The AI War: Agents vs. Automation

Introduction

In this episode, we talk with Dan Boyles about AI agents, automation, Copilot, security, data, and the future of autonomous systems.

Meet the Guest

Dan Boyles is an AI strategist, Microsoft Copilot expert and speaker. At the time of recording, he is Head of AI at First AI Group, where he helps organizations with AI strategy, security, adoption, governance and implementation. He previously spent around eight years at Microsoft working on Azure, cloud, modern workplace and Copilot. He's also known as "The Copilot Guy."

Setting the Stage

The word "agent" gets used for a lot of very different solutions. A chatbot, a workflow, or a Copilot Studio solution doesn't automatically make something a fully autonomous agent.

Episode Highlights

• Dan uses the route to Gatwick as an analogy: a workflow follows instructions, while an agent independently chooses the best route to the goal.
• His advice is to start at the bottom of the AI pyramid — with individual problems and everyday pain points.

Deep Dive

A workflow follows fixed rules. An agent can combine different sources of information, reason, and act independently within limits. The more autonomy a system is given, the more important data, governance, exception handling and human oversight become. An autonomous HR agent, for example, could kick off a formal procedure without understanding the human context behind it.

Real-Life Stories & Examples

• An HR agent helps employees find policies and procedures.
• A sales solution uses hundreds of old pitches to produce a client document that's roughly 80% finished.
• Onboarding can coordinate licenses, hardware, accounts and access rights.
• A brand agent checks visual identity, fonts, colors and copy.

Key Takeaways

• Not every solution labeled "agent" is actually autonomous.
• Start with a concrete pain point.
• HR, sales, onboarding and brand control are practical starting points.
• Data, SharePoint, permissions and governance need to be in order first.
• ROI can also come from innovation, creativity and reduced workload.
• Semi-autonomy is often more realistic than full autonomy.

Closing Thoughts

The value isn't in the "agent" label — it's in solving a problem people deal with every day. Start small, make the value visible, keep people involved, and make sure your data is in order.