What changes when you plug AI into your product process

The biggest change AI made to how I work wasn't the one I expected.

I thought it would be research. Faster market analysis, quicker competitor reviews, better synthesis of user interviews. All of that is true and all of it is useful. But the change that actually shifted how I work day to day was simpler: AI removed the friction between having information and being able to think clearly about it.

The messy middle

Every product decision has a messy middle. The part where you have the inputs — the brief, the research, the feedback, the competing opinions — but you haven't yet found the shape of the answer.

That part used to take longer than it should. Not because the thinking was hard, but because the organising was hard. Getting everything in one place. Finding the through-line. Working out what actually matters versus what's noise.

AI compresses that. Not by doing the thinking — it doesn't — but by getting you to a starting point faster. A rough synthesis, a first draft of the argument, a structure you can push back against. The output is never right first time. But working from something concrete is faster than working from a blank page, every time.

Where it changed specific parts of the process

Discovery used to mean a day of synthesis after every round of user interviews. Patterns, themes, quotes — all manually organised before you could draw any conclusions. Now that synthesis takes a fraction of the time. The conclusions still require judgment. The organising doesn't.

Narrative and presentation work changed the most. Taking scattered project files — briefs from different sources, research at different stages, stakeholder feedback in different formats — and getting to a coherent story used to take most of a day. AI cut that significantly. First draft is never right. But it's a real draft.

Spec writing got faster too. Not because AI writes the specs — the judgment calls about scope, edge cases, and acceptance criteria still require someone who knows the product — but because the structural work around a spec is faster. The format, the sections, the questions you need to answer. That scaffolding is quick now.

What didn't change

The hard decisions are still hard.

Whether the problem is real. Whether the user insight holds up or whether it's what you wanted to hear. Whether the business case makes sense. Whether the scope is right. Whether you're solving the right thing at all.

AI has no opinion on any of that. And if you use it as a shortcut past those questions — if you let a well-structured output substitute for genuinely answering them — you'll move faster toward the wrong answer.

The judgment calls belong to you. AI just gets you to the point where you have to make them faster.

The honest version of what it's worth

AI is worth a lot if you already know how to think about product problems. It removes friction, compresses timelines, and gets you to clarity faster than you'd get there alone.

It's worth less than you think if you're hoping it will do the thinking for you. It won't validate your strategy. It won't tell you whether you have PMF. It won't catch a flawed assumption dressed up in a well-written paragraph.

Use it to move faster. Keep the judgment. That's the honest version.

If you want the practical decision-making angle, I wrote about how AI actually helps product decisions. And for the broader strategic framing, AI won't build your product — but it will sharpen your strategy.