What was the result?
self-reportedBrian Davis, founder of One Stop CPA, used the tax research app BlueJ plus ChatGPT on a client's merger and acquisition question tied to a move to a new state. He developed an answer within two hours, "much less time than it would have taken previously," according to the Journal of Accountancy.
One Stop CPA is a fully digital firm of four accountants and three non-accountants, serving doctors, real estate investors and entrepreneurs. Brian Davis puts the client's scenario into BlueJ for cited, source-backed tax research, then reviews it and applies his own judgment: "I make the ultimate decisions." He then uses a library of saved ChatGPT prompts and Google Docs templates to turn the analysis into client deliverables like road map summaries, recap documents and one-page PDFs. His trick: after getting an output just right, he asks ChatGPT to turn it into a reusable prompt so next time it takes one step.
Who did this?
How do I try it?
- Write the client's problem and facts in plain words.
- Run the research in a tool that cites its sources, and read the sources it cites.
- Review the answer yourself and decide; the AI is a starting point.
- Use ChatGPT to turn your conclusions into a client-friendly summary, road map or one-page PDF.
- When an output comes out right, ask ChatGPT to turn it into a reusable prompt and save it in your template library.
Brian Davis's own reuse trick, from the article: "Can you turn this into a reusable prompt so next time I can get a similar output in one step?" Editorial adaptation for the client summary: Here is my analysis of [client situation]: [paste]. Turn it into a one-page summary for the client in plain English with: the question, what we found, the options, our recommendation and next steps. Do not add any facts or numbers that are not in my analysis.
Make it run every week
A play works once. Turn it into a routine in three small steps:
- Run it once and check the result yourself.
- Give your AI a schedule. Tell it when to run this play (every week, or every month) and to message you with what it found before you act.
- Keep a log. One line per run: what was checked, anything flagged, what you did.
What should I watch out for?
Self-reported. The two hours is for one client question, and the earlier time isn't given. He uses enterprise versions so client data isn't used for training; don't paste client details into consumer AI tools. AI research must be checked by a professional before advice goes to a client.
Did it work for you?
PageMotor keeps this site free.