The Rise of the AI Product Engineer

The hybrid role quietly replacing the traditional PM — and what it means for your career in 2026.

By Lukas Kaminskis — CEO & Co-founder of Turing College

There's a role that has been rising in tech hiring for the last eighteen months without most people noticing it had a name. You see it in job descriptions that ask for product instinct and shipping ability in the same paragraph. You see it in compensation bands that don't fit cleanly into either the PM or the engineer ladder. You see it in the kinds of people getting hired into senior roles at AI-native companies, who used to be PMs and now spend half their week in Cursor.

The role is the AI Product Engineer, and its rise is the single most consequential shift in product work since the PM role was invented.

I want to make the case for why it's happening, what it actually is, and what a senior operator should do about it before the market sorts everyone into the people who made the transition and the people who didn't.

What is an AI Product Engineer?

An AI Product Engineer is a hybrid operator who can specify, build, and ship a product with AI tools — combining the judgement of a product manager, the technical fluency of a software engineer, the taste of a UX designer, and the analytical instincts of a data lead into one person. The role exists because AI has collapsed the time between idea and working prototype from weeks to hours, and that collapse rewards generalists who can hold all four disciplines at once.

The title varies by company. Some call it Product Engineer. Some call it AI Engineer with product responsibility. Others still call it Senior PM, but the job description now quietly requires shipping working code. The label matters less than the underlying shift in what companies are paying for.

Why this role is rising now

The role is rising because the economics of building changed first. When one person with AI tools can produce what used to require a small team, the bottleneck moves from execution to judgement. Companies are now paying more for fewer people who can do the whole loop themselves, and paying less for specialists whose work AI now automates. The hiring data has started to reflect what the building data already showed.

According to Gartner's forecast, 75% of new enterprise applications would be built using low-code or no-code platforms by 2025. That number was under 25% in 2020. A four-year jump from a quarter of new applications to three-quarters. Most industries do not change that fast. Software does.

What's surprising is how slowly the staffing model has caught up. Most companies still write job descriptions designed for the 2020 building stack — separate PM, separate engineer, separate designer, separate data analyst — even though the work itself has consolidated. In my analysis of recent hiring conversations with senior operators, the question I hear most often is some version of: “Do I hire a PM with technical skills, or an engineer with product instincts?” The honest answer is that the two roles are converging, and the company that figures out how to develop one person to do both will move faster than the company still hiring two.

Nikhyl Singhal, formerly of Meta and Google, made a sharper version of this point on Lenny Rachitsky's podcast. The PM whose day is spent moving information around — writing status updates, reconciling roadmaps, translating engineering shorthand for marketing — is, in his framing, “essentially going to become a dinosaur.” Over the next 12 to 24 months, he predicts, companies will experience a massive shedding of staff followed by rehiring, where the new hires will all be “AI first.” That is not a prediction about layoffs. It is a prediction about replacement. The same headcount, possibly fewer, but with a different skill profile.

The AI Product Engineer is the shape of that new skill profile.

The Venn diagram: four disciplines collapsing into one role

The AI Product Engineer is what happens when four older specialisms — AI/software engineering, product management, UX/UI design, and data analysis — stop being separate jobs and start being four skills held by one person. The diagram below shows where the role sits.

There's a subtle move inside Nikhyl's argument that I think gets missed in the broader discourse about AI replacing knowledge work. He doesn't say information work is dying. He says the mover of information is. Those are two different claims. Information still needs to flow between specifications and code, between user research and roadmap, between metric and decision. What's changed is that the cost of moving it has dropped to near zero. Which means the value of the judgement applied to it has gone up.

This is what the four-circle diagram is really showing. Each discipline still exists. AI engineering, product management, design, data analysis — all of them still required. What's collapsed is the assumption that each one needs its own dedicated human. When the mechanical parts of every discipline get automated, what remains is the judgement layer at the centre, and one person with enough range can hold the judgement for all four.

Old PM vs. AI Product Engineer: what actually changed

The old PM role optimised for coordination across a team of specialists. The AI Product Engineer role optimises for end-to-end ownership with AI doing the specialist work. Same goal — shipping a product that matters — but the org chart and the day-to-day look different. The table below shows where the work moved.

The last row is the one that matters. Everything above it explains what AI now does. The last row explains why humans are still hired.

Why judgement became the scarce resource

When AI can produce a working prototype in an hour, the bottleneck shifts from execution to evaluation. The hard question stops being “can we build this?” and becomes “should we?” Judgement — knowing what to build, recognising whether the thing you built is good, and deciding how it fits into the broader system without damaging the brand — has become the highest-paid skill in product organisations.

Nikhyl put this almost mathematically. He cites Einstein: genius is 1% inspiration, 99% perspiration. Then he updates the quote for the current moment. In the AI era, AI handles the 99% perspiration. Humans get to focus, completely, on the 1% inspiration. That sounds like a slogan until you sit with it. If a tool removes 99% of the labour from a job, the remaining 1% becomes the entire economic basis of the role. The 1% is also the part that's hardest to do, and hardest to teach.

The hiring market has noticed. Compensation for builders who can hold judgement across the full stack — product, code, design, data — is at an all-time high. Compensation for pure coordinators is not. This is the new equilibrium.

