AI That Works, Episode 5: Building AI for Your Business – AI Assistants That Take Action

An AI assistant that knows your expenses policy is useful. One that can take your train ticket and raise the claim for you is a different thing. This is Episode 5, the final episode of series one of BCN's AI That Works webinar series, following Episodes 1 to 4 on safe adoption, prompting, everyday Copilot wins and building your first agent. It shows what happens when agents stop answering questions and start taking action.

BCN's head of AI and data innovation builds a customer service agent live in Microsoft Copilot Studio. It looks up real orders, checks returns against policy, opens support cases and alerts the team in Teams, with the full conversation transcript attached to each case. The session covers the five building blocks of an agent that can act (knowledge, topics, actions, triggers and channels), how to connect it to live business data, and how the same pattern applies to HR onboarding and IT triage.

It's frank about the limits: an agent is only as accurate as the data behind it. It also covers keeping a person approving consequential actions, giving each agent an owner, and what it costs. It comes from a Microsoft partner and is built around Microsoft's tools, so expect some promotion of BCN's AI services.