Companies that treat AI as a replacement for product and design thinking are going to produce a lot of fast, mediocre, undifferentiated product.
There’s a narrative taking hold in tech right now that goes something like this: AI is making designers and product managers obsolete. Engineers can vibe-code entire products. Agents can draft specs, generate prototypes, and ship features without the slow, expensive humans who used to gatekeep the process. Design thinking had its moment. Move fast, ship everything, let the data sort it out.
I understand the appeal. And parts of it are true — AI is compressing the time it takes to go from idea to working code. Dramatically. I’ve watched PMs who couldn’t write a line of code six months ago build functional prototypes in an afternoon. That’s real and it’s not going back in the box.
But here’s what I’ve learned leading AI transformation at scale: the companies that treat AI as a replacement for product and design thinking are going to produce a lot of fast, mediocre, undifferentiated product. The companies that treat AI as an amplifier for design thinking are going to build things their competitors can’t catch.
Speed without direction is just faster waste.
When you give a large organization AI tools that make shipping faster, the first instinct is to measure velocity. How many PRs merged? How fast did we go from spec to production? How many experiments launched this quarter?
These are reasonable metrics. They’re also dangerous in isolation.
I’ve seen what happens when you optimize for throughput without maintaining the discipline of asking hard questions upfront: you ship more things, but not better things. You run more experiments, but with weaker hypotheses. You produce more prototypes, but they’re built on assumptions nobody validated. The output volume goes up. The learning velocity stays flat.
This is the velocity trap. You feel faster. Your dashboards confirm it. But your product isn’t compounding — it’s just accumulating.
The thing that surprised me most about working with AI at scale wasn’t what got easier. It was what got harder — or more precisely, what became more important because everything else was accelerating around it.
Choosing which problems to solve. When it costs almost nothing to build a prototype, the bottleneck shifts from “can we build this?” to “should we build this?” The judgment about which customer problems actually matter, which ones represent durable market opportunities, and which ones are just interesting distractions — that judgment becomes the highest-leverage activity in the entire product development lifecycle. That’s product thinking. That’s design research. That’s the stuff some companies are trying to cut.
Knowing when you’re wrong. AI is very good at generating confident-looking output. It will give you a spec that reads well, a prototype that looks polished, and a plan that seems coherent. It’s much worse at telling you the whole thing is built on a flawed assumption. The discipline of testing your thinking against real customer behaviour — going broad before going narrow, validating before scaling — that’s not overhead. It’s the thing that prevents you from scaling a bad idea at unprecedented speed.
Maintaining coherence across a system. When individual contributors can move faster, the risk of fragmentation goes up. Every team ships faster, but nobody’s ensuring the pieces fit together. The customer experience becomes a patchwork of locally optimized but globally incoherent features. Systems thinking — the ability to see how the parts relate to the whole — becomes more valuable, not less.
Asking questions that AI can’t generate. AI is extraordinary at answering questions. It’s not great at asking the right ones. The most valuable thing a designer or PM does isn’t produce artifacts — it’s frame problems. “What if we’re solving the wrong problem entirely?” “What would this look like from the perspective of a user who doesn’t trust us yet?” “What are we optimizing for and is that actually what matters?” These are the questions that change the trajectory of a product, and they come from human judgment, empathy, and experience.
I’ve spent the past year building the infrastructure that lets designers and product managers use AI tools at scale. And the pattern I keep seeing is this: the teams that get the most value from AI are the teams that already had strong design and product thinking discipline.
They use AI to explore more broadly in less time. They generate more concepts, test more variations, and get to validated insights faster — because they know how to frame experiments and interpret results. AI amplifies their methodology rather than replacing it.
The teams that struggle are the ones that skipped the methodology and went straight to output. They’re producing a lot of stuff. They’re not learning much from it.
This maps to something I think about constantly: design thinking isn’t a phase of the process that AI can compress away. It’s the steering system that determines whether acceleration creates value or just creates noise.
When AI can generate prototypes in minutes and draft specs in hours, the quality of the questions you ask and the problems you choose matters more, not less. The thinking is the product. The AI is the accelerant. Get that backwards and you’ll move very fast in the wrong direction.
If you’re a company investing heavily in AI-driven development — and you should be — here’s what I’d argue based on what I’ve seen work:
Invest in AI AND in the design and product thinking that gives it direction. Don’t cut the people who frame problems and validate solutions because you’ve automated some of the people who execute on them. Those are different functions and you need both.
Encode your methodology into the tools, not just the training. If you have a design thinking framework, don’t just teach it in workshops and hope people apply it when they’re moving at 3X speed. Build it into the agent workflows. When a PM starts a new initiative, the tools should prompt for customer evidence, suggest research synthesis, scaffold experiment design. Make the methodology the default path, not an optional detour.
Measure learning velocity, not just output velocity. Track how fast you go from hypothesis to validated insight, not just how fast you go from spec to shipped feature. If your experiment volume goes up but your insight quality stays flat, your AI investment is generating motion, not progress.
Treat AI tooling as a platform, not a rollout. The companies that will sustain their velocity gains are the ones building coherent agent platforms with shared context, governance, and quality standards. The ones that deployed a bunch of point tools and called it transformation will be dealing with fragmentation and technical debt within a year.
Here’s what I’d say to designers and PMs who are worried about being replaced: you’re right to be paying attention, but you’re probably worried about the wrong thing.
The risk isn’t that AI replaces design and product thinking. The risk is that your company believes AI replaces design and product thinking, and makes bad organizational decisions based on that belief. That’s a real risk and it’s happening at some companies right now.
But the opportunity is enormous. If you can operate at the intersection of design methodology, product strategy, and AI capability — if you’re the person who ensures that AI acceleration creates customer value rather than just output volume — you are in the most strategically important seat in the building. There aren’t enough people who can do this well.
The designers who will thrive aren’t the ones who resist AI or the ones who just learn to prompt better. They’re the ones who understand that their real value was never the artifacts they produced — it was the thinking that determined what to build and why. AI makes that thinking more valuable, not less, because the cost of building the wrong thing just dropped to near zero while the cost of building the wrong thing at scale went up dramatically.
The fastest teams I’ve worked with aren’t the ones with the most AI tools. They’re the ones with the clearest thinking, amplified by AI. That’s the model. That’s where this is going.
Speed is cheap now. Direction is what’s scarce.