Skip to content

You’re Not Behind (Yet): How to Learn AI in 19 Minutes

2026-02-12 · Long video (solo, talking head or whiteboard) · Ali · recording, 19:56 · watch on YouTube

Topic: Tools, apps and tech · mode: Explanation and tutorial · tags: AI, Sponsor segment present, His own numbers

Ali sets out a five-phase plan, first made as a training for his team, to become fluent with AI in about three months: build daily habits and tools, use AI as a coach, then as a worker under the 10-80-10 rule with your own taste as the check, then as a system through a prompt library, and finally as infrastructure through automations.

What is said

The parts of the video and what is said in each (★ — how much it matters to the video).

00:00:00 — Five phases to become fluent with AI in about three months

  • ★★★ Ali gives a five-phase process to become fluent and productive with AI in about three months: foundations in week one, AI as coach in week two, AI as worker in weeks three and four, AI as a system over one to two months, and AI as infrastructure from month four. — 00:00:00
  • ★★ Business owners increasingly decide whom to hire, fire and promote by their level of AI fluency, Ali says, and the gap between businesses that use AI properly and those that do not is widening. — 00:00:00

00:00:24 — Phase 1: five foundations for the first week

  • ★ Ali considers the five AI foundations non-negotiables for everyone in his team. — 00:00:29
  • ★★ Ali's first AI foundation: use AI for everything you would first have used Google for. — 00:00:29 · Google
  • ★★ Ali's personal preference is Claude by Anthropic for most things, but he says ChatGPT, Grok or Gemini are fine, whichever you already have access to. — 00:00:29 · Claude, ChatGPT
  • ★★ Ali's second AI foundation: keep your AI chat website open in a pinned tab, to build the habit of always having an AI chat window open. — 00:00:29
  • ★★★ Ali's third AI foundation: talk to AI rather than type, with Wispr Flow, built-in dictation on Windows and Mac or the AI tools' own; you speak faster and ramble more, and he says this one habit transforms how much value you get. — 00:00:58
  • ★★ Ali's fourth AI foundation: download the AI mobile apps so you can use AI as a Google replacement or thinking buddy anywhere, while walking or commuting. — 00:01:27 · Google
  • ★★ Ali's fifth AI foundation: automatically record your online meetings, and ideally your in-person ones too. — 00:01:27
  • ★★ To record and transcribe Zoom calls, Ali's team has used Grain for the last five years; for a free tool he usually recommends Fathom. — 00:01:56 · Zoom

00:02:18 — Phase 2: using AI as your coach, and its limits

  • ★★★ In phase two of learning AI, AI as your coach, you do not yet ask AI to do your work but to help you think better about the work you already do, Ali says. — 00:02:26
  • ★★ Ali's example coaching prompt for his social media manager Nicole: describe your role and goal (growing his Instagram from 1 million to 1.2 million followers in 90 days) and ask for the highest-leverage things to focus on, common mistakes and questions to ask your manager. — 00:02:26 · Instagram
  • ★★ In his own example prompt to Claude, Ali gives his goal for 2026: growing his business's revenue from $5 million to $10 million, with the new Lifestyle Business Academy as the biggest lever. — 00:03:24 · Lifestyle Business Academy, Claude
  • ★★ Whether or not you have a coach (Ali works with a couple), it is very helpful to use AI as a thinking buddy and coach that asks you questions, Ali says. — 00:03:24
  • ★★ Ali suggests putting the transcript of a meeting with your manager or team into AI for insights, for example asking for a two-week curriculum to improve your skills after a coaching talk with your manager. — 00:03:53 · Angus (general manager)
  • ★★ Ali puts the transcript of a Lifestyle Business Academy coaching session into AI to draw out the students' key themes and struggles for the curriculum, and asks for feedback on his own teaching style and blind spots. — 00:04:22 · Lifestyle Business Academy
  • ★★ Ali suggests asking AI to interview you about what you actually do in your role and help you find what is high-leverage and what is a waste of time. — 00:04:52
  • ★★★ Ali thinks of AI tools as a very smart colleague who reads many books but has no other context: useful to mirror your thoughts and ask you questions, but its advice should not be treated as gospel. — 00:05:21

00:06:05 — Brilliant sponsor read

00:07:34 — Phase 3: using AI as your worker with the 10-80-10 rule, and the idea of taste

  • ★★ The mistake Ali sees most people make is jumping straight to having AI do work, such as "write me this Instagram post", which gives generic, poor output. — 00:07:47 · Instagram
  • ★★★ In phase three, AI as your worker, Ali applies the 10-80-10 rule, which he thinks he got from Dan Martell's Buy Back Your Time: you do the first 10%, AI the middle 80%, and you the final 10% as a taste and quality check; asking AI for 100% gives garbage. — 00:08:16
  • ★★ Ali advises giving AI a lot of context, such as a video transcript, competitors' best reels and the content strategy doc, before asking for 20 hook ideas, then choosing among them with your own taste. — 00:08:45
  • ★★ Paste the AI ideas you liked back in and ask for more along those lines, Ali says: liking 5 of 20 and then 10 of 50 gives 15 ideas that are human-vetted rather than AI slop. — 00:09:34

  • ★★★ Taste, an intuitive feel for what is good and bad in your work, is the key thing that separates people who use AI well from those who don't, Ali says. — 00:09:34

    “This is the key thing that separates the people who use AI well from the people who don't.”

