A map/reduce workflow for LLMs, with what looks like local caching. To me build systems and data processing pipelines like this one have a big intersection.

A tool from Google that uses LLMs to extract structured data. I suppose it’s quite reliable!