This is a pretty good presentation, using Python, of the ability to improve vanilla LLM tools with info relevant to a particular scenario that the LLM might be blind to.
@Geoff first mentioned this to me/us at one of our meetings.
It also highlights some of the open source tooling that’s available.
