"Last year, product teams shipped more than ever. Now, the chaos is catching up.” Malte Scholz, Head of Product and co-founder of airfocus by Lucid, believes that product management is becoming a bottleneck. "And it's not just the product managers, it's the leaders, the product ops… the entire function is slowing product development down.”
Andrew Ng’s analogy is the typewriter. It made it easier to write, but it didn't help us know what to write. The same dynamic is playing out in product teams and AI today, but with higher stakes and faster consequences.
"AI speeds up the engineering, coding, and delivery, but the decision-making just hasn't caught up. That gap very often exposes a weak strategy, and weak product management principles and workflows, in many cases," he says.
The severity varies by organization, but regardless of operation size, it’s time for the product industry to pay attention. "The slow-moving companies don't have this bottleneck problem so much, but they're also moving very slowly. In a medium to fast-moving world, product management is a huge bottleneck, and it's going to get worse.”
AI helps you do, but not decide
To understand how we got here, Malte points to a historical parallel that cuts to the heart of the problem.
"PMs are now doing way more prototyping work, design work, and sometimes they're even shipping small things themselves with new AI tools like Cursor or Claude Code. It's fun, it gives you something in your hands afterwards, but it doesn't make you think about the problem a lot, which is actually the role."
The issue, Malte thinks, isn't about capability but stems from human nature. "People tend to pick work that makes them feel busy and productive, and people tend to avoid the uncomfortable thinking work,” he observes. “It's just built into humans. But the fundamental job hasn't changed: Product managers need to make sure that they think about the problem space and that the individual problem spaces are also linked to the broader strategy." Let’s not forget why we’re here. As Marty Cagan describes it: "Product management is about discovering a product that is valuable, usable, and feasible – and then working with a team to deliver it."
The 30/70 split
While AI has dramatically accelerated development velocity, product management faces a different reality.
This isn't just a temporary mismatch. "AI speeds up the engineering and the coding and the delivery, but the decision-making just hasn't caught up. And that gap very often exposes a weak strategy, and weak product management in many cases."
The contrast with engineering becomes stark when you examine the work itself. "You want to use AI for all the stuff that is easy to solve, and we’re doing that already. Nobody's writing sentences for a release note or a user story. I can tremendously speed up the process in this whole discovery and de-risking process. I can test solution ideas and experiment much faster with these new prototyping tools."
But there's a critical caveat where the job is still to have enough time to think, “about the hard part that the AI cannot do.” This is like this evergreen wisdom that, for whatever reason, product people struggle with, perhaps he suggests, “because it's not human behavior to leave the comfort zone."
When you dig into the mechanics, the difference between accelerating code and accelerating product decisions becomes clear.
"AI isn't really good at helping with getting the relevant context. What's happening in the entire product arc? What are the problems that I hear and want to solve? How do they match the company strategy? All these complicated things that require constant back and forth and meetings, and all of that AI is not ready to help yet," explains Malte.
The compounding misalignment effect
Perhaps the most insidious consequence of increased velocity is what Malte calls the "compounding misalignment effect."
Shipping faster as a result of accelerated build creates its own feedback loop: "When development speeds up, you're shipping more. But that also has a boomerang effect, because there's also more product management work on the launching and the alignment. The communication internal to marketing and sales. There's more work coming back," says Malte.
The current limitations of AI solutions in product management
Given AI's capabilities, why can't it simply solve the strategic alignment problem? "In order to build the perfect product strategy, you need to have the context,” Malte explains. “Very good or perfect insights into product analytics and what the customers really want and need now, and in three years. That's obviously an end state that you will probably never reach."
AI is limited in its usefulness in the creation of strategy. The problem is often about context that exists but isn't documented. "At Lucid, our company objective is to drive enthusiastic user adoption of Lucid's broadly applicable platform. There's so much loaded into that statement, and context needed to understand it, that AI can’t know: the last five years of Lucid, where they're weak, where they're strong, how they got to that statement, what they told us at the offsite meeting in a three-hour presentation that wasn't recorded. It's not systematically written down."
There's also a trust problem: "People don't trust the AI with this kind of stuff. They don't want to put in the effort to collect all of the information to then give it to the AI."
Even when teams do provide context, evaluation becomes difficult: "When you collect all these artifacts, you put it into the AI, and then you let it come up with something. For some reason, it's harder because the output looks so good, so shiny, so well thought through. It's a bit harder to criticize it."
How to keep up in AI acceleration
So what's the way out of this bottleneck for product teams? "The vision and the components are 100% true. You need these things in order to do product strategy with AI. What I'm now saying is that, in reality, this is often breaking and it ends up being more of a manual task, and a very human hand-holding basis."
There's a behavior change required: "If everyone would document, perhaps everyone would be better and faster. But people don't do it so much because there's mistrust, and it would require heavy behavior change."
So what can we do as we wait for the necessary documentation and behavioral shifts to happen?
1. Insight quality is the new moat
"In a world where AI can generate infinite ideas, the only edge left is knowing which signals matter,” advises Malte. “Many companies don't have their feedback under control – they make gut decisions all day. You could get away with that in the past, but not anymore.”
The economics have shifted because building software is increasingly becoming cheaper. In the past, you were winning by building alone. “If everyone can build now, the quality standards go up, and you increase the benchmark by really knowing what problems to solve.
2. Focus on alignment and product operations
Alignment was always important, but it's becoming more important now. "Alignment is a ‘new’ superpower. In a high velocity environment, teams ship faster, dependencies multiply, and shipping cycles compress, which means without a real-time shared source of truth, misalignment essentially compounds," says Malte.
"The conversation has shifted. Product ops isn't justifying its existence anymore. It's becoming the backbone. Marketing and sales have had ops for 10 years. Product is finally catching up."
The threshold is clear: "The moment you have around 10 product managers, and one of them is not a product ops person or does 50% of his time is doing product ops, you're really doing yourself a disservice."
The question facing product leaders isn't whether to adapt, it's whether they'll adapt fast enough. The teams that figure out how to maintain strategic clarity while increasing velocity will be the ones that are going to win.
The rest? They’ll just ship faster into chaos.
Emma-Lily Pendleton
Read also
Experience the new way of doing product management
Experience the new way of doing product management