We have provided LLMs with reference materials we created ourselves. Additionally we are running validators to check the output.
I agree that the LLM will still not be 100% accurate but it should work pretty well
If you have validators check that the output is near 100% accurate, you could just cut out the AI and just use the validators as tutors.