What product teams become in the age of AI

The lazier version of the AI conversation asks which roles will survive.

The more useful version asks what the work becomes.

For product teams, that distinction matters. AI is already compressing large parts of the mechanics of modern knowledge work. Microsoft and LinkedIn's 2024 Work Trend Index found that 75% of global knowledge workers were already using generative AI at work, while 79% of leaders said AI adoption was necessary to remain competitive. Yet 60% of leaders also said their organisation lacked a clear vision and plan for implementation. In other words, adoption is happening faster than operating models are adapting.

This is why the question is not whether product teams matter in an AI-enabled organisation. They do. The question is where the scarce value moves once the production of documents, drafts, prototypes and code becomes faster, cheaper and more widely distributed.

The answer, I think, is upward.

Less value will sit in formatting, translating, documenting and manually producing first drafts. More value will sit in choosing the right problem, making sound trade-offs, maintaining coherence across the experience, protecting quality, interpreting signal, and deciding what should happen next. AI does not make product judgement less necessary. It makes poor judgement easier to scale.

Execution is becoming cheaper

The evidence for this is already strong enough to move beyond speculation.

In a controlled experiment published by Microsoft Research, developers using GitHub Copilot completed a standardised coding task 55.8% faster than the control group. That does not mean all software development is suddenly effortless, nor that speed is the only metric that matters. But it does mean a meaningful part of implementation work is now materially more compressible than it was.

The same pattern is showing up across knowledge work more broadly. Microsoft's Work Trend Index reported that employees using AI said it helped them save time, increase creativity and focus on more important work. Productboard's 2025 survey of 379 enterprise product professionals found that 94% use AI daily or often, and respondents reported saving an average of four hours per task across core product activities. The tasks most affected were not especially surprising: presentations, PRDs, competitive research and roadmap creation.

That is the first important shift. A great deal of product work has historically been consumed by translation: translating customer language into requirements, strategy into documents, ideas into prototypes, concepts into tickets, priorities into decks. AI is getting increasingly good at the first draft of all of that.

Which means the centre of gravity moves.

The value shifts from production to judgement

When the cost of producing a draft falls, the premium rises on deciding whether the draft is worth pursuing.

That may sound obvious. It is still easy to underestimate.

For years, product roles accumulated a large administrative surface area: writing, reframing, coordinating, formatting, summarising, chasing, packaging. Some of that work matters. Much of it merely sat between thought and action because the tools were clumsy and the organisations were fragmented.

AI strips away some of that friction. But it does not remove the need for product teams. It exposes what they are actually there to do.

They are there to exercise judgement.

To decide which customer problem is strategically worth solving. To determine which signal is real and which is noise. To see when something technically possible is commercially pointless. To understand how one decision affects the wider system. To maintain coherence across an experience that can otherwise become fragmented very quickly.

That is why the future of product work is not less demanding. It is more exacting. It asks less for administrative stamina and more for sharpness.

"AI does not make product judgement less necessary. It makes poor judgement easier to scale."

Product management becomes more strategic, or it becomes procedural

This is probably the most significant implication.

A product manager who is primarily a document machine is now highly exposed. If the role is mostly writing tickets, repackaging stakeholder requests, running status updates and polishing roadmap language, AI will increasingly outperform the human on speed and often on adequacy.

The role survives by moving upward.

That means stronger problem framing. Better prioritisation. Clearer economic judgement. Better synthesis of customer insight. More rigorous sequencing. More confident decision-making under uncertainty.

Productboard's 2025 survey is useful here. When product professionals were asked which skills were becoming more important as AI spread through the function, the top responses were data literacy, synthesising customer insights, systems-level thinking and strategic thinking. In other words, the valuable work is moving away from paperwork and toward interpretation.

This is the quiet shift AI is forcing on product management. It is making the mediocre middle harder to hide in. If the mechanics of PM work become faster and more automatable, then the differentiator becomes the quality of thought.

Not whether a PM can produce a PRD. Whether they can make a good call.

"As execution gets cheaper, coherence becomes more valuable."

Product design becomes more editorial

Design will change too, though not in the simplistic way people often predict.

