How does AI change PM-UX-Tech workflow?

Surprise surprise... AI doesn't make the trio obsolete.

First of all lets understand why does PM-UX-Tech trio even exist in product organizations.

Not because someone drew an org chart and decided to split the work three ways.

Each role exists because building something useful for real people is genuinely hard in three distinct ways.

Someone needs to understand the problem deeply enough to know which problem is even worth solving for the customer, and for the business. That is the PM. Not because they are the smartest in the room, but because holding the problem is a full time job on its own.

Someone needs to understand people well enough to know how a solution will actually feel to use. That is UX. Not just aesthetics. The logic of how a human moves through something, where they get confused, where they give up, where they finally feel understood.

Someone needs to understand the material well enough to know what is actually buildable, at what cost, with what tradeoffs. That is the engineer. Constraints are not obstacles. They are the thing that shapes what the solution actually becomes.

Three hard problems. Three specializations. That is why the trio exists.

In today’s issue we talk about how AI might change the collaboration between Product, Tech and UX.

Before that, hear from our partner Fabi, which lets you get data and dashboards with plain English questions.

---

Your Data’s Spread Across 6 Tools. Your Answers Shouldn’t Be.

HubSpot, Stripe, Attio, Google Ads, spreadsheets, your data warehouse… pulling it all together is a full-time job you didn’t sign up for.

Fabi connects your entire stack and lets you ask plain-English questions: Which campaigns are working? What’s driving churn? You get answers, charts, and full dashboards in minutes. Just describe what you want and Fabi builds it.
No SQL. No legacy BI. No waiting.

Prefer to work in Slack? You can ask questions right there as well.

Don’t forget to use code PRODUCTIFY for 20% off your first 3 months.

---

Now back to today’s newsletter.

Before AI, most product teams ran on a simple contract.

The work moved in one direction.

PM defines problem  →  UX designs solution  →  Tech builds it  →  Ship

Each role had a lane. Hand-offs were the norm. So were the arguments that came with them. The engineer sitting out until the spec was ready. The designer seeing feasibility problems nobody flagged early. The PM solving a problem tech could have shaped better from day one.

Waterfall between PM, Tech and UX was already costing teams before anyone named it as the thing to fix.

Then AI arrived. And the lanes did not just blur. They started to feel optional.

Here is what actually changed.

A PM can now generate a wireframe in minutes. A designer can write a working prototype without touching a developer. An engineer can ship a UI without a single design review.

And so a tempting idea started spreading.

Maybe one good person with the right tools can just... do it all.

It sounds efficient. It feels modern.

It is also, in most cases, a quiet way to ship worse work faster.

Here is why that assumption breaks down.

When one person collapses all three roles, they stop questioning themselves. The PM who designs never gets told the flow is confusing. The engineer who makes product calls never gets challenged on whether this is even the right thing to build. The UX person who scopes the feature never gets pushed on feasibility.

AI removes friction. But some friction was load-bearing.

The arguments between PM, UX and Tech about what the user actually needed were not inefficiencies. They were how bad assumptions got caught before they became bad products.

So when is solo work with AI actually fine?

When the stakes are low and scope is small. An internal tool. A quick experiment. One screen. One flow. A proof of concept nobody will rely on yet.

When you are validating, not building. In these moments, AI-enabled solo work is a superpower. Use it.

And when does the full trio still matter?

When interpretation matters more than execution.

Low-risk does not mean low-interpretation. You can run a small experiment with five users and nothing riding on it. But what you see, record, and conclude is always filtered through whoever is in the room. AI can help one person do more. It cannot help one person think differently than themselves.

The trio earns its place when decisions are hard to reverse. Core user journeys. Architectural choices. Anything expensive to undo. And when the team needs to own the outcome together, because work produced alone does not automatically earn the confidence of the people who have to maintain it.

What the workflow looked like. And what it looks like now.

Before AI, the flow was linear and slow.

PM writes brief  →  UX designs (2 weeks)  →  Tech flags issues  →  Loop back  →  Ship

Each person solved in isolation. By the time all three were aligned, a month had passed and the problem had already shifted.

Now, with AI, the same feature starts differently.

PM brings prototype  →  Trio reacts together  →  UX generates directions  →  Tech stress-tests  →  Align early  →  Ship

Each person shows up differently now.

The PM arrives with a rough AI-generated prototype instead of a brief. Something tangible to react to instead of imagine. The job is to make the problem visible and think out loud, not to pre-solve the experience. The moment the prototype starts answering UX questions, put it down and hand the problem over.

The UX person does not wait to be handed a brief anymore. They arrive with AI-generated user flows, rough concepts, or early research synthesis pulled together before the first conversation happens. Three directions where there used to be one. Questions about the experience formed before the room even meets.

The engineer comes not with an estimate but with a proof of concept. A quick AI-assisted spike that already tells the team what is feasible, what is risky, and where the hard edges are. Constraints surface in the first conversation instead of the fifth.

When all three show up this prepared, nobody is waiting. Nobody is reacting to just one person’s framing. The conversation starts in the middle instead of at the beginning.

Disagreements surface faster. Bad ideas die earlier. The good ones get sharper, sooner.

So no, AI did not make the PM-Tech-UX trio obsolete.

It made solo work viable in the small moments and collaboration more important in the ones that matter.

But it also handed every role a new kind of power. And with that comes a new kind of restraint. The PM who can now design must know when to stop designing. The engineer who can now prototype must know when to hand the problem back. The UX person who can now scope must know where their lane still ends.

AI gave everyone a longer reach. It did not give anyone a better reason to overstep.

The question is not, can I do this alone now?

It is, what is the cost if I interpret this wrong?

Answer that honestly, and you will know exactly when to work solo and when to bring the room back together.