Hacker News Reader

Because of stuff like this I don't get how you can trust it not to hallucinate when it summarizes your inbox.

IME LLMs hallucinate much less when given a text that it can summarize or answer questions about vs. asking it to generate something