CODE IN TOKYO

Will That Hot New AI Tool Still Be Around in 5 Years?

The AI world has been moving at a frankly insane pace lately.

Every day, there’s a new “revolution.”

New AIs, plugins, frameworks, development methodologies. Open YouTube and you get flooded with videos like “You’re falling behind if you don’t know this.”

I get it. It’s genuinely fun.

I use AI heavily myself, and I honestly don’t think I could develop without it anymore.

But lately, I keep thinking:

Maybe everyone should calm down a little.

Will That Tool Still Exist in 5 Years?

That tool you’re frantically chasing right now.

How much of it will still be around in 5 years?

The IT industry has always been like this — things hyped as “the next generation” routinely vanish without a trace a few years later.

And in AI, that cycle is even faster.

Last year’s optimal solution is already outdated this year. With the models themselves evolving, entire tools can become unnecessary overnight.

I use Claude Code now, but whether I’ll still be using it in 5 years is anyone’s guess. Probably not.

So if “keeping up with the latest” becomes the goal itself, it gets pretty dangerous.

You spend the whole day just catching up.

And before you know it, you’ve built nothing.

Which is the worst.

Look at Your Actual Work Before Hunting for Plugins

I’m not the type to try every new plugin that drops.

I do touch new things, of course. But my baseline is:

“Fix what’s actually annoying me at work right now.”

For example:

If that’s the case, fix that.

In the end, the highest-value improvement isn’t the new tech trending on X — it’s reducing daily friction.

People Who Chase Everything Usually Build Nothing

This isn’t a “stop using AI” message.

The opposite, actually. Use it hard.

But the people chasing every piece of latest information? They usually haven’t built anything.

It’s almost a law.

Trying new tools every day, posting hot takes on X, watching “god update” videos on YouTube — they feel like they’re at the cutting edge, but from the outside, they’re just information-processing clerks.

And it gives them a “feeling of working,” so they’re the last to notice.

Pretty bad, right?

So lately, every time I see a new tool, I run it through one filter:

“Does this actually help my work right now?”

If yes, use it. If no, pretend I never saw it.

Don’t go further than that.

Knowing all the latest tools without building anything makes you “well-informed,” not “impressive.”