This guide gives you a reusable starting pattern. It is designed to help you see the work more clearly; it is not a substitute for judgment, source checking, or responsibility for the result.
Prepare the inputs before you ask for output.
The model only sees what you give it. Spend a few minutes naming the reader, desired result, and uncertain information. This makes a first draft easier to assess and reduces the need for decorative rewriting later.
Give the task a useful brief.
Turn these rough notes into a decision-ready email. Recipient and role: [who will read it] Context: [why this matters now] Notes: [paste labeled notes] Decision or response needed: [what the recipient may need to decide] Tone: [direct, warm, concise, or other] Use this structure only if the information supports it: context, what changed, options or recommendation, request, and details to confirm. Preserve names, dates, costs, and commitments exactly as supplied. If the notes do not support a recommendation, present the options and write “recommendation not established.” End with one specific response requested from the recipient. List any missing information before the email body.
Replace every bracketed field with your real context. Read the output before reuse.
Four steps that keep the result usable.
- 1
Name the recipient’s decision or response before drafting.
- 2
Separate confirmed facts from suggestions in the notes.
- 3
Ask AI to expose missing details before it writes the email.
- 4
Check every name, date, amount, and commitment against the source notes before sending.
Start with the response you need
Many work emails become long because the writer begins with a full history. Start by naming what the recipient needs to decide, approve, answer, or simply know. That gives the draft a useful boundary and helps the reader scan for the next move.
Keep options distinct from a recommendation
A list of possible paths is not the same as a recommendation. Ask the model to keep those sections separate, especially when the notes contain preferences but no confirmed decision. This prevents a polished email from making an informal suggestion sound approved.
Use a confirmation line as a safety rail
When a detail is missing, a visible confirmation request is more useful than a plausible guess. Keep questions concrete: which date, which owner, which amount, or which option should be recorded? A short verification line protects the accuracy of the entire message.
A worked example: a budget question
Rough notes about a tool upgrade might say “costs are fine, need approval”. The email should ask the specific question — “can you approve the annual license?” — state the amount only if the notes state it, and put the missing renewal date in a “to confirm” line. That turns an ambiguous note into a decision the recipient can answer in one reply.
Trim the history to the decision
Recipients rarely need the full timeline. Keep the context to one or two sentences that change the answer, then move to the request. If the notes contain a long investigation, keep it out of the email and offer the record as a link. A shorter message is more likely to get the decision you need.
Run a human check.
- Can the recipient tell what response is needed within the first few lines?
- Are facts, options, recommendations, and open questions separated?
- Did the draft preserve supplied names, dates, amounts, and commitments exactly?
- Are missing details visible before the email is sent?
- Does the email end with one specific next response?
AI is strongest here when it makes missing information, structure, and options easier to see. The moment an output becomes a claim, commitment, or decision, bring a person back into the loop.