We want to give every engineer a Claude subscription so we can hit our deadlines

A founder of a growing startup told me this recently. They're under pressure from investors to reach a certain ARR, but the team can't keep up with the feature requests. The fast pace of development has turned the product into a complex system. Jumping on the hyping, the founder's solution is more AI.

I asked them how engineers know what to build. The answer was telling. "I just tell them what to do. Short communication lines, no need for documentation, everything is in my head."

We know by now that AI can help generate more code, more quickly. But just as your engineers need ample context to build the right thing, so do the models. Feature descriptions, current architecture, repository conventions. The more an LLM knows about the product, its codebase and the intent behind a feature, the better its output will be. If your engineers are flying blind, adding AI to the mix just means you're producing the wrong code faster.

LLMs can be a great multiplier. It is on you to make sure they are multiplying your team in the right direction.