Andrew Miles Davis answers three questions from recent training sessions and keynotes. He starts with whether your next manager could be AI, concluding that it could handle functional work like allocating tasks, tracking performance, and analysing data, but that accountability is the real barrier, since you cannot discipline or hold an AI personally responsible when something goes wrong. He then explains how to make AI forget something you have told it, focusing on temporary chat modes found in ChatGPT, Claude, and Copilot, while warning that temporary does not necessarily mean no record exists, so confidential company information should never be treated as safe. He also mentions deleting saved memories in settings and asking the model not to remember as alternatives. The episode closes with the question he hears in almost every workshop, which single AI tool he would pick, and his answer that there is no single best tool, only the right tool for what you are trying to do. For him that means ChatGPT and Whisper Flow, for stylisation-heavy presentation work it would be Midjourney, and for video producers it would likely be Seedance. Subscribe to In AI Nutshell for daily ten-minute episodes built around the questions real people are asking about AI right now.

Andrew Miles Davis marks week 82 of Cool Tools Tuesday by revisiting Napkin AI, a visual graphics tool he first covered in episode two, which has found new relevance with a presentation feature. It asks how many slides you want, offers four options per slide, follows your chosen style as it goes, and gives more control than prompting a whole deck in Copilot or ChatGPT. He then covers Ahrefs' free AI visibility checker, long known for SEO, which shows how many times a brand is mentioned across ChatGPT, Perplexity, Copilot, AI Overviews and AI Mode, plus top topics, most cited domains and most popular pages, and he tests it on a company he trains for. The episode closes with GPT Watermark Remover, a free tool that detects and strips invisible characters such as zero-width spaces from text pasted out of AI chat tools, useful for anyone worried that hidden markers could trigger AI detection or cause formatting issues. Subscribe to In AI Nutshell for three new tools every Tuesday with honest verdicts from someone testing them in real work.

Andrew Miles Davis argues that the real story behind OpenAI's new $500 a month ChatGPT plan is not the price itself but the reduced allowance quietly applied to the existing $200 tier, and what that signals for everyone further down the ladder. He sets out three ways of paying for AI emerging in parallel, with money, with advertising, or with reduced access, and warns that features currently usable for free may gradually move behind subscriptions. He explains why this matters more than a Netflix-style tier split, since better AI is an economic capability rather than a nicer evening in front of the television, and walks through how to judge whether a pricier plan is worth it by asking what it would let you do and whether you can put a financial value on that difference. The closing advice is practical: use the cheap era to experiment hard, build real workflows, make the mistakes now, and avoid waiting for AI to improve or for a company strategy before starting. Subscribe to In AI Nutshell for daily ten-minute episodes tracking exactly how and when the golden era of affordable AI ends.

Andrew Miles Davis covers a week of releases that somehow topped the previous one, led by OpenAI launching Dot, an always-on assistant that takes action rather than just answering, positioned as a direct response to Meta Muse overtaking ChatGPT as the most downloaded app. He breaks down OpenAI's new $500 a month tier, more than double its previous top plan, framing it as confirmation of a pricing trajectory he has predicted for over a year where the $20 tier eventually offers what free currently does. Other releases include Claude's new Sonnet 5.5 model, which is already performing impressively enough that Andrew plans to test it against his usual Opus setup, Kling 4.0 re-entering the AI video race against Seedance, Eleven Labs releasing version four, Copilot introducing its own coding tool to compete with Codex and Claude Code, Fireflies launching a Whisper-style universal dictation feature, and Gemini's upcoming move to replace Gems with universal skills compatible across models. He also covers Apple raising phone prices partly due to AI-driven chip and memory shortages, Reid Hoffman criticising the tech sector's apocalyptic AI messaging, an Ofcom report showing 21% of UK adults now use tools like ChatGPT as a primary news source despite lower trust than traditional journalism, and Google's publisher payout test revealing that most media outlets are recovering less than 0.1% of lost referral revenue. Subscribe to In AI Nutshell for the AI news that matters to marketers, every Friday in ten minutes.

