Can Claude realistically generate tax planning reports (e.g., presenting strategies to a client) without a lot of manual review and correction?

One member said they'd love to use AI as a tax planning report generator — feeding in a set of strategies (e.g., 'here's the seven strategies I want to present to a client') and having it output a clean, client-ready format with numbers filled in, since building these reports manually eats up a lot of time. Another member pushed back, saying they hadn't had good luck with this kind of use case, citing past disagreements with ChatGPT over accuracy on certain topics, and expressed skepticism that the output could be trusted without significant time spent proofing it. The group acknowledged this is a fair concern and noted that there's real skill involved in what you feed the model — the quality of your input strongly affects the quality of the output — but that this isn't intuitive when you first start using the tool. The group agreed to cover, in a later part of the session, specific techniques for improving prompt input to get more reliable, presentation-ready output, but no concrete method was given yet in this segment.