Crixus Blog

I recently prepared a talk that opened with a single question:

If AI can do everything — what do you still need us for?

I didn’t answer it right away in the talk. I won’t answer it right away here either. But I’ll start where I started there: with a clarification that sounds obvious and isn’t.

You’re already a product manager

If “product management” makes you think of people with roadmaps and Jira boards who have nothing to do with you, I have to disappoint you. Product management simply means: making sure the right thing gets built at the right time for the right people. Not that something gets built — that’s project management (And I repeated multiple times that I am not a great Project Manager). But that it’s the right thing.

And by that definition, you’re all running the same loop, whether you call it that or not:

Understand → Decide → Build → Show → Learn → and back to the start.

Your agency is a product. Every client website is one. Your plugin, your hosting offer, your newsletter, your own blog, this post… all products, all with an audience, an expectation, and exactly this loop. Usually without anyone calling it that.

Remember this loop. We’ll come back to it at the end of this series, and it’ll give up more than it promises today.

“Code is free” is true

The sentence you hear at every conference right now: code is free. And it’s true. Building costs almost nothing anymore.

But it’s thinking too small. It’s not just code:

• Design — free
• Copywriting — free
• Research — free
• Data analysis — free
• Prototypes — free
• Testing — free

All of that might not yet be perfect or on the level of a master in their class but it’s doable.

Everything executed has become nearly free over the last twelve months.

And yes — before I go on, I want to say this plainly: part of our work genuinely isn’t needed anymore because of this. Anyone who was paid to implement something someone else had decided has a real problem right now. Let’s not going to sugarcoat that.

The interesting question is: what’s left? Look at the loop again. For four of those five steps, AI can now handle the execution. For one, there’s nothing left to hand off.

And that one thing is harder than it sounds — because we never really made it consciously in the first place.

The filter nobody noticed

Here’s the insight the rest of this series is built on:

We thought we’d always been deciding. In reality, cost was deciding for us.

You always had ten ideas and time for two. A feature cost two weeks of developer time, so you thought twice about it. A new skill on the team cost a hire, so you had to prioritize. The filter that screened out nine of ten ideas was never a conscious decision. It was simply scarcity. We just pretended, after the fact, that we’d weighed things carefully.

AI didn’t take work away from us. It took the filter away.

And a filter that disappears, disappears quietly. You only notice once everything starts slipping through.

The picture: a jar full of marbles

Picture a jar. Every marble in it is a discipline a product needs: engineering, design, sales, marketing, UX, requirements. You used to have to hire or commission someone for every single marble. That was the filter I just described.

AI hands you every marble today. All of them. Instantly. Free.

Is the jar full now?

No. There’s always air between the marbles. The water that fills those gaps is product management itself: the connective tissue, the direction, the decision about what belongs together and what doesn’t. That’s why product management looks different at every company, and the marbles are never distributed the same way.

You get every marble handed to you for free and think the jar is full. It isn’t. It’s just fuller.

The thesis

Code is free. Decisions aren’t.

Cheap building doesn’t make decisions cheaper. It makes them more expensive — because you’re now making so many more of them. Feasibility used to be the filter. That filter is gone. What’s left is judgment.

I learned exactly this the hard way a few months ago, building a product from scratch with a colleague at Automattic during our Radical Speed Month. Four weeks, zero developers on the team, just the two of us and a lot of AI.

That’s the story in the next part.