Andrew Miles Davis opens October by reviewing a genuinely productive September, where a return to full training routine, quiet progress developing the Story Lens into proper intellectual property, and a 21-day daily LinkedIn video run all delivered on what he had hoped for. For October he sets out three priorities, starting with finding the right YouTube format after three attempts, a talking head episode, then two use case videos that were useful but not entertaining, landing on a plan to aim for genuine edutainment rather than pure utility. On LinkedIn he is scaling back from daily posting to a more sustainable weekly rhythm of one FAQ video and one step-by-step tutorial, filling remaining days with curated content and, in a habit he openly admits he has neglected, actually posting about his own keynotes and training sessions. The third focus is pushing agentic AI tools like Muse and Codex properly, setting two or three genuinely significant tasks with a clear end goal rather than spreading himself across too many half-finished experiments, a pattern he is consciously trying to avoid this time. Subscribe to In AI Nutshell for daily ten-minute AI insight from someone building all of this in real time.

Andrew Miles Davis wraps up September with his monthly tool review, led by Meta's Muse, an agentic assistant he had held off testing for a while given his earlier frustration with Open Claw, which he had bought a Mac Mini for specifically and ultimately abandoned due to complexity. Muse's WhatsApp and iMessage style interface removed that friction entirely, and after pushing it with increasingly difficult tasks he found it succeeding roughly 80% of the time, enough that he says he would genuinely pay for it once the current free access ends, provided the same capability isn't matched elsewhere in tools he already pays for. Claude Design returns for a second consecutive month, credited as the driving force behind a 21-day daily LinkedIn video challenge where motion graphics that would previously have required outsourcing now take seconds to produce. Copilot in Excel also holds its place, now underpinning nearly every operational spreadsheet in Andrew's business from podcast tracking to LinkedIn planning to daily habit discipline, run consistently across both his Mac Mini and laptop. Subscribe to In AI Nutshell for a daily ten-minute update on what is actually worth using, from someone running these tools in real client work every day.

Andrew Miles Davis covers three quick, self-explanatory tools on week 81 of Cool Tools Tuesday. He starts with Canva's AI Title Generator, a feature surfaced by an attendee in a recent training session who found it genuinely useful for brainstorming hooks and headlines, though Andrew personally still prefers training a large language model directly for that level of control. He covers Good Vibes, a free Chrome extension that lets users specify exactly what they never want to see across platforms like X, Reddit, LinkedIn, and YouTube, then quietly filters it out of the feed, offering both a capped free tier and an unlimited option via a Typesafe API key. The episode closes with IrisAgent, a personal AI assistant built around iMessage that connects apps like Slack, Notion, and Gmail so tasks can be delegated by text, which Andrew frames as an early signal of the agentic assistant trend he has been predicting will become mainstream by 2027, pointing to Meta's Muse as further evidence that these tools succeed by working inside interfaces people already know how to use. Subscribe to In AI Nutshell for three new tools every Tuesday with honest verdicts from someone testing them in real work.

Andrew Miles Davis ranks the five large language models he pays for as of 28 September 2026, after being asked the question twice in a single week. Gemini lands at number five, strong in its wider Google ecosystem but still weak as a standalone model for how he works, with YouTube handling as its one clear advantage. Copilot takes number four on the strength of its native integration with Excel, Word, and PowerPoint, which he now considers a game changer for spreadsheet formulas and document building, despite occasionally taking longer than doing the task manually. Claude sits at number three purely because of Claude Design, which he says has transformed his ability to create motion graphics for LinkedIn video, while its memory inconsistency and habit of either forgetting or over-obeying training instructions is why it would be the first cut if budgets tightened. Perplexity takes number two as his first port of call for research and trustworthy answers, and ChatGPT holds number one thanks to the combined strength of its chat, work mode, and Codex. He also flags Meta's Muse as a likely top three contender in six months once the honeymoon period passes. Subscribe to In AI Nutshell for daily ten-minute episodes that tell you honestly which tools earn their place in a real working week.