The Cost of Iteration Has Collapsed

Everyone seems to be asking whether AI can write better PRDs, user stories, or prototypes.

After a year of building with AI, I think we're asking the wrong question.

The biggest change isn't that AI creates artifacts faster.

It's that AI has dramatically reduced the cost of iteration.

Product Management Was Never About Documentation

Looking back, I don't think our biggest challenge was writing requirements.

It was getting everyone to understand the same thing.

Before AI, a typical workflow looked something like this. A stakeholder would describe a problem. We would discuss it. I would take notes, then later translate the discussion from memory into user stories, PRDs, PowerPoint mockups, or screenshots edited in Microsoft Paint. Developers and QA would review those artifacts, ask questions, and surface gaps. I would clarify the intent, revise the documents, and the cycle would continue.

Those artifacts were useful because they gave the team something concrete to react to. But they were always an approximation of the conversation and the thinking behind it. Static text and images left room for different interpretations, especially once the work moved from one person to another.

Documentation was never the goal. Shared understanding was.

The challenge wasn't that the documentation was poor. The challenge was that documentation is a poor substitute for shared understanding.

Reading about a product is fundamentally different from experiencing one.

Cheap Iteration Changes Shared Understanding

Today, the conversation is still the most important part of the process. The difference is what can happen immediately afterward.

After a stakeholder conversation, I can turn what I heard into an interactive prototype with AI assistance. Within 20-30 minutes, we can review an interactive version of the idea together. Before AI, reaching that point could take days.

The stakeholder can click through the experience while the conversation is still fresh. Developers and QA can see the behaviour, challenge assumptions, and expose edge cases. The prototype becomes a shared artifact for everyone involved.

The prototype becomes the fastest path to shared understanding.

That does not make documentation irrelevant. It changes its role. Documentation can record the decisions, constraints, and intent that emerge after the team has explored the problem together.

Documentation increasingly becomes the output of shared understanding rather than the mechanism for creating it.

When Iteration Becomes Cheap

Cheap iteration changes the conversation.

Stakeholders become more engaged because they are reacting to something they can experience, not trying to imagine a feature from a document. Bad ideas can be discarded earlier, before the organization has invested heavily in them. People become more creative because exploring an alternative no longer feels like a major commitment.

It changes planning as well. Instead of trying to perfect an idea before anyone can see it, teams can explore several directions and learn what each one reveals. Meetings move from "Did we capture the requirement correctly?" to "Is this actually the right experience?"

It also gives product managers a better way to negotiate trade-offs. We can show the ideal solution, the feasible version, and the path between them. The long-term direction stays visible even when delivery has to happen in stages.

Most importantly, cheap iteration makes learning dramatically less expensive.

Not failing. Learning.

Sometimes we discover an idea won't work. Sometimes we validate an assumption. Sometimes we uncover a better solution than the one we started with.

Every iteration produces something valuable: better understanding.

The Bottleneck Has Moved

Ironically, AI hasn't eliminated the hardest part of product management. It's simply moved it.

Creating artifacts has become dramatically easier. Understanding hasn't.

AI can generate possibilities. Product managers still determine which possibilities deserve to become products.

The bottleneck today isn't writing user stories. It's asking better questions. Understanding context. Making trade-offs. Facilitating alignment. Exercising judgment.

Those have become more valuable precisely because artifact creation has become cheaper.

The Real Challenge Isn't AI

One thing surprised me over the past year.

The hardest part wasn't learning the technology. It was helping experienced professionals become comfortable working with an iterative system.

Many people expect AI to produce the correct answer on the first attempt. When it doesn't, they conclude the tool isn't useful.

That hasn't been my experience.

The value comes from the conversation.

Prompt. Review. Challenge. Refine. Repeat.

Each round improves more than a single artifact. It improves the team's shared understanding of the problem.

That's a fundamentally different way of working.

You're no longer crafting documents one at a time. You're shaping an evolving understanding of the product.

What This Means for Product Managers

If I were hiring product managers for an AI-native organization, I wouldn't start by asking which AI tools they know.

Tools change.

I'd want to understand how they think.

• Can they maintain context across a complex problem?
• Can they facilitate productive conversations?
• Can they use prototypes to accelerate learning?
• Can they distinguish between generating artifacts and generating understanding?
• Most importantly, are they willing to experiment?

The product managers who thrive won't be the ones who produce documents the fastest.

They'll be the ones who can turn ambiguity into shared understanding, guide teams through rapid learning cycles, and apply judgment when the number of possible answers explodes.

Their value will come from helping organizations learn faster without losing sight of the problem they are trying to solve.

A Bigger Shift Than Faster Documentation

Every generation of software has reduced the cost of building.

AI may be the first technology that dramatically reduces the cost of learning what should be built.

That's a much bigger shift than writing user stories faster.

The story isn't faster documentation. It's that the cost of iteration — and therefore the cost of shared understanding — has collapsed.

Documentation helped us explain ideas.

Cheap iteration helps us discover them.

And when organizations can learn faster together, they stand a much better chance of building the right products.