AI can generate a recap that sounds like a podcast. The real question is whether it captures what you actually meant.

We’re back for Season 2 of The Boring AI Show with a hands-on experiment: we feed Season 1 transcripts into NotebookLM and listen to the AI-generated audio summary. It’s impressive, fast, and genuinely useful, and it also shows a common failure mode of generative AI summarization: it can compress so hard that it drops context, blurs specific stories, and quietly changes the meaning. That becomes our jumping-off point for a practical lesson you can use immediately at work.

From there we break down the difference between prompt engineering and context engineering. Prompts are what you ask the model to do. Context is what the system knows and what it is allowed to use, including your files, constraints, and approved definitions. We share an easy framework (who, do, what, how), talk about why “What is AI?” is a terrible prompt, and explain how better constraints reduce generic outputs and hallucinations. We also touch on choosing tools based on the job, from ChatGPT 5.5 and newer agent features to Claude, Miro slide decks, and creative workflows that support better storytelling instead of cookie-cutter slides.

We close with what we’re watching for Season 2: more real use cases, real organizational change, the coming AI backlash, and how education and community efforts shape responsible adoption. Plus, our AI wins of the week include students emphasizing human-in-the-loop design and a “second brain” knowledge graph using Obsidian with Claude to speed up case studies and slide decks.

Subscribe, share the show with a friend, and leave a review if this helps you cut through the hype. What topic or use case do you want us to test next?

Thanks for listening! Visit us at @mindovermachines.com to learn more about us!
Want to stay up to date on our latest AI insights? Subscribe to our newsletter here!