Implementing LLM in Operations

I am a technical (web/mobile) and operational founder exploring implementing AI in specific work areas and tasks in a in home health agency.

I am dipping my toes into the ecosystem and would love to hear the experience of other people or teams that have implemented something in testing or production.

We are looking for two primary use cases: 1. An internal knowledge assistant for clinical and operational teams (not doctors or diagnostic in nature, mostly our own policies and internal data) where accurate and trustworthy info is necessary, 2. A customer facing chat bot capable of taking pressure off our existing team (burnout and churn of that team is a common occurrence).

I am looking for info on best practices, possible foot guns, maybe even resources that would be essential.