Copilot Gets Smarter: What’s New in Microsoft’s Wave 3 AI Update

Microsoft’s latest Wave 3 updates for Copilot mark a clear shift in how AI supports everyday work. During our recent webinar, we explored what’s new, what it means in practice, and how organisations can start seeing real value from these tools. 

The key theme is simple: AI is moving from helping with tasks to completing them for you.

From Prompts to Productivity: The Shift to Smarter Workflows 

Earlier versions of Copilot focused on simple interactions. You would ask a question, receive an answer, and then move on. 

With Wave 3, that model is evolving. Copilot now handles multi-step tasks across your apps in one go, reducing manual effort and saving time. 

Instead of asking for help step by step, you can now brief Copilot once and let it complete the work behind the scenes. This includes pulling data from files, creating documents, drafting emails, and even setting up meetings automatically. 

This shift is what makes the latest update feel practical rather than experimental.

Copilot Cowork: Delegating Real Work to AI

One of the standout features is Copilot Cowork. 

This new capability allows you to give Copilot a single, more complex instruction, and it will break it down into actions. For example, in the webinar, we demonstrated how one prompt could: 

• Create a summary presentation 

• Draft an internal briefing email 

• Schedule a follow-up meeting 

all in one go.  

What makes this powerful is not just the output, but how it happens. Copilot works directly within your Microsoft environment, accessing tools like Outlook, Excel, Word, and PowerPoint to complete the task fully. 

It removes the need to copy, paste, and switch between apps, which is where many workflows lose time. 

More Choice, Better Results: Introducing Multiple AI Models 

Another major change is the introduction of multiple AI models working together. 

Previously, Copilot relied mainly on GPT models. Now, Microsoft has introduced Claude models, giving users access to different strengths depending on the task. 

For deeper research and more complex thinking, Claude can provide more detailed responses. For general use, GPT still performs strongly. 

By combining them, users can benefit from better quality outputs without needing to choose manually every time. 

Compare and Decide: The Model Council Approach

To help organisations make the most of these options, Microsoft has introduced Model Council within the Researcher agent. 

This allows you to run the same request across different models and compare the results side by side.  

In practice, this is useful for: 

• Exploring different perspectives 

• Validating outputs 

• Making more informed decisions about AI use 

It also reinforces an important principle: AI outputs should always be reviewed and verified, especially when decisions matter. 

Deeper Insights, Faster: What Research Mode Can Do

One of the most impressive use cases shared in the session was around AI-powered research. 

Using Copilot’s Researcher agent, a detailed risk assessment was created from multiple documents in a fraction of the usual time. The process produced a long, structured report with clear sources and references.  

This highlights where AI delivers the most value: 

• Analysing large volumes of information 

• Producing detailed summaries 

• Supporting faster decision-making 

For many organisations, this is where the real return on investment begins. 

Edit With Copilot: Working Directly in Your Files

Another update now available to all users with a paid licence is Edit with Copilot. 

This allows Copilot to work directly inside your documents and spreadsheets, rather than simply suggesting outputs. 

In Excel, for example, it can: 

• Analyse data 

• Build tables and formulas 

• Create visualisations that update automatically 

Importantly, these outputs are not static. They remain linked to your data, meaning any changes are reflected immediately. 

This makes it a practical tool for day-to-day work, not just a one-off assistant. 

What This Means for Your Organisation

The overall message from Wave 3 is clear. 

AI is no longer just about experimentation. It is about embedding smarter ways of working into everyday processes. 

But getting value from tools like Copilot is not just about having access. It is about: 

• Understanding how to use it effectively 

• Training your team 

• Aligning it with your internal workflows 

How Pugh Can Support Your AI Journey 

At Pugh, we work with organisations at every stage of their AI journey. 

Licensing and readiness: We help ensure you have the right Copilot licences in place and set up correctly. 

Training and adoption: Our sessions cover everything from basic AI use to advanced Copilot features, helping your team build confidence quickly. 

Strategy and implementation: We support you in identifying real use cases and embedding AI into your processes. 

Whether you are just getting started or looking to scale your use of AI, we can help you move forward with confidence. 

Ready to Get More from Copilot? 

If you would like to explore how Copilot could support your organisation, get in touch with our team today. 

We are happy to run a short, practical session to show what is possible and help you take the next step. 

Contact us here