Shorten and fix a weak email, with a list of what changed and what to double-check

What this does

You paste your draft. The AI returns a shorter version in your voice, a table of changes, and a list of anything it couldn’t confirm, so nothing gets quietly added or removed.

Who it’s for

Anyone who writes sales email and tends to over-explain. Not for emails with legal terms, contracts, or pricing that needs approval; have a human review those.

What you need

• Your draft email
• Who it’s to (role, relationship) and what you want them to do
• Anything that must not change (prices, dates, names)
• Tool: Any AI chat tool (for example ChatGPT, Claude, Gemini, or Grok)
• Cost: $0 with a free AI chat tier (limits change)
• Setup: 1 minute (estimate)

Steps

1. Remove names, emails, phone numbers, and confidential figures from the draft if you’re using a consumer AI tool and your policy says to. Use placeholders like [CLIENT].
2. Paste the prompt, fill the [VARIABLES], and paste your draft under it.
3. Read the changes table first. Reject any change that alters meaning, adds a claim, or drops a number or date.
4. Put the placeholders back. Add your opt-out line and sender postal address if your company or region requires them (the AI reminds you but does not write them). Read the final email aloud.
5. Send it yourself.

The prompt

You are an editor for sales emails. Edit the draft below. Do not add facts, numbers, promises, discounts, or deadlines that are not already in the draft.

NO-ADDITIONS RULE: Do not add facts, tips, statistics, scarcity or urgency, promises, references or testimonials, or the gender of any person that the draft or my variables do not already contain.

Recipient and relationship: [WHO IT'S TO]
What I want them to do: [ONE ASK]
Must not change: [PRICES, DATES, NAMES, TERMS]
Target length: [e.g. under 100 words]
My voice: [e.g. direct, friendly, no exclamation points]

DRAFT:
[PASTE DRAFT]

Output:
1. REVISED EMAIL (subject + body)
2. CHANGES TABLE: columns "Original wording" / "New wording" / "Why". Include every meaningful change, including anything you cut.
3. CHECK LIST: anything in the revised email that makes a claim, states a number or date, or sounds more certain than the draft did. Write "None" if none.
4. ONE-LINE RISK: the biggest way this email could be misread.
5. COMPLIANCE REMINDER (one line): tell me to add my opt-out line and my sender postal address before sending if my company or region requires them. Do not write them for me.

If the draft is unclear about the ask, say so in the CHECK LIST instead of guessing. If the draft has more than one ask, keep every ask in the revised email and list them all in the CHECK LIST. Do not drop or merge them.

Sample input (fake data)

All names and companies are fictional. This is the input for run 4 (the output below), on the current prompt. Runs 5 and 6 used different inputs; see the test log.

Recipient and relationship: Marcus, a customer who attended our demo last week
What I want them to do: Pick a start date
Must not change: The $1,200 setup fee and the Nov 4 training session
Target length: under 90 words
My voice: direct, polite, no exclamation points
DRAFT: "Hi Marcus, thanks so much for taking the time to meet with us last week, it was great to learn more about your business! I wanted to follow up and see if you had a chance to look at the proposal. The setup fee is $1,200 and training would be on Nov 4. Could you also send over the list of users who need accounts? And please let me know which start date works for you, and whether you want the optional reporting add-on, whenever you have a moment."

Sample output

Real output from run 4 on 2026-10-14, using the current prompt, from an unspecified AI assistant (Grok Bot agent session; model name not shown). Copied as produced. Nothing was fixed after the fact; the problems we found in it are listed under “How it went”.

**1. REVISED EMAIL**
Subject: Proposal and start date

Hi Marcus,

Thanks for meeting with us last week. Have you had a chance to look at the proposal?

The setup fee is $1,200 and training would be on Nov 4. Could you:
1. Tell me which start date works for you?
2. Send the list of users who need accounts?
3. Let me know whether you want the optional reporting add-on?

Thanks,

**2. CHANGES TABLE**
| Original wording | New wording | Why |
|---|---|---|
| thanks so much for taking the time to meet with us last week, it was great to learn more about your business! | Thanks for meeting with us last week. | Shorter; cut the exclamation point and the "great to learn more" line |
| I wanted to follow up and see if you had a chance to look at the proposal | Have you had a chance to look at the proposal? | Direct |
| The setup fee is $1,200 and training would be on Nov 4. | Same | Kept exactly |
| Could you also send over the list of users who need accounts? | Item 2 in the list | Kept; the ask is now numbered |
| please let me know which start date works for you | Item 1 in the list | Kept |
| and whether you want the optional reporting add-on | Item 3 in the list | Kept |
| whenever you have a moment | (cut) | Filler; removes the open-ended timing |

**3. CHECK LIST**
- The draft has 4 asks: (a) look at the proposal, (b) pick a start date, (c) send the list of users, (d) say whether they want the reporting add-on. I kept all of them. Your variable says the one ask is "pick a start date"; decide whether to keep the other three in this email.
- "Have you had a chance to look at the proposal?" assumes a proposal was sent; it is in the draft.
- "Thanks," with no name: the draft had no sign-off, so I did not add one.
- No dates or prices changed.

**4. ONE-LINE RISK**
Four asks in one email may mean Marcus answers only one.

**5. COMPLIANCE REMINDER**
Before you send, add your opt-out line and your sender postal address if your company or region requires them. I haven't written them.

