AI Gave Us the Speed. Now What Do We Do With the Thinking?

For years, the product development life cycle had one predictable friction point: engineering.

Design would define the experience. Product would prioritize it. Engineering would struggle to ship it. The bottleneck was downstream, in code, in build cycles, in sprint capacity. Everyone upstream had time to think. Time to research. Time to run the process.

That constraint is dissolving.

AI-assisted development is collapsing engineering timelines in ways we are only beginning to reckon with. Code that once took weeks is being scaffolded in hours. Prototypes that required full sprints are being shipped before the product brief is finalized.

And I’ll say it plainly: I think this is extraordinary. The ability to close the gap between thinking and building, to prototype at the speed of a conversation, to ship with a velocity that would have seemed unreasonable two years ago, is one of the most exciting shifts I’ve seen in my career. This isn’t a case against speed. Speed is finally here. The question is what we do with it.

The Bottleneck Didn’t Disappear. It Moved.

When AI removes friction from engineering, it doesn’t remove friction from product and design, it exposes it.

Product and design are now the bottleneck. Not because the people doing that work are slow, but because the system was never designed for them to be fast.

The seams that used to hide in development backlogs now live inside product and design processes. And unlike engineering bottlenecks, which show up in velocity metrics and sprint reviews, these bottlenecks are invisible until the experience fails.

What Gets Lost When We Skip the Thinking

The product and design process was never just methodology. It was a structural argument: slowing down to understand the problem saves time downstream. Empathy before ideation. Discovery before definition. Research before resolution.

AI is compressing those phases. Generative tools skip the divergent phase. Pattern-matching accelerates decisions that once required deliberation. And the invisible cost is the reasoning that never happened, the insights never surfaced, the edge cases never considered.

This is what Un-Design calls Decision Debt. And it’s no longer just the weight of accumulated organizational choices. It’s now also the weight of skipped thinking. Of people whose needs get optimized into a pattern that was never actually built with them in mind.

Speed without thinking isn’t efficiency. It’s liability at scale.

A Confession From the Middle of This

I lead both product and design delivery. And I’ve spent years building something specific: a way of working where understanding people comes first. Where journey maps, research, and documentation aren’t bureaucracy, they’re the argument. The evidence that earns the right to build something. The foundation that makes product decisions defensible and design decisions meaningful.

Standing inside that practice right now, watching AI compress the very phases I’ve spent a career defending, is one of the strangest professional experiences I’ve had. I’ll be honest, part of me is terrified. Not of the technology, but of what gets quietly lost if we move through this transition without intention. The insights that don’t get surfaced. The edge cases that get assumed away. The product decisions that get made at speed before the question has been properly asked.

And at the same time, I am genuinely thrilled.

Because I’ve also seen what happens when good thinking has no velocity behind it. When the right answer arrives too late to matter. When the research is thorough, the strategy is sound, and the moment has passed. AI doesn’t have to be the end of rigorous process. It can be the thing that finally makes rigorous process fast enough to be useful.

I don’t have a clean resolution to that tension. I’m living inside it. And I think leaders who pretend they’ve figured out this moment are deceiving the people who depend on them.

This is what transformation actually feels like from the inside. Exciting. Unsettling. Both at once.

The Real Question

If product and design teams are now expected to both think and ship, to run discovery and direct AI-generated development, then something else has to give. The cognitive load doesn’t disappear. It doubles. And organizations that treat this as a headcount optimization opportunity will find they’ve traded one bottleneck for a different kind of collapse.

The question isn’t “how do we slow AI down?” It’s “what structures do we need to build so the thinking survives the speed?”

That means:

• Processes that protect reasoning, not just output

• Accountability for system-level outcomes, not just delivery velocity

• Incentives that don’t let speed become a proxy for quality

• Governance that can honestly distinguish what AI can skip from what it cannot

This is Un-Design applied to an AI-native world: not automating what we already do, but asking whether what we already do was ever designed to produce the outcomes we’re claiming to pursue.

The Closing Truth

AI gave us the speed. That’s real, and it’s remarkable.

Now the question is whether we build the structures that make that speed mean something, or whether we use it to produce broken experiences faster than we ever thought possible.

The organizations that will win aren’t the ones that just move faster. They’re the ones that use this moment as a forcing function to finally build the delivery architecture they should have built years ago, one where thinking is protected, coherence is accountable, and speed doesn’t come at the cost of the reasoning that makes experience matter.

We finally have the velocity. The work now is making sure we’ve earned it.