How to Write Emails That Get Replies, with a Little Help from AI

Most emails don't get ignored because they're badly written. They get ignored because the reader can't tell, in the first three seconds, what you want and how much work you're asking of them. AI is genuinely good at fixing that, but only if you use it as a drafting partner instead of a magic reply button. Here's the method I teach clients who send twenty emails a day and hear back from six of them.

Why emails go unanswered

Before you open ChatGPT, it helps to know what you're fixing. In my experience coaching people through their inboxes, unanswered emails almost always have one of four problems:

• No clear ask. The reader finishes reading and isn't sure whether a reply is even needed.
• Too many asks. You buried three questions in a paragraph, so the reader postpones until they can "deal with all of it," which means never.
• Too long. A wall of text on a phone screen gets a mental "later."
• No deadline or next step. Nothing tells the reader when or how to respond, so it drifts.

AI can help with all four, but you have to feed it the raw material. It can't guess what you actually need.

The core technique: brain-dump, then compress

The single most useful thing you can do is stop trying to write a polished email in your head. Instead, dump everything messy into the AI and let it do the compression. Open ChatGPT, Claude, or Gemini and type something like this:

"Turn this into a short, friendly email. I want one clear ask and a deadline. Keep it under 90 words. Here's the situation: [paste your messy notes]."

For the messy notes, just talk. For example: "I emailed Priya two weeks ago about the venue contract, never heard back, event is June 12, I need her to either sign the attached PDF or tell me what's holding it up, I don't want to sound annoyed but I am a little."

You'll get back something tight and neutral. The word limit matters more than people expect. Without it, these tools pad. With "under 90 words," you get an email someone can read on a phone at a bus stop and answer in one line.

Always specify the one thing you want

The instruction that changes everything is "one clear ask." If you have three questions, tell the AI to make them a numbered list, not a paragraph. People answer numbered lists because they can see the shape of the reply. Try:

"Rewrite this so my three questions are a short numbered list. Add one line at the top saying what I need and by when."

The result reads like: "Could you help me with three quick things before Thursday?" followed by 1, 2, 3. That format roughly doubles reply rates in my clients' experience, because it converts a vague request into a checklist the reader can knock out.

Subject lines: let AI write five, you pick one

The subject line decides whether your email gets opened today or next Tuesday. Don't agonize over it. Ask:

"Give me 5 subject lines for this email. Make them specific, not vague. No clickbait. One should mention the deadline."

You want subjects like "Venue contract: need your sign-off by Fri" rather than "Quick question" or "Following up." Specific subjects that name the topic and the action outperform clever ones. Pick the clearest of the five and move on.

Match the tone to the relationship

The default AI email tone is a little corporate and eager. Fine for a stranger, wrong for a colleague you've known for years. Tell it explicitly who you're writing to:

• "Warm but brief, we're friendly coworkers."
• "Polite and a bit formal, I've never met this person."
• "Direct and no fluff, my boss hates long emails."

You can also give it a sample. Paste an email you've sent before and say "match this voice." This is the fastest way to stop AI emails from sounding like AI emails. If you use a tool that already knows your style, like Gemini inside Gmail or Copilot inside Outlook, it will lean on your past messages automatically, but you can still nudge it with a one-line tone instruction.

The follow-up that isn't annoying

Chasing a non-reply is where people freeze. They either sound passive-aggressive or apologize so much the message disappears. Give the AI the real emotional brief:

"Write a 40-word follow-up. It's been a week with no reply. Friendly, no guilt-tripping, give them an easy out so it's not awkward."

"An easy out" is the key phrase. It produces lines like "Totally understand if this slipped, or if it's not on your plate anymore, just let me know either way." That single sentence gets replies because it lowers the cost of answering.

A quick before-and-after

Here's a real pattern. The original draft someone showed me:

"Hi, I hope you're doing well and had a great weekend! I wanted to reach out because I've been thinking about the report we discussed and I have a few thoughts and also some questions about the timeline and whether we're still aligned on the deliverables and I know you're busy but..."

Fed into Claude with "under 80 words, one ask, friendly, numbered questions," it became:

"Hi Sam, quick check-in on the report. Two things before I keep going: 1) Are we still targeting the 15th? 2) Do you want the customer data section in or out? A one-line answer to each is all I need. Thanks!"

Same content. One is a chore to read, the other gets answered before lunch.

Where AI won't help, and can hurt

A few honest limits. AI doesn't know your relationships, so it will sometimes be too warm with someone you're annoyed at or too formal with a friend. Always read the draft as if you were receiving it. It also invents specifics if you're vague, so never let it fill in dates, names, or numbers you didn't give it. Check those every time. And for anything emotionally loaded or high-stakes, treat the AI draft as a first pass, not a final one. The words are yours the moment you hit send, and the reader can tell when a message was outsourced without a second thought.

A repeatable routine

Once this clicks, your email process becomes: dump the situation, ask for under 90 words with one clear ask and a deadline, request five subject lines, set the tone for the specific reader, then read it aloud once and send. The whole thing takes ninety seconds and produces an email that respects the reader's time. That respect, more than any clever phrasing, is what gets you the reply.