Solution

We delivered a focused 2-hour session on AI-assisted and agent-based development. The aim was to give a practical view of how these tools work today and how they can support real engineering tasks. The session included a short intro, a hands-on demo, and time for questions.

The demo had a fun meta angle. The trainer used their own coding agent as the example project, and then improved that agent with Cursor. In other words, we used an AI-assisted environment to develop an AI-powered agent.

The live demo walked through the full lifecycle of building with agents: defining behavioural rules, co-designing tasks with the agent, implementing those tasks in practice, working with prompt/command files, and operating MCP-based tools for actions such as task retrieval and PR creation or review.

Although the session used Cursor as the primary environment, the concepts and workflows directly map to GitHub Copilot, and we highlighted the parallels throughout.

Value

The training gave developers a clear picture of how coding agents actually work, not just how they appear on the outside. By developing the trainer's agent further during the demo, participants saw the moving parts and learned patterns they can reuse in their own work. The session offered practical takeaways, helping teams understand what these tools can do right now and how to fold them into everyday development with minimal friction.

Yle – the Finnish Broadcasting Company is Finland's national public service media company. They produce content and services for audiences in equal conditions regardless of their residence status, wealth, age or gender. For all of us, for each of us.