Reports and summaries are where AI saves the most time at work — and where vague prompts waste the most. A summary that misses the decision, or a status report pitched at the wrong reader, has to be redone by hand.
These templates are built for the documents most teams produce every week. Replace the parts in brackets and paste your material at the end.
Meeting notes from a transcript
Template: "Turn the transcript below into meeting notes for [people who missed the meeting]. Structure: 1) decisions made, 2) action items as a table with owner and due date, 3) open questions. Leave out small talk and repeated points. If an owner or date was not stated, write 'not agreed' instead of guessing. [transcript]"
The last instruction matters: without it, models tend to invent plausible owners and deadlines.
Weekly status report
Template: "Write a weekly status update for [project name] for [the steering committee]. Use these headings: Progress this week, Risks and issues, Next week, Decisions needed. Keep each section to three bullets or fewer. Flag anything behind schedule clearly at the top. Use only the information in my notes below. [notes]"
Executive summary of a long document
Template: "Write a one-page executive summary of the document below for [a director who will decide whether to fund it]. Start with the recommendation in one sentence, then the three strongest reasons, then the main risk and how it is managed. Plain language, no jargon, under 300 words. [document]"
Leading with the recommendation is what makes an executive summary useful. Asking for it explicitly changes the whole structure of the answer.
Reviewing a draft before you send it
Template: "Review the draft below as if you were [a sceptical finance manager]. List the five weakest points — unclear claims, missing evidence, inconsistent numbers — with a suggested fix for each. Do not rewrite the whole document. [draft]"
Giving the model a specific reviewer persona produces sharper, more useful criticism than "check this for me".
Turning notes into a first draft
Template: "Turn my rough notes below into a first draft of [a two-page project update] for [the regional manager]. Keep my facts and figures exactly as written, don't add new ones, and mark anything that needs confirming with [CHECK]. Tone: clear and factual. [notes]"
Comparing two documents
Template: "Compare version A and version B below. List every substantive change — added, removed or reworded commitments, dates and figures — in a table with columns: section, version A, version B, why it matters. Ignore formatting and typo fixes. [A] [B]"
Rules that keep AI reports accurate
- Restrict the sources. "Use only the information below" is the single best defence against made-up facts.
- Ask it to flag gaps. Tell the model to mark missing information instead of filling it in.
- Name the reader. A summary for the board and one for the field team should look very different.
- Check every number. Treat figures in AI drafts as unverified until you've compared them with the source.
Skip the template-hunting
Templates help, but they're one more thing to keep track of. With IacuWise you can type the short version — "summarize this report for the board" — and it expands it into a complete prompt with the reader, structure, limits and source rules already in place. It works with ChatGPT, Claude, Gemini and other models, and the Chrome extension puts it right inside the chat window.