A software engineering truism that feels weaker now after AI codegen:
Brooks’s Law from The Mythical Man-Month - "Adding manpower to a late software project makes it later"
When a project was running late it would be tempting for the eng manager to just assign a new engineer to the project. Generally adding people to an already late software project would slow it down though. You have to ramp up someone from scratch and there is now a larger coordination tax.
The coordination tax is still there, but with AI and if you have a high agency level engineer the cost of onboarding someone to a project has decreased dramatically.
If you have been developing software in an LLM friendly way (I can touch on how in future posts but think mermaid charts, `.http` files, terraform, project specific skills, etc) an engineer can spend a day or two asking questions against the codebase and very quickly find their footing in a new repo.
So I think no longer is the equation:
- Cut features (generally the right answer)
- Delay release
But now there is a valid third option:
- Throw another engineer at the problem.
If they are skilled and you develop in an AI friendly manner they should slot right in.
That said, generally the right answer is probably still to cut features. haha.