In A(i) Nutshell

Hosted by Andrew Davis

Business Marketing

Website RSS feed

Episodes

718

Latest episode

Aug 2026

Language

EN-GB

About the show

A 10-minute daily podcast about the world of Generative AI for marketers and the everyday person.

Listen to episodes

60 recent

August 21, 202610 min

Anthropic's CEO's Own Wife Barely Shows Up in Claude's Training Data and That Should Worry You (AI News)

Andrew Miles Davis covers a genuinely quiet news week that still surfaced some meaningful stories, starting with the observation that Anthropic CEO Dario Amodei's wife appears almost nowhere in Claude's outputs despite being easily findable elsewhere online, prompting a wider concern about what large language models choose not to train on and why having multiple competing platforms matters as a safeguard against any single company controlling what information surfaces. He covers Robin Williams' children reactivating his Instagram account to give fans an authentic reference point against the growing spread of AI-generated versions of his voice and likeness, ChatGPT launching a dedicated version for teenagers with study tools and safety guardrails around self-harm and explicit content, Google DeepMind releasing Backstory, a verification tool built to help journalists and fact-checkers investigate manipulated images and content, and LinkedIn quietly adding an AI slop reporting option that appears to have no actual consequence for the content or its author. Subscribe to In AI Nutshell for the AI news that matters to marketers, every Friday in ten minutes.

August 20, 202611 min

When Will Using AI at Work Feel as Normal as Sending an Email (YouTube Edition)

You can find this episode on YouTube: https://www.youtube.com/watch?v=Pkp-wTHGjMQ Andrew Miles Davis opens a new video series by tackling a question he hears in almost every training session he delivers: when will using AI at work stop feeling unusual? Drawing on his Story Lens framework, built around the "stranger comes to town" story structure he has used with organisations for fifteen years, he argues that AI sits firmly in the adaptation stage rather than the new normal stage, and that most people waste time in this stage by spreading themselves across too many tools rather than going deep on one. He introduces a practical five-day test, reviewing the last five working days to separate AI use that has become instinctive from moments where AI could have helped but was not considered until after the fact, as a simple way to identify what to build into a genuine habit next. This is the first instalment in a planned series walking through all seven stages of the Story Lens framework. Subscribe to In AI Nutshell for daily ten-minute AI insight and to the YouTube channel for the full Story Lens series.

August 19, 202610 min

If AI Disappeared Tomorrow This Is the Main Thing I Would Actually Miss (Random Questions)

Andrew Miles Davis answers eleven randomly generated questions covering AI, marketing, and personal reflections in his favourite recurring format. He shares what he would genuinely miss if AI disappeared overnight, landing on convenience rather than capability, and revisits a frustrating recent experience where AI silently produced incorrect duplicate entries partway through a major data entry project despite performing well initially, a moment he still describes as a bad surprise rather than a good one. He also covers lighter ground including the craziest experience of his career, involving a missed flight, a broken mini-disc player, and an unplanned DJ set opening for Kanye West in front of 16,000 people, along with his picks for the brand with the most personality online, the piece of popular technology he thinks will eventually disappear, and why he could deliver a twenty minute TED talk with zero preparation any day of the week. Subscribe to In AI Nutshell for one of these unscripted, unfiltered episodes every month alongside daily ten-minute AI insight.

August 18, 202610 min

I Built a Website in Claude Design in Under 30 Minutes & I'm Genuinely Impressed (Cool Tools 75)

Andrew Miles Davis covers three tools on this week's Cool Tools Tuesday, leading with a genuinely impressed first experience of Claude Design after using it to build a full brand kit and website UX in under 30 minutes for a major upcoming project, having tried and failed to get the result he wanted through ChatGPT and Codex first. He gives an honest take on what this means for professional designers, arguing the tool will most benefit people who already understand strong design principles rather than replace that judgement entirely. He also introduces Capptivo, a free open source alternative to Screen Studio and Cursorful for creating polished screen recordings with automatic cursor following and zoom, currently Mac only and untested by Andrew but flagged as one to watch. The episode closes with VideoToBlog.ai, a tool that converts YouTube videos or audio files into properly structured blog posts built with AEO and GEO discoverability in mind rather than simply dumping a transcript onto a page. Subscribe to In AI Nutshell for three new tools every Tuesday with honest verdicts from someone testing them in real work.

August 17, 202610 min

Nobody Told Me How Angry AI Would Make Me So Here Are 5 Red Flags Worth Watching (Top 5)

Andrew Miles Davis steps away from the well-worn territory of privacy, environmental impact, and job losses to share five more personal red flags that have been genuinely bothering him about AI recently. He covers the flood of mediocre AI-generated content now that publishing requires almost no effort, the growing gap between how fast the technology moves and how well organisations actually understand the consequences of deploying it, and the habit of outsourcing thinking simply because AI offers a faster route than sitting with a problem. The most personal entries come from a recent frustrating experience where AI silently produced fifteen duplicate answers across a long data entry project before admitting the error only when directly challenged, and the broader realisation that there is no meaningful way to hold AI accountable when it repeatedly gets things wrong, unlike a colleague, a friend, or a child. Subscribe to In AI Nutshell for daily ten-minute episodes that stay honest about what using AI is actually like, not just what it claims to be capable of.

