Anyone ever found anything that reads unstructured documents well, to extract structured information?

Been reading a lot about AI - and the generic LLMs like OpenAi’s. But my question is about enterprise or corporate use of “AI”. Specifically about use cases like analyzing vast quantities of unstructured data found in documents, which may sometimes be digitized and sometimes not quite (like scanned copies of hand written documents). They contain varied language but we need to extract the same info that’s mentioned in them in different formatting, and varied language.

This would surely require creating our own LLM? Has this been done successfully anywhere? Would welcome examples of how companies are using this tech. Or any pointers to reading.