With more ways than ever to build AI agents, how do you know which approach is right for your business? In this episode of Mastering Mondays, Matt and David explore AI harnesses in Microsoft Copilot Studio, comparing the Standard Harness with the GitHub Copilot Harness. They discuss orchestration, skills, memory, file creation and governance, using real-world scenarios to explain when each approach makes sense.

The key takeaway? Don't choose the newest technology just because it's available. Choose the right architecture for the job.

In this episode, David and Matt are joined by Microsoft MVP and AI Workforce Lead at Phoenix Software, Lewis Thompson, to explore the next big evolution in AI: Microsoft Copilot Autopilot. Lewis shares his hands-on experience with the technology, explaining how AI is moving beyond answering questions to independently carrying out everyday tasks. From automating time-consuming admin to understanding the importance of governance, security and responsible AI adoption, the conversation explores what Autopilot could mean for the future of work. Don't miss this insightful episode packed with real-world examples, practical advice and tips for getting the most out of Microsoft's latest AI capabilities.

How much is AI actually going to cost your organisation?

As Microsoft moves further towards usage-based billing, answering that question is becoming increasingly difficult. Copilot credits, consumption charges, different agent harnesses and costs associated with building, testing and running AI all mean organisations need to think differently about how they budget for Microsoft technology.

In this episode of Mastering Mondays, Matt and David welcome back Rich Gibbons, the first returning guest to the podcast, to make sense of the changing commercial landscape around Microsoft Copilot and AI.

The conversation explores how Copilot credits work, where unexpected costs can appear and why having a Microsoft 365 Copilot licence does not necessarily mean every Copilot experience is covered. Rich also explains why the shift towards consumption-based AI creates new challenges for finance, IT and development teams trying to predict and control spend.

More importantly, the discussion looks beyond the cost itself. If an AI agent costs £100,000 but creates £1 million of value, is it expensive? How do you build a business case when AI usage can be unpredictable? When should you use an agent, and when could a traditional automation achieve the same result at a lower cost?

Join Matt, David and Rich for a practical look at the real cost of AI, why organisations need to start designing with cost in mind, and what you can do now to avoid an unexpected AI bill.