I want to create a learning assistant agent to help me learn technical concepts faster. To that end I have a bunch of resources I believe would be helpful but I do not know if fine-tuning on the resources would be better or creating a vectordb would result in a better agent.
What would be arguments for and against the two approaches?