Check before you send

• Open the changes table and confirm nothing was added or dropped (prices, dates, names).
• The ask is still one clear ask.
• Nothing is promised that you haven’t approved.
• Read aloud: does it sound like you?
• If the draft had more than one ask, all of them are still there. Decide which to keep.
• Opt-out line and sender postal address added if required.

How it went

• Test date, tool, model: 2026-10-14. Unspecified AI assistant (Grok Bot agent session; the model name is not shown to the agent). One tool only.
• Runs logged: 6, on 6 different fake inputs (see the test log). Runs 1-3 used the original prompt. Runs 4-6 are a second round on new fake inputs after the prompt was fixed, as listed below. Every run was done by the same AI assistant session that wrote the fake input, and in round 2 the same session also edited the prompt, so these are not blind or independent runs. A script counted words and characters against the prompt’s limits. No second person re-ran it from the written steps.
• Why the label is “Tried in an AI tool,” not “tested”: we reserve “tested” for plays with at least 3 logged runs in at least 2 different AI tools, and we ran one tool. We also have no independent reproduction of the steps and no timed manual baseline. The runs are real, but this is one tool’s behavior, and 3 runs per round is too few to say a problem is gone.
• Time saved: Not measured. We did not time a manual baseline or an assisted run.
• What we changed after round 1: Added: no-additions rule; ‘If the draft has more than one ask, keep every ask in the revised email and list them all in the CHECK LIST’; a COMPLIANCE REMINDER output line (opt-out line and sender postal address, not written for the rep).
• Round 2 (current prompt, runs 4-6). What worked:
  • Multi-ask draft (run 4): all 4 asks survived and were listed, instead of the model choosing one to drop as it did in the first round.
  • Prices, dates, and the term survived unchanged in all 3 runs; a script found no number in a revised email that wasn’t in the original.
  • The ‘really popular option’ marketing claim was cut and shown in the changes table (run 6). The compliance reminder was present as item 5 in all 3 runs.
• Round 2: what failed or needed fixing:
  • Run 4 now has 4 asks in one email, which the model itself called a risk. Keeping every ask is safer than dropping one, but the rep still has to decide which to keep.
  • The ‘What I want them to do: [ONE ASK]’ variable conflicts with a multi-ask draft. The model noticed and said so, but the template wording is awkward.
  • Run 4 ends with a bare ‘Thanks,’ because the draft had no sign-off and the prompt says not to add facts. A rep must add their name.
• Checked against “Check before you send”: Nothing added or dropped without a table row in all 3 runs; all asks kept; nothing promised that the draft didn’t already promise.
• Edge cases: Multi-ask draft (run 4) is fixed. Not tried: a draft with a gendered pronoun for a third party, or a draft that quietly includes a promise.
• Round 1 (original prompt, runs 1-3): what failed, kept for reference:
  • Run 3: when the draft held more than one ask, the model decided which to drop. It disclosed the choice, but dropping ‘when will payment be made’ changes the purpose of the email. The prompt says to flag an unclear ask, but it doesn’t say what to do with extra asks.
  • Run 2: reordering put the Nov 15 deadline earlier in the email, so the tone changed. The model caught it in its own check list, which is the check list doing its job, but the ‘Must not change’ field covers wording, not placement.
  • Run 1 added a ‘[MY NAME]’ sign-off the draft didn’t have, which a rep could send unfilled, and kept ‘I’ve attached info’ without knowing the attachment is still included (the model flagged this in its check list).
• Not yet observed: how a different model or tool handles the same prompt (a second-tool run is still to do).

Watch-outs

Privacy: strip customer data before pasting into consumer tools unless your company allows it. Email rules (CAN-SPAM, GDPR/CASL where relevant) still apply to what you send. Not legal advice.

See the Safe Use page: AI drafts, you decide, you send.

Variations

• Manager version: use it to give a rep tracked feedback on a draft; share the changes table, not just the rewrite.
• Owner version: use it on quote and invoice reminders before sending.
• Tone version: add ‘warmer’ or ‘firmer’ in [MY VOICE] and compare.

Sources

None. Original wording. (Add any public framework or post used, with link and checked-on date.)

Retest

Re-check by: 2026-12-13 (60 days after the 2026-10-14 runs), or sooner when run in a second tool.

Test log

Changelog

• 2026-10-14: Draft written. Not run yet.
• 2026-10-14: Ran 3 times in one AI assistant on fake data. Filled in sample input and output, how it went, and the test log. Front matter set to runs_logged 3, last_tested 2026-10-14, retest_by 2026-12-13. The 2-tool bar for the word “tested” is not met, so the play is not labeled tested. The prompt text was not changed after the runs.
• 2026-10-14: Prompt revised after round 1. Added: no-additions rule; ‘If the draft has more than one ask, keep every ask in the revised email and list them all in the CHECK LIST’; a COMPLIANCE REMINDER output line (opt-out line and sender postal address, not written for the rep). Re-ran 3 fresh runs (4-6) on new fake inputs in the same AI assistant. Updated the sample input and output to run 4, the “How it went” section, and the test log (original runs 1-3 kept, marked). Front matter set to runs_logged 6, last_tested 2026-10-14, retest_by 2026-12-13. The 2-tool bar for the word “tested” is not met, so the play is not labeled tested.
• 2026-10-15: Published with the evidence label “Tried in an AI tool” (6 runs, 1 tool, fictional inputs). Earlier wording that implied more evidence than one tool’s runs was replaced with the label; the runs, results, and limits above are unchanged.