The course includes a white paper in PDF format and over 1.5 hours of video content, focused on practical examples of how to use these technologies. The video content complements the white paper, which provides links, references, and more detailed descriptions of the concepts discussed. In contrast, the video is more effective in illustrating the results obtained and the effort required.

First, we will give an introduction to the foundational concepts and generative AI. This will provide a strong, common understanding before we go deeper into specific concepts and scenarios. This introduction is tool- and domain-agnostic, but we have chosen examples relevant to BI professionals.

Next, we will provide a detailed overview of the five key “building blocks” to consider to effectively utilize AI and agents: model, context, prompt, tools, and environment. This is the most significant part of the whitepaper, providing practical guidance and information on these concepts, their importance, and the high-level decisions you make when managing them.

Finally, we discuss four scenarios in this whitepaper that are possible and emerging in today’s market:

  • Scenario one: chatbot tools. This scenario involves using the out-of-the-box chatbot experiences either available in generic tools like ChatGPT or specialized experiences in Copilot for Power BI. In these scenarios, the
    user makes changes with assistance from chatbot suggestions.
  • Scenario two: augmented chatbots. This scenario involves enhancing chatbots by customizing the context and tools that they have in order to provide better outputs and affect the real world. In this scenario, the chatbot might make minor changes, but has limited use in real development tasks.
  • Scenario three: agentic development. This scenario involves using coding agents like GitHub Copilot agent mode or Claude Code to read and write changes to artifact metadata under human orchestration. Here, the agent is a worker and partner for the human developer to complete tasks.
  • Scenario four: asynchronous agents. This scenario involves the delegation of asynchronous coding agents like GitHub Copilot coding agent tasks, Claude Code GitHub actions/SDK, or Google Jules. These agents work asynchronously on a task in the background without supervision, and a human expert reviews and approves their output.

Note that these scenarios do not focus on models or tools, which are changing too fast. Instead, we focus on the high-level concepts and approaches we find interesting, valuable, and least likely to change. This whitepaper explains these concepts practically as they pertain to BI development, so that you can efficiently understand, experiment, and evaluate whether this might be valuable for yourself, your teams, and your organization.