On the other hand, working at a large company I saw a many-line SQL query that had been run monthly for a long while, looked at and edited by quite a few people, which down in the middle of it was comparing (for equality) a city id and a probability.
There is absolutely room for improved tooling. I have no idea whether Prisma is it.
Well better tools wouldn't have helped that, since most of the them just expose the underlying low-level DB types. Until databases support a proper type-system that doesn't just describe memory layout that kind of bug will always be an issue.
There is absolutely room for improved tooling. I have no idea whether Prisma is it.