Hello Reader,

The conversation about AI in product has been a conversation about tools for two years.

Which assistant for which task?

Which model for which workflow?

Does the latest release finally closes the loop?

In our survey of 309 product leaders, the conversation that matters next is somewhere else.

We just published the State of AI in Product 2026 report, co-published with Product Circle. We surveyed Product Leaders from around the world across April and May. The headline is not about adoption. Adoption already happened. Here's what we learned.

The tools changed. The work did not.

87.7% of respondents have adopted AI coding assistants.

85.4% use AI tools for product work.

69.9% have shipped AI-powered features inside their own product.

But, only 36.1% say AI is strengthening their operating model. Almost a third say it is exposing weaknesses that were already there, or making things actively worse.

The tooling layer ran ahead. The operating model never caught up. I am not surprised, and this is what I feared would happen.

The bottleneck is upstream, it always has been

When you ask where AI is actually landing, the pattern is sharp. Around 50% report high impact in engineering and development. 45% in design and prototyping. Discovery, ideation, and documentation sit in the middle. Strategic planning, QA, customer support, and cross-team collaboration sit in the single digits.

So building got faster. Deciding what to build did not.

One survey respondent said it better than I could:

While I don’t believe this bottleneck is necessarily new, I want every product leader to sit with these words. If engineering is now twice as fast and discovery is exactly as slow as it was a year ago, you have not made your product process faster. You have moved the traffic jam upstream, to the part of the work that decides whether any of that speed is pointed at something worth building. The next operating-model investment lives there, not in another tool.

AI is a multiplier, not an equalizer

A lot of people assumed AI would level the playing field. The data says the opposite.

Organizations with mature operating models are about 1.7x more likely to say AI is strengthening them, and 3x less likely to say AI is making things worse. The orgs that were better run before AI are also converting it better now. This is key.

Size makes this even clearer. Smaller organizations, the 1 to 50 group, report AI strengthening their operating model 48.1% of the time. For organizations of 500 or more, that number drops to 19.6%, even though the large orgs run formal AI training at more than twice the rate of the small ones.

Why? Larger organizations usually have bigger budgets. They do more training. Yet, they have less conversion. This is where we see the large orgs are measuring adoption rates of using the AI tools as a measure of success whereas the smaller teams quietly redesign how work moves through the pipeline using the best tools fit for purpose. Whatever your operating model was before AI, AI just gave you more of it. If you were optimizing for the wrong things then, you’re certainly doing more of that now.

The strategy gap nobody is looking at

Here is the finding that I find most interesting and critical in the research, especially because it’s not a new problem. It’s something I’ve been very passionate about. It is the single largest gap in the entire dataset, bigger than any split by tool, industry, geography, or company size. And it runs straight down the middle of the org chart.

61.9% of product managers name “no clear AI strategy” as a top challenge. Only 19.0% of C-level leaders say the same. A 42.9% gap between the people setting direction and the people doing the work.

This is where strategy deployment comes into play. When a gap this big exists, it’s saying that the strategy at the top is not being translated into something the teams can digest and run with.

It also means that people working in the same org are looking at completely different problems, and each are assuming everyone else sees what they see.

I watch this play out every time I walk into a product org mid-transformation. Ask the CPO and ask the PMs the same question about the strategy, and you get two different companies. One is confident the direction is set. The other cannot tell you what it is.

The gap does not mean leadership has no strategy. It means whatever strategy exists at the investment level has never been translated into the operating rules a team needs on a Tuesday to be able to execute. Strategy that lives in a board deck or an executive’s head and never reach a sprint in a way it can be executed on and traced back to outcomes is not yet strategy. It is intention.

Clarity is the accelerator

Here is something many leaders get backwards about governance. They treat it as the thing that slows people down, when a clear mandate is usually what frees them up. Give a team a clear sense of what they are allowed to do, where the limits sit, and who owns the call, and they stop hesitating and start moving. A guardrail is not a cage. It is permission to run inside it.

And when you want autonomous teams, whether that is with humans or machines, this becomes a critical part of keeping everyone aligned. Autonomy is what scales companies. Alignment makes sure it scales in the intended way.

None of this requires a new platform. It requires the harder work of redesigning how decisions, reviews, and ownership move through your organization, now that the tools are already in everyone’s hands.

You can read the full report, with all five findings and the methodology, at productcircle.co/state-of-ai-2026. And if the strategy gap is the one that hits closest to home, our product strategy course at Product Institute is built around exactly that translation problem. We also can help with operating model assessments to show you where your gaps lay.

Which of the five findings lands the hardest in your org right now, and what would change if you fixed the one underneath it?

Until next time, Melissa Perri Founder Product Institute, Board Member, and Teacher

​