August 14, 202611 min

Claude Is About to Hide Invisible Watermarks in Everything It Writes (AI News)

Andrew Miles Davis covers a week of stories that point toward what he sees as the coming new normal across education, law enforcement, and entertainment. A University of St Andrews medical student was flagged by AI detection and forced to defend his own dissertation before an academic panel, despite having written it entirely himself. New Orleans police have begun using AI to triage overflow 911 calls, while Florida sheriff's departments have quietly purchased Meta Ray-Ban smart glasses for officers. Meta faces a growing trust gap around personal data in both the UK and US, arriving the same week Mark Zuckerberg published a lengthy essay advocating for open AI models and broader public access to AI capabilities. Claude announced plans to embed invisible watermarks into generated text that persist through editing, sparking debate over surveillance versus authorship verification, while researchers found ChatGPT now displays sponsored ad placements on roughly 26% of commercial-style prompts. The episode closes with data showing over 10% of Hollywood job postings now list AI skills as a requirement, reinforcing Andrew's ongoing argument that AI competency is shifting from niche advantage to expected baseline across every industry. Subscribe to In AI Nutshell for the AI news that matters to marketers, every Friday in ten minutes.

August 13, 20268 min

The Best Prompts Are Not the Prompt. They Are What You Add at the End. (AI Prompt Hacks)

Andrew Miles Davis returns to prompt hacks after a break spent building out a growing resource of his favourite prompt endings, arguing that the real leverage in prompting often comes not from the core request but from what gets added at the end of it. He shares the ELI shorthand for requesting age-appropriate explanations, a powerful ending that asks AI to surface the best question you are not currently asking about a strategy or scenario, and a simple but underused instruction to check your work for mistakes, weaknesses, or blind spots, optionally framed from the perspective of a teacher or boss. He connects each prompt back to his broader AI Essentials framework, particularly research and education and strategic planning, and points listeners to an ongoing Notion document where he continues to add new prompt endings over time. Subscribe to In AI Nutshell for prompt hacks and daily ten-minute AI insight built for marketers who want sharper results.

August 12, 202610 min

Master These 12 Things and You Have Mastered 95% of AI (AI Essentials 2026)

Andrew Miles Davis returns to a framework he built and taught extensively last year, updating it from the original eight essentials to twelve as he has continued refining how people actually use generative AI. He argues that nearly everything anyone does with a large language model falls into one of these categories: creating content, generating ideas, reformatting, analysing, strategic planning, summarising, optimising, seeking advisory input, exploring scenarios, research and education, asking or being asked questions, and companionship. He flags research and education alongside advisory as the two areas where hallucinations are most likely to cause real damage, given how confidently AI delivers incorrect information in those contexts. Rather than chasing every new tool or model release, Andrew makes the case that mastering these twelve categories covers roughly 95% of practical AI use, and challenges listeners to review their own recent prompts to see the pattern for themselves. Subscribe to In AI Nutshell for daily ten-minute episodes that simplify AI down to what actually matters.

August 11, 20268 min

One of My Favourite AI Tools Just Added a Game Changing Feature (Cool Tools 74)

Andrew Miles Davis covers three tools on this week's Cool Tools Tuesday, starting with Clara, a free YouTube transcription and summarisation extension offering clickable chapters and the ability to chat directly with video content without manually uploading it elsewhere, alongside a set of free supporting tools including a timestamp link generator and subtitle downloader. He also covers Caption Bolt, a standalone video finishing tool for short form content that handles resizing, trimming, B-roll, branding, and captioning in multiple styles, built around the simple promise of publishing more while spending less time editing. The episode closes with an update to Whisper Flow, Andrew's most used AI tool, which has added a note taking feature positioned directly against Granola, recording and summarising any audio in a room without needing to be formally invited to a meeting, complete with a personal thoughts section. Andrew notes he is actively trying to consolidate his toolkit rather than expand it, and this update may let him do exactly that. Subscribe to In AI Nutshell for three new tools every Tuesday with honest verdicts from someone testing them in real work.

August 10, 20267 min

Is Microsoft Copilot Actually Better Than ChatGPT? My Honest Answer (FAQs)

Andrew Miles Davis answers three questions from recent training sessions, starting with one of the most common questions he receives from corporate teams issued Copilot licenses by their IT department: is it actually better than ChatGPT. His honest answer is that for the vast majority of practical use, the leading tools perform similarly once properly trained, and the real differentiator becomes personal preference and familiarity rather than raw capability. He then shares his current top three AI tools, splitting the answer between what he finds most useful and what he genuinely enjoys using, landing on ChatGPT for its voice feature, Whisper Flow for accessibility given his RSI, and a split between MidJourney and Magnific for video and image work. The episode closes with a lighter question about whether Andrew uses a timer to consistently hit the podcast's ten minute format, revealing his radio background as the reason he can naturally pace an episode without watching the clock. Subscribe to In AI Nutshell for daily ten-minute episodes built around the questions real people are asking in training rooms right now.