Build a strong corpus of job writing examples

It would be super helpful to have a large (well, large by our standards) corpus of good job examples.

These should be individually vetted and approved to ensure that they meet modern practice.

The examples don't necessarily need to be part of apollo - they could live in the docsite, for example we already have lots of job examples there. The problem is that the average quality is dangerously low. The API docs also have examples, which are probably better quality on average, but have very little composition.

For our most popular adaptors, we need at least one example of every function, but we also need examples of composition.

Later, we may want to embed this corpus into the vector DB to assist RAG. Or use it to train our own model

An obvious concern here is that the quality of the examples will degrade over time as our standards and APIs inevitably change

Start by inventorying the existing docsite and API documentation examples, then compare coverage across the most popular adaptors. Define how examples will be vetted, including function coverage and composition, and record how future API or standards changes will be handled. Done means an approved corpus with the required examples and a maintenance approach.