AI will make it easier to generate screens, flows, variants and interface concepts. That does not make design irrelevant. It makes design less about producing isolated artefacts and more about governing the quality of the whole.

Figma's 2024 AI design report captures this transition well. It found that 89% of respondents expected AI to affect their products and services within 12 months, yet fewer than half of those working on AI features had actually launched anything. It also points to the growing importance of context, storytelling, persuasion and choosing between options — not merely generating them.

That is why I think design becomes more editorial in the age of AI.

Less time will be spent being the sole gatekeeper to pixels. More time will be spent ensuring that what gets made is coherent, legible, trustworthy and appropriate to the product's wider logic. Designers become more responsible for experience integrity: does this make sense in the broader journey, does it respect the user's context, does it introduce inconsistency, does it degrade trust, does it solve elegantly rather than merely exist?

As generation becomes easier, taste matters more.

So does restraint.

Engineering becomes more architectural

Engineering shifts in a similar direction.

AI will not remove the need for engineers. But it will continue to change the ratio between typing and judging. Between writing and governing. Between local implementation and system-level accountability.

The GitHub Copilot experiment is one useful signal, but McKinsey's 2025 work on AI suggests the broader organisational point: many companies are beginning to redesign workflows and governance around AI, yet fewer than one-third report following most of the adoption and scaling practices needed to capture value at scale. That implies a widening gap between using AI tactically and operating well with it structurally.

This is where engineering becomes even more important, not less. Someone still has to hold the line on architecture, reliability, observability, maintainability, security and system integrity. Someone still has to distinguish between code that functions and code that should exist in production. Someone still has to ensure that the convenience of acceleration does not quietly create a brittle system.

As AI lowers the cost of generating code, the quality of technical judgement becomes more consequential.

The work moves upward here too.

The operating model has not caught up yet

This is the part many companies are avoiding.

They are adopting AI, but not redesigning work around it.

Microsoft's Work Trend Index shows strong employee-led adoption. McKinsey's 2025 State of AI research shows that organisations are still early in putting in place the management practices required to scale value from generative AI. Fewer than one-third of respondents said their organisations were following most of the relevant adoption and scaling practices, and fewer than one in five said their organisations were tracking KPIs for gen AI solutions.

That is a familiar pattern: tool enthusiasm without operating-model maturity.

In product teams, it shows up as isolated experimentation rather than role redesign. AI is used to draft documents, summarise interviews and generate ideas, but the organisation still evaluates roles as if the old mechanics remain the centre of the job.

They are not.

If AI reduces the cost of producing artefacts, then the team should not simply produce more artefacts faster. It should become more selective, more strategic and more accountable for the quality of its choices.

Otherwise the organisation will merely accelerate clutter.

What good product teams become

The strongest teams, I think, will become more leveraged and more opinionated at the same time.

Product managers will become more strategic and more economically literate. Designers will become more responsible for coherence, judgement and trust. Engineers will become more architectural, more supervisory and more accountable for production quality. And all three will need to be much better at synthesis.

This is the part of the AI conversation I find most underdeveloped. People speak as if the main question is whether humans remain in the loop.

Of course they do.

The more interesting question is whether the humans in the loop are good enough.

Because AI does not remove the need for discernment. It raises the bar for it.

"The role survives by moving upward."

A simple test

A good way to judge whether a product team is adapting well is to ask what has changed in its definition of excellence.

Is the PM still being rewarded for document production, or for decision quality? Is the designer still being valued mainly for output volume, or for experience coherence and judgement? Is the engineer still being measured by lines shipped, or by the resilience and integrity of the system? Has the team become more thoughtful as execution gets faster, or merely more prolific?

Those questions matter because AI can make mediocre teams look busy very efficiently.

What it cannot do is give an organisation taste, strategic judgement or a coherent point of view. Those remain stubbornly human responsibilities.

Which is why product teams do not disappear in the age of AI.

But they do lose the luxury of being vague about where their value lies.

The work becomes less about producing things manually and more about deciding, with precision, what deserves to be produced at all.

That is the shift.

And it is not a reduction in the importance of product teams. It is a clarification of it.

"AI can make mediocre teams look busy very efficiently. It cannot give them taste."