Turn [link or file of the article or transcript] into drafts I can edit. Read it in full, then use my past posts at [links to two or three past posts] as a guide to my voice. In one message in ChatGPT, give me: a thread draft of 5 to 8 short posts that follows the piece's main argument, a newsletter blurb of about 80 words with one line on why readers should click, and three separate post ideas, each with a one-line hook. Use only claims the piece actually makes, and flag any point you had to shorten. Don't post, schedule or send anything. If the piece is too thin for a thread, say so and give me the blurb and ideas only.
Paste it into your dot's conversation in ChatGPT. Edit anything in [brackets] first.
How it works
- Share the article or transcript with your dot: a link, a file in connected Google Drive, or an upload in the conversation. Add two or three past posts so it has your voice to match.
- Replace the bracketed parts and paste the prompt. The drafts come back together in one ChatGPT message.
- Edit them, then tell your dot what to change (“shorter hooks”, “less formal”). Your feedback in the same conversation shapes the next batch.
- You publish. Nothing in the prompt lets your dot post or schedule on your behalf.
Working from interviews every week? See content kit from an interview.
Repurpose a long post: FAQ
Which apps do I need to connect?
It works with your files. Connect them before you paste the prompt; your dot only uses apps you've connected and permitted.
Will my dot send or change anything without asking?
It shouldn't if you keep the suggested rule: Keep posting and sending on Ask before taking action. The prompt asks for drafts only, and asking for a draft doesn't give your dot permission to publish it. OpenAI says a dot can still make mistakes, so review its first runs.