  • ★★ Cringing at AI output is a good sign that your bar for taste is higher than what the AI produced, Ali says; your job is to give it feedback as you would a junior team member or intern. — 00:10:26

  • ★★ By the end of week four Ali wants the habit of asking, for any task, whether AI could do it alongside you so you can test its response: a process of experimentation and testing. — 00:10:56
  • ★★ Phase four of learning AI is using AI as a system, so that it gets better over time instead of starting from scratch; Ali expects most people to need one to two months to get comfortable with it. (hedged) — 00:11:25

00:11:28 — Phase 4: using AI as a system, the cake recipe and the prompt library

  • ★★★ As a baker improves a cake recipe bake after bake, Ali says, you build your own prompt library through prompt engineering, updating each prompt into a new version as you see its results. — 00:12:52
  • ★★ In Ali's example, Nicole improves her Instagram hook prompt version by version, for instance adding that each hook be under 20 words, then never rhetorical questions, testing the output each time. — 00:13:22 · Instagram
  • ★★ Ali suggests saving a finished prompt as a TextExpander shortcut, such as IGH for Instagram hooks, so typing it expands the whole prompt. — 00:13:51 · Instagram
  • ★★ Applied to your own work, Ali says, by month three you will have a prompt library whose prompts keep improving as you add context. — 00:14:20
  • ★★ With a prompt library you can test the same prompts on different AI models, such as a trial of ChatGPT 5.1, and you find that certain models work best with certain prompts, Ali says. — 00:14:49 · ChatGPT
  • ★★ Ali pays for subscriptions to ChatGPT, Claude and Gemini because he finds them incredibly useful. — 00:15:18 · ChatGPT, Claude
  • ★★ For work that a text chat cannot do, such as slide decks, Ali points to AI tools like Gamma, Beautiful.ai and the AI generation in Figma Slides. — 00:15:18

00:15:48 — Phase 5: AI as infrastructure, four levels of automation

  • ★★ Don't get overwhelmed by the hundred new AI tools a week, Ali says: find the AI tools that help with your specific use cases at work or in your business. — 00:15:48
  • ★★★ Phase five of learning AI is AI as infrastructure, set up once to run in the background without you; Ali expects it from month four onwards and says you could spend years going deeper. (hedged) — 00:15:48
  • ★★ Ali's first level of AI automation is the automation built into tools you already use. — 00:16:46
  • ★★ Ali's editors edit in Premiere Pro with the AI plugin Firecut, which automatically generates a transcript and can put it into Google Drive. — 00:16:46 · Google Drive
  • ★★ Ali's second level of AI automation is simple connector tools such as Zapier or Make, for example sending each new Zoom recording's transcript through ChatGPT with a library prompt and posting the result to Slack. — 00:16:46 · Zoom, ChatGPT, Slack
  • ★★ Ali's third level of AI automation is a more powerful tool such as n8n, which he has been playing with a lot: more granular control and more sophisticated workflows, but more technical knowledge needed. — 00:17:15
  • ★★ Ali's team is trying to build an automation for the Lifestyle Business Academy in which every student coaching call is automatically recorded and transcribed with Grain and processed weekly. — 00:17:45 · Lifestyle Business Academy
  • ★★ The automation would combine coaching transcripts, the students' one-on-one Slack support and their place in the Notion CRM into a weekly report per student of wins, struggles and support needed, a manual job that now takes hours every Friday. — 00:18:14 · Lifestyle Business Academy, Slack, Notion
  • ★★ Ali's fourth level of AI automation is building your own AI apps, not to sell, which is hard, but as internal tools for your workplace or business. — 00:18:43
  • ★★ Much of advanced AI automation is overkill, Ali says: for most things you don't need, at least for now, to go much beyond learning to connect apps with Zapier. — 00:18:43
  • ★★ With automation the discipline is deciding what is worth automating, what is worth doing manually, and better still which processes you don't need and can delete, Ali says. — 00:19:19

00:19:40 — Pointer to his video on his own AI workflow

In the description and comments

  • ★ Ali first made his five-phase AI training for his own team, then realised it might help anyone. — video description

Threads in this video

Where the stories that moved here went next

Mentioned

Central: 10-80-10 rule · artificial intelligence

Also: People: Angus (general manager) · Dan Martell

Projects: Lifestyle Business Academy

Works: Buy Back Your Time

Organizations: Anthropic · Brilliant · Google · Instagram

Products: Adobe Premiere Pro · Beautiful.ai · ChatGPT · Claude · Fathom · Figma · Firecut · Gamma · Gemini · Google Drive · Grain · Grok · Make · n8n · Notion · Slack · TextExpander · Wispr Flow · Zapier · Zoom

Also centred on artificial intelligence: AI wrote this Short 🤖 #shorts · How I use AI to save 10+ hours per week · The Ultimate Beginner's Guide to Claude Code · How to Learn So Fast People Assume You’re Naturally Gifted (with AI) · How a plastic surgeon could use AI

← Previous: 5 years from now, today won't matter · Next → Managing 3 girls in their 20s be like…