Using ChatGPT to write emails you actually want to send

The first dozen times someone uses ChatGPT to write an email, the result has the same problem. It sounds like an email a marketing intern would send. Slightly over-friendly. Slightly over-long. Slightly off in a way you cannot quite name but can definitely smell.

That is fixable. The trick is not finding a better prompt — it is changing how you use the model. Drafting with AI is a workflow, not a one-shot request. Here is the workflow that works.

Step 1: Skip the polite "hello, can you help me write..."

When you sit down to draft an email with ChatGPT, do not start by asking it to "write an email." Start by pasting in your situation.

Compare these two openings:

Write me an email to my client about the deadline.

versus

I am the project manager. My client is a friendly but detail-oriented marketing director at a Tallinn agency. We have been working together for six months on a website rebuild. We were due to deliver the homepage today but we just discovered a major issue with the analytics integration that pushes delivery by three days. The client is going on holiday in two days. I want to tell her now, sound calm and in control, not over-apologize, and propose a concrete plan.

The second prompt has not asked for an email yet. It has just laid out the situation. That alone produces a far better first draft when you do ask for one.

So the first habit: describe the situation before asking for the output.

Step 2: Ask for variants, not "an email"

The most underrated word in AI prompting is "three." Three drafts, three tones, three lengths. Comparing options is more useful than picking the first thing that lands.

A reliable continuation of the prompt above:

Now draft three versions:

>

1. Warm but professional, around 100 words. 2. Direct and short, under 70 words. 3. More detailed, with the plan as a short bulleted list, around 150 words.

>

End each with a clear next step. Do not start any of them with "I hope this finds you well."

A few things to notice. The length cap is doing real work — without it you will get something twice as long as you want. The "no I hope this finds you well" is doing real work too — every AI starts with that sentence by default. The "bulleted list" constraint shapes the structure without you having to write it out.

After reading the three drafts, you will almost always have a clear preference, and you will almost always know exactly which two sentences to edit.

Step 3: Critique what came back

Now reply to the model. Don't write a fresh prompt — reply in the same conversation with what is wrong.

Version 2 is closest. Tighten the opening sentence, drop the "we appreciate your patience" line, and end with one sentence instead of two. Keep the rest as-is.

The model edits the draft you both have in front of you. This is faster and produces better results than re-prompting from scratch, because the model already has all the context.

You can be blunter than you would be with a colleague. The model does not have feelings; you are not insulting it. "That sounds robotic. Make it sound like a normal person." "Cut the apology — I did not do anything wrong." "Less formal. We have been on first-name terms for six months."

A useful question to ask the model after a critique: "what tone would the average reader hear this in?" The reply will tell you whether the draft is going to land the way you want it to.

Step 4: Strip the AI smell

There are five tells that betray AI-written email. If your draft has any of these, replace them:

• "I hope this email finds you well." Nobody talks like this. Open with the actual point.
• "I wanted to reach out regarding…" Cut. Just say what you wanted to say.
• "Thank you for your patience and understanding." Hollow. Replace with something specific or remove.
• "Please let me know if you have any questions." Always true, never useful. End with a concrete next step instead.
• Em-dashes everywhere, perfectly balanced sentences, suspiciously even paragraph lengths. AI-written text is often too tidy. Vary the rhythm.

A useful prompt to strip these in one pass:

Edit this so it sounds like a normal person wrote it. Remove anything that sounds like the opening of an AI email. Vary the sentence lengths. Keep the meaning and structure.

After this pass, the email almost always passes.

Step 5: Save the patterns you use repeatedly

Most professional email falls into a small number of recurring patterns:

• Declining politely
• Chasing politely (first, second, third time)
• Apologizing for a missed deadline
• Confirming arrangements
• Pushing back on a request without offending
• Asking a hard question without making it feel like an interrogation

After a week or two of drafting with AI, you will have a few favourite framings for each. Save them — as a Custom GPT, a Claude Project, a note in your password manager, a Notion page, anywhere. The next time the same pattern comes up, paste the saved framing in and start there. Each time you do this, the loop gets shorter.

A small example. Once you have written a few "polite chase" emails you like, save the model's instructions:

Whenever I ask for a "polite chase" email, follow these rules:

>

- One paragraph, three to four sentences. - Do not apologize for following up. - Acknowledge the previous request was sent on a specific date. - End with a clear soft deadline ("Would Friday work?"). - Tone: warm, not formal. Match the language and register of the original thread.

Now any time you say "polite chase, here is the thread," the model knows what you mean and produces something usable on the first try.

A few situations where AI helps most

Some types of email AI saves you the most time on:

Hard conversations. Layoff messages, ending a contract, asking for a raise, complaining without burning a bridge, declining a former friend's request. The hard part of these is not the words — it is overcoming the inertia. Once you have three drafts, you can edit and send in five minutes.

Languages you write in less often. If you live in Estonia and need to write a customer reply in fluent Estonian, German, or English depending on the recipient, AI is faster than you and usually correct. Always read carefully — language errors happen — but the time saved on draft one is enormous.

Threads you keep restarting. Anything you have written four times and keep deleting. Paste in what you have so far and ask "what's stopping this from landing — what would a more confident version of me write here?" Sometimes the unblock is from the rephrasing alone.

Long technical explanations to non-technical people. Drafting the version that is honest and accessible takes time. AI is good at compressing for an audience when you tell it who the audience is.

What not to use AI for

Genuinely warm, personal notes to people you know well. The hand-written feel matters more than the polish. AI-drafted condolence notes and birthday messages always sound a beat off, no matter how good the prompt. Write them yourself.

Anything that requires you to look someone in the eye later and have meant it. AI is a tool for writing faster, not for deciding what to mean.

The whole workflow, in one line

Describe the situation, ask for three variants, critique in-conversation, strip the AI smell, save the pattern. Five steps, takes ninety seconds the second time. Once it becomes a habit, the email backlog stops being a backlog.