Internal tools also help get around the "9 babies can't make an engineer in a month" (or whatever it is) problem. Netflix (where I work, listed in the article) is maybe the best example. There's a limit to how many engineers you can have working on Netflix's 1 flagship service (it's a big limit). You can increase that limit by having other engineers come in and build tools for the whole company. I'm also surprised what percentage of engineering teams are directly "developer experience" at big orgs. It's probably < 1% but I think it should be more like 5% (probably an obscenely wrong opinion).