I saw this shift before I could fully name it.
Moving into a fractional CPO role, I set out to build something that scaled me: a system that could give me back time. On a Saturday morning I sat down and tried — not with a developer, but me and Claude Code. I understood the problem and what I wanted. By lunchtime I had a working voice interface on my phone that let me speak an update and have it automatically log to the database tracking my business development pipeline. No app. No engineering team. A front-end interface, a database, and a voice prompt, built in a morning by someone who had never written a line of code.
That was the moment. Not because of what I'd built — it was basic. But because I no longer needed a developer. That was empowering.
Andrew Ng named it directly: "The invention of agentic coding assistants has led to a new builder's block, where the holdup is deciding what to build." His teams are now proposing a ratio of one PM to 0.5 engineers.
Mike Krieger, co-founder of Instagram and CPO at Anthropic, said it plainly: "The models today are good at adding features. They're not necessarily good about figuring out what to cut out of the product."
The obvious response — and why it doesn't hold
If building is cheap, mistakes are cheap too. Fail fast, learn, move on.
That works at the feature level. It doesn't work at strategy. When building is expensive, cost is the forcing function that makes you say no. That constraint creates discipline — and counterintuitively, it creates innovation. Constraints force the question that matters most: if we can only do one thing, what's worth building?
When cost disappears, every request looks reasonable. Sales asks for a custom integration. You build it. A client wants a slightly different version of the core feature. You build it. Someone at the last board meeting had an idea. You build it. The product doesn't collapse — it sprawls. Three years later you can't explain it in one sentence, Sales doesn't know how to sell it, and the original value proposition is buried under a decade of "yes, we can build that."
That cost doesn't show up on a burndown chart. It shows up when you try to raise a Series B and can't tell the story. Or when a PE firm does diligence and finds a different product at every client site.
What this means for founders and CEOs right now
If you're scaling a product business in 2026, the question has changed. It's not "do we have the engineering capacity to build this?" It's "do we have the product clarity to know what's worth building?"
Most companies don't have good infrastructure for that second question. They have backlogs. They have roadmaps built in quarterly planning cycles. They have PMs who are excellent at managing delivery and less practised at interrogating whether what they're delivering will actually sell, move revenue, or hold up in a board meeting.
The pattern that shows up repeatedly
In scaling B2B businesses, the technology is strong. The product is hard to sell. Not because the engineering is wrong — because the decisions upstream of the engineering were never made rigorously enough.
Companies that treat AI as a feature release will move faster in the wrong direction. The ones that get this right will build less and compound more.
