Retrieval-augmented generation
Looking up relevant documents and putting them in the prompt, so the model reads facts rather than recalling them.
Also written: RAG, retrieval augmented.
Rather than hoping the model memorised your documentation during training, you search your own documents for the relevant passages and paste them into the prompt. The model then answers from material in front of it.
This is the single most effective remedy for invented answers, and it has the useful property that you can show the reader which passages the answer came from. Its weaknesses are in the search step: if retrieval surfaces the wrong passage, the model will answer confidently from the wrong passage.