Hey! Ankur here, and this is the 22nd edition of Lazy AIthe one where doing less is the whole strategy.

Yes, the email subject seemed counter-intuitive. An AI newsletter saying DON’T keep up with AI?

But hear me out..

Today’s piece is an attempt to calm the panic that hits when you open LinkedIn and every third post is about an AI tool you’ve never used.

Open any feed for 30 seconds and you’ll see it — someone teaching prompt engineering, someone evangelising agentic workflows, someone showing off a Veo 3 clip, someone selling a Claude skills course. Every one of them is good. Every one of them is making you feel like you should be doing what they do.

So you start dabbling. A bit of this, a bit of that. Two weeks later you’ve forgotten most of it, a new tool drops, and the cycle repeats. I’ve done this. In fact most days, I’m still doing it.

But the panic is built on two assumptions that fall apart the moment you examine them.

The two assumptions

Assumption one: the things you’re missing today will still matter tomorrow.

They won’t. Most of them. Because AI tools have a brutally short half-life — most of what’s trending this month is forgotten in month twelve.

Half-life is the time it takes for something to lose half its relevance

Think about it. 18 months ago, people were paying for ChatGPT prompt packs on Gumroad. Entire newsletters were dedicated to AutoGPT. “Prompt engineer” was a real job title at high salaries.

Today? Prompt packs are obsolete. AutoGPT is a footnote. Prompt engineering as a standalone role is mostly gone. And Sora — the AI video tool everyone was obsessed with six months ago — already got shut down.

Look at what’s trending right now. Some of it will matter in 2027. Most of it won’t. You just don’t know which is which — and neither does anyone teaching it on LinkedIn (this is why everyone sounds so confident — confidence in this space is mostly performance).

Spread your 5–7 hours a week across six trending things, and you’re memorising train timetables that get rewritten every quarter.

You don’t need to be an engineer to talk AI. Subscribe to this newsletter — every other day, you’ll understand AI well enough to be the smartest person in the room at work.

Each issue is just 5 minutes — less than the time you spend doomscrolling before bed. Except, this actually moves your career forward. Join 8,000+ subscribers now.

Assumption two: Each new AI skill is a separate thing you should learn from scratch.

Also wrong. And this one’s the good news.

The skills underneath AI tools are supremely transferable. Spend 100 hours building Claude skills — really learning how to scope a model’s behaviour, structure instructions, handle edge cases — and you’ve also picked up most of what you need for agentic workflows, custom GPTs, and even Claude Code. Because at the core, they’re all the same skill: Telling a model precisely what you want, in a way it can actually act on.

Same with image and video generation. Get genuinely good at one model — Midjourney, Veo 3, Nano Banana, pick your weapon — and the iteration loop you’ve internalised (prompt, judge the output, adjust, repeat) is exactly the loop you’ll use for every visual model that ships next year. The interface changes. The instinct doesn’t.

The generalist stays a beginner everywhere. The deep person becomes a fast learner everywhere. Those are very different futures.

Here’s what actually bothers me

We’ve been sold the idea that staying current with AI means staying broad. It doesn’t. Broad familiarity is what everybody already has. It’s worth nothing.

And here’s the part that should make you exhale — the depth you need is probably already sitting in front of you.

According to a study conducted last year, 66% of ChatGPT users only use it for basic stuff. They type a question, copy the answer, close the tab. They’ve never set up a project. Never written a custom instruction. Never used artifacts properly. Never chained two prompts. Never built a reusable system prompt for the work they do every single day.

The gap between a casual user and a deep user of the same tool is enormous. Way bigger than the gap between a casual user of one tool and a casual user of five.

So the highest-leverage move isn’t a new tool. It’s going embarrassingly deep into the one you already open every day.

You don’t need to be an engineer to talk AI. Subscribe to this newsletter — every other day, you’ll understand AI well enough to be the smartest person in the room at work.

Each issue is just 5 minutes — less than the time you spend doomscrolling before bed. Except, this actually moves your career forward. Join 8,000+ subscribers now.

So what should you take from this?

Tonight, don’t bookmark another tutorial. Just open the AI tool you already use. Ask yourself how many of its features you’ve actually pushed against. Projects? Custom instructions? Artifacts? Memory? Whatever your tool’s version of those is.

That’s where your depth is hiding. Not in the next thing on your feed.

You’re not behind on AI. You’re behind on the AI you already have. That’s a much smaller, much kinder problem to solve.

Breathe. It’s okay.

If this got you thinking, please share it with someone who keeps bookmarking AI tutorials they’ll never finish.

It takes a lot of time to research and write this. Sharing means a lot 🙂

See you next time..

Cheers,

Ankur

Keep Reading