I don't get asked about AI much, which still surprises me....
And when I do, I can usually feel how quickly people want the conversation to be over.
The coverage is noisy. The tools change constantly. Most people already feel overloaded.
Still, when it comes up, it tends to be with people I like. Smart friends. Founders. People who are good at what they do.
They'll say, "How's your AI stuff going?" and you can tell they're hoping for a one-line answer.
So I give them the polite version.
'AI is moving fast. If you have real domain expertise, your output can jump in quality when you learn how to use the tools properly. I've never felt more creative. I can finally build what I had in my head.' I'll also add the reassuring line: 'AI won't replace humans. It'll be human plus AI. It'll become part of how we work.'
That's the polite version. The honest version is harder to say out loud.
My journey with AI started last August, and I still feel behind. I don't think anyone feels up to date. It moves too fast. The distance between an idea and a working prototype is collapsing. I know this because I've experienced it first-hand.
Last September I tried to build an app that would add brand graphics to video assets. It was a steep learning curve. To get anything working, I had to understand databases, APIs, FFmpeg, and how the parts connect. My lack of technical knowledge was the bottleneck. Today, I could build a version of that in under an hour.
Not because I became an engineer. Because more of the technical setup now happens quietly in the background.
A few weeks ago I built a tiny app that receives my AI newsletters each day, lets me swipe through them tinder-style, and includes text-to-speech so I can listen as a podcast on the way to the tube. That took about three prompts. A year ago, I wouldn't have been able to do that at all.
Once you've seen what's possible, you start seeing automation everywhere.
I no longer wonder if something is possible. I assume it is, and I start thinking about how I'd build it. I now have an 'AI setup' that covers work I would have had to spread across an accountant, a research function, and a CRM. I've also noticed I've stopped looking for software that does one specific task. More and more, I'd rather build a simple version myself that fits what I need.
And that's the part that worries me. Because the people doing those jobs often cannot see what's coming.
So why are so many people still dismissing it? A few misconceptions come up again and again.
First: "AI isn't that good." For a lot of people, that opinion comes from one bad test. They tried it once in 2024, asked a generic question, got a generic answer, and decided that was the ceiling. Or they only use the free tier like a search engine and assume it reflects what the tools can really do. It doesn't.
Second: "AI output is generic." AI output reflects your input. If your direction is vague, the output will be vague. If your direction is clear, the quality improves fast. The Coca-Cola AI Christmas ad is a good example. It got a lot of hate. That wasn't a technology limitation. It was a weak ad. A poor creative decision is a poor creative decision, with or without AI.
Third: "You have to be technical." This is the biggest barrier, because it stops people from starting. You don't need technical skill. You need curiosity, reps, and the willingness to iterate until the output is usable.
And starting now doesn't mean you're "six months behind". Yes, I can push further today because I understand more of what's happening. But the tools have changed. In September, I had to stitch components together just to get a basic version of something working. That knowledge mattered then. The technology is moving so fast that a lot of it is already obsolete.
So if you're hesitating because it feels like too much to learn, you're thinking about it the wrong way. Start now, because the barrier is dropping.
If you start now, you can still learn the fundamentals while they're still visible.
Cost versus quality.
Speed versus accuracy.
Wrapper versus underlying capability.
How to set up a simple automation.
Why context matters more than clever prompting.
The people who learn those things now will be the ones who can lead with this technology inside a business. They'll be the ones who can choose the right tools, set the right expectations, and get reliable outcomes.
If you arrive once the barriers are gone and everything is hidden behind the scenes, you'll still be able to use AI, but you won't get the same chance to learn what goes into driving the best results.
So don't wait to start. Not because you're behind. Because you're early enough to learn the fundamentals while they still hold their value.
Now the uncomfortable bit.
More and more, I can see roles that can be automated or radically restructured. That has nothing to do with whether people are good at what they do. It's because a lot of modern work centres around drafting, summarising, researching, analysing, rewriting, reporting.
These are the areas the tools are strongest, and they're getting better all the time. This is why I believe the timeline is short.
From where I sit, we have 6 to 12 months before the shift becomes undeniable. It will vary by sector, but the pressure is the same everywhere. Smaller, more adaptable businesses will produce better work with fewer people and shorter cycles. That becomes visible. Then it becomes unavoidable.
For leaders, what changes first is a capability gap inside the organisation.
They'll realise their workforce is strong at yesterday's tools, but not equipped for what's now available. They'll start to notice which employees become more valuable, and which become harder to justify. They'll see competitors shipping more, testing more, learning faster, often with smaller teams.
So what stays valuable?
Knowing what good looks like, and being able to get there reliably. Clear direction. Accountability. Trust and relationships. Distribution. The ability to teach others how to use these tools properly. And the ability to implement change so the business actually improves, not just talks about AI.
If your work happens on a screen, assume the shape of your job or business is going to change fast.
For employees, the safest move is to become the person who can use these tools properly and show outcomes. Not because it's trendy. Because it makes you harder to replace and easier to promote.
For founders, it's the same shift in a different form. Smaller teams will ship faster, test more, and compete with companies that used to have far more resources.
Either way, this is a moment to take your own positioning seriously.
If you take anything practical away from this,
Do three things this week:
Pay for ChatGPT Plus or the Claude Pro plan. Don't judge this from the free tier.
Use ChatGPT or Claude once a day for seven days on something real. Iterate until you get a result you can use. Save the prompt and reuse it.
Build one small thing. Use a tool like Lovable and create a simple app. Ask AI to help you refine the idea, hit 'build' and watch it come together in front of you.
If you do that, you won't just "understand AI". You'll understand what has changed.