A few months ago, most conversations about AI in product still felt exploratory: pilots, proofs of concept, a lot of experimentation.

That phase is ending. What I’m seeing now, fairly consistently and across organisations, is AI moving directly into the product layer. Not as an experiment sitting off to the side but as something shaping how products are actually built, shipped, and iterated day to day.

When you look across enough examples, clear patterns start to emerge:

1. Product teams are getting smaller — and more powerful

Engineer-to-product ratios are shrinking, and output is increasing. The traditional model — one product manager orchestrating engineering teams of six to 10 — is starting to change. Now you’re seeing tighter, more autonomous groups where product, design, and engineering begin to blur.

In many cases, product managers are no longer just defining requirements; they’re building working prototypes themselves. Not rough wireframes but something tangible enough to align a team around. It changes the role fundamentally. Product moves from writing documents to creating things, and engineering becomes more about refinement and scale than starting from zero.

2. The build process is compressing — dramatically

The pace of building is accelerating. Work that was scoped across quarters is now happening in weeks, sometimes faster. AI is collapsing the early stages of development: prototyping, code scaffolding, documentation, testing.

The pattern is consistent: Teams can get to 70%-80% of a product very quickly, but the final stretch — integration, edge cases, polish — still takes time. Even so, the overall cycle is significantly shorter, and the cost of experimentation is much lower. Trying something — and abandoning it — is no longer a major investment.

3. PRDs are becoming alignment tools, not instruction manuals

The role of the Product Request Document (PRD) is evolving. Historically, it was a detailed instruction manual for engineering. Now it’s becoming lighter, more about alignment than execution. 

When product managers can build high-fidelity prototypes, teams align around something real rather than something described. In some cases, AI is even generating the documentation itself — summarising intent rather than dictating how something should be built.

It’s a shift away from abstract planning toward shared, tangible understanding.

4. AI is reshaping how engineering work happens

AI is becoming embedded into everyday engineering workflows. Writing and refactoring code, generating tests, improving documentation, even supporting legacy rewrites — all of this is starting to normalise.

What’s notable is the approach to adoption. Most effective organisations aren’t necessarily mandating usage for engineers, they’re framing it as a professional responsibility to explore. And the pattern is familiar: A small group resists, a quiet majority experiments, and, over time, it becomes part of the baseline.

The biggest gains are often in the least visible areas — the work that used to slow teams down.

5. AI is accelerating product thinking — not just building

AI isn’t just speeding up execution; it’s shaping how teams think. Product arguments are becoming clearer, faster. Teams are using AI to explore multiple directions, test assumptions, and refine ideas before anything is built.

This has a very practical impact. Alignment cycles are shorter. Communication with non-technical stakeholders improves. Decisions get made faster and with more confidence.

It’s not just about building more quickly. It’s about thinking more clearly.

6. Organisations are formalising AI capability unevenly

There’s a growing recognition that this is not just about access to tools; it’s about capability. Some organisations are investing in AI literacy across teams. Others are creating internal groups to capture and share what’s working.

But progress is uneven. Much of the momentum still depends on a small number of highly capable individuals, i.e. the people who know how to actually use these tools effectively. Scaling that capability across an organisation remains a challenge. AI advantage, for now, is still uneven and, if we are being honest, fragile.

Taken together, these shifts point to something bigger than individual use cases. AI isn’t just adding efficiency to existing workflows. It’s changing the shape of product development itself. Smaller teams with broader capabilities. Faster cycles with lower risk. Less reliance on documentation, more on working prototypes. Product thinking is happening earlier with more clarity.

We’re still early. The patterns aren’t fully settled, and there’s more to work through, particularly around scaling, governance, and differentiation. But one thing is already clear: AI is no longer a future consideration for product teams. It’s part of the present. And, increasingly, it’s where the real change is happening — not in strategy decks but in the everyday decisions about what gets built and how.

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

Banner photo: Adobe Stock By ipopba.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .