Andrew Miles Davis kicks off a new series drawing on patterns he has observed across thousands of face-to-face training sessions, starting with two of the most common reasons people struggle with AI tools despite genuinely wanting to use them well. The first is an expectation problem, where people go in expecting the accuracy of a database, the creativity of a human, and the consistency of software, and then lose patience when AI delivers something messier than that. The second is a habit carried over from two decades of search engine use, treating large language models like Google by typing short queries, reading the first answer, and moving on. Andrew argues that shifting these two mindsets alone would improve most people's results overnight. Subscribe to In AI Nutshell for daily ten-minute episodes built around what actually happens when real people use AI at work.

Andrew Miles Davis covers this week's AI news with a focus on what it means for marketers, starting with new data showing bot traffic now accounts for over half of incoming visits for some publishers as AI crawlers scrape content to train large language models without permission. He explains the shift from SEO to GEO, generative engine optimisation, and why marketers are already gaming the new system in ways that could poison the quality of information AI models produce. He also covers a MIT and Stanford study finding that AI systems frequently prioritise agreement over accuracy, a behaviour known as sycophancy that makes training your model to push back more important than ever. The episode rounds off with the impact of AI short-form video tools on the editors who built careers clipping long-form content for influencers. Subscribe to In AI Nutshell for the AI news that actually matters to marketers, every Friday in ten minutes.

Andrew Miles Davis shares three prompt extensions that consistently produce sharper, more useful answers from any large language model, starting with a technique that forces the AI to identify what you might be getting wrong before it even answers. He also breaks down a comparison framework that goes beyond asking what competitors do, by asking what a third player would do to beat them both, and rounds off with a prompt built around the idea that success leaves clues, using reverse engineering to unpack how someone or a company achieved a specific result. All three are practical, immediately usable, and explained with real examples from Andrew's own prompting sessions. Subscribe to In AI Nutshell for prompt hacks, tool reviews, and daily AI insight built for marketers.