Graham McNicoll

A couple of years ago, the first AI tool I used was Copilot. We saw 20 to 30 percent improvement in engineering productivity. Real, but incremental. Now with Claude Code and similar tools, we are managing multiple agents simultaneously. Teams are seeing 5x to 10x productivity gains. That is wild, and it changes the question. Your team's productivity is going exponentially higher. Software development is no longer the hard part of the process. It has become the easy part. Which means the bottleneck has moved. When building is no longer the constraint, the other steps get squeezed. What should you be building? What should it look like? How does it integrate with the other parts of your product? What effect did that product change have on your metrics? Is this actually worth shipping? The new hard problem is not "how fast can we write code." It is "how do I build the right product for my users" when I can ship ten times as much. When you can generate product MVPs in hours instead of weeks, you can get real signal on what should be built and what should not, at a pace that was impossible two years ago. The teams that figure this out early will pull away from the teams who just ship more. If your team is shipping faster than ever and the roadmap is still a guess, send me a DM. Happy to talk through what the new bottleneck looks like in practice.