moment, this is where the good news stops for lots of organisations but does give some insight into the strategy of pushing the functionality to the end-users. It’s important to note that this powerful fine-tuning capability is available for organisations with 5,000 or more Microsoft 365 Copilot seats. Always verify the specific rollout plans, licensing details, and any prerequisites..
- Identify Use Cases: Think about tasks in your organisation that involve generating specific documents (like contracts or agreements), or require deep domain expertise for Q&A. These are prime candidates for fine-tuning. Combine the fine-tuning with RAG to access regularly changing data to get the best of both worlds – a model that speaks your corporate language and understands your business combined with access to your most up to date data.
- Engage Your Subject Matter Experts: These are the people who will be key to successful fine-tuning. Start conversations about how a tailored GenAI model could help them and what data would be most valuable for training.
- Explore Copilot Studio: Familiarize yourself with Microsoft Copilot Studio and its new capabilities. Look out for the “highly automated recipes” that simplify data selection and processing for fine-tuning.
- Plan for Governance: While the tools are becoming easier to use, ensure you have a strategy for governing the creation, deployment, and maintenance of fine-tuned models to maintain quality, accuracy, and alignment with business objectives.
The introduction of low-code model fine-tuning in Microsoft Copilot Studio is a significant step in making advanced AI accessible and practical for businesses. It empowers organisations to create highly specialised AI agents that truly reflect the organisations unique operational landscape, all while maintaining robust security and governance.
What are your thoughts on this new capability?
Further Reading
Microsoft 365 Copilot Blog – Introducing Microsoft 365 Copilot Tuning (Microsoft Ranveer Chandra)