What the rise of this role means for mid-career operators

There's a generational risk in this shift that doesn't get named often enough. The people most exposed are mid-career PMs and engineers — exactly the cohort with the most institutional knowledge, the strongest product instincts, the deepest sense of what makes a product good. Their judgement is exactly what the market wants more of. But their hands have, in many cases, gone soft. They haven't shipped code in five years. They haven't designed a screen in eight. They haven't run their own analysis since the data team was hired.

Nikhyl is honest about this. Mid-career reinvention is hard. People are exhausted. They have mortgages and parents and kids and a life that already fills the day. The threshold to learn new tools — to actually open Claude or Cursor and build something for the first time in years — feels enormous when measured against the energy left over at 9pm on a Tuesday.

The reinvention is also non-optional. Or rather, it's optional in the same way that learning to use email was optional in 1998. You can decline. The market just stops paying for what you do.

How to apply this tomorrow

Five concrete moves a mid-career PM or engineer can make this week to start the transition from information mover to AI Product Engineer. None of them require quitting your job. All of them compound.

• Build one thing this week that solves your own problem. Not a tutorial. Not a course project. A small tool you actually want — an inbox triage script, a meeting-notes summariser, a smart-home dashboard. Nikhyl makes this point directly: PMs need to find joy in using new AI tools to build things. Joy is not a soft suggestion. It's the mechanism by which the reps stack up.
• Obsolete one of your own recurring tasks. Pick the most repetitive thing on your week — status reports, backlog prioritisation, weekly metrics roll-ups — and automate it with an AI tool. A great engineer or PM, in Nikhyl's framing, uses technology to obsolete themselves from their daily, repetitive work. The freed time is the entire point.
• Update your CV to lead with what you've built, not where you've worked. This is the brand-versus-modernity shift Nikhyl describes. Companies now care more about how current your skills are than which logos sit at the top of your CV. Put the AI builds first.
• Find a feedback loop with people doing this work daily. Whether it's an open-source community, a Discord, or a structured programme like our Building with AI cohort — proximity to other builders is how the tacit knowledge transfers. Reading about how to build with AI tools is not the same as building alongside people who do it for a living. If you're a German resident eligible for Bildungsgutschein funding, or a UK employee whose company pays into the apprenticeship levy, the programme costs you nothing — explore Building with AI here.
• Treat the next 12 months as a deliberate rebuild. Nikhyl predicts a massive shedding and rehiring over the next 12 to 24 months. The window to position yourself as one of the people being rehired, rather than one of the people being shed, is open now and won't be open indefinitely.

What about engineers? Does the same shift apply?

The same convergence runs in the other direction. As AI accelerates coding, engineers increasingly need product skills — deciding what to build, evaluating whether it worked, understanding the user well enough to make taste calls. The role isn't asymmetric. PMs need to become more technical, and engineers need to become more product-minded. They're meeting in the middle, and the middle is the AI Product Engineer.

This is the part that surprised me most when I started watching the hiring patterns. The shift isn't only about PMs adding code skills. It's about engineers adding product judgement. As AI takes over the mechanics of writing functions, the engineer's value moves to deciding which functions are worth writing — which is, by definition, product work. The two roles are converging because the work itself has converged.

Nikhyl makes a related point worth sitting with. He predicts AI Product Engineers will become “agents of change” — people who bring modern, AI-native building practices into non-tech industries. Marketing teams that need someone to build internal tools. Sales orgs that need automation. Legal departments that need document workflows. The hybrid role is about to be a cross-industry one.

FAQ

Is “AI Product Engineer” a real job title or just a buzzword?

It depends on the company. Some — Vercel, Linear, smaller AI-native startups — use the title directly. Others embed the same skill profile inside a Senior PM or Staff Engineer role without renaming it. The title varies; the underlying skill demand is real and rising. Watch what the job description requires, not what the title says.

Do I need to learn to code from scratch to make this transition?

You need enough technical fluency to ship working software with AI tools — not enough to pass a Google L4 coding interview. The relevant skills are reading code well enough to verify what AI produces, understanding system design well enough to make sensible architecture choices, and using tools like Claude, Cursor, and similar agents to write the actual code. The bar is “can you ship?” rather than “can you write a B-tree from memory?”

Will the rise of this role reverse if AI hype cools off?

The Gartner data showing 75% of new enterprise applications built on low-code or no-code platforms by 2025 was forecast in 2021, before the current generative AI wave. The shift toward fewer, more multi-skilled builders predates this AI cycle and is underwritten by it. A cooling of AI hype would change the timeline, not the direction.

What's the fastest path for a mid-career professional to make this transition?

Build something real, in public, every week — even if small. Combine that with a structured environment that forces reps and provides feedback from people already doing the work. Self-study alone is slow because the tacit knowledge of when to trust the AI, when to override it, and how to evaluate output quality only transfers through doing the work alongside others doing it well. For German residents and UK employees whose companies pay the apprenticeship levy, the Building with AI programme provides exactly this kind of environment at zero personal cost.

The rise of the AI Product Engineer is the most consequential career inflection point in product work since the role was invented. The good news is that the underlying skills — judgement, taste, building instinct — are not new. They are the same skills that have always defined the best operators in this profession. What's changed is that those skills now come with a build button attached.

For the people willing to press it, this is the most interesting moment in the history of product. For the people who don't, it's the moment the market quietly stops looking for them.