← All posts
Real Talk5 min read

I Asked AI to Help Me Write a Business Email. Here's What Happened.

April 17, 2026

Let me be honest about the expectation going in.

Most people who've seen an AI-written email have a story like this: a colleague forwarded them something that was clearly generated, full of phrases like "I hope this message finds you well" and "Please do not hesitate to reach out." It was technically correct and completely lifeless.

So when someone suggests using AI to help with your business emails, the instinct is: yeah, no thanks. I'd rather just write it myself.

That instinct makes sense. But it's based on a very specific kind of AI output — the kind you get when you don't give it anything to work with.

Here's what I mean.


The first attempt: exactly what you'd expect

Say you need to follow up with a potential client — someone you met at an event two weeks ago, had a good conversation with, and haven't heard from since. You want to check in without being pushy.

You open ChatGPT and type:

"Write a follow-up email to a potential client."

Here's roughly what you get:

Subject: Following Up

Dear [Name],

I hope this email finds you well. I wanted to follow up on our recent conversation and see if you had any questions. Please feel free to reach out at your convenience. I look forward to hearing from you.

Best regards, [Your Name]

Technically an email. Sounds like it was written by a very polite robot in 1997. You'd never send this.

This is where most people stop and conclude: AI can't write emails.


What actually went wrong

The problem isn't that AI can't write. It's that you handed it almost nothing.

"Write a follow-up email to a potential client" gives AI no idea:

  • Who the client is or what you do
  • What you talked about
  • What tone you want
  • What you actually want them to do after reading it

Without that, it fills in the blanks with the safest, most generic possible version. That's what you asked for, technically.

The fix is to give it context. Not a lot — just enough to make it real.


The second attempt: the same task, done differently

Same situation. This time, the prompt:

"Write a short follow-up email to a potential client named David. We met at a local business event two weeks ago and talked about his landscaping company needing help with their social media. I want to check in, remind him of what we discussed, and offer to set up a quick call — but keep it low-pressure. Friendly, not salesy. Under five sentences."

Here's what came back:

Subject: Good to connect the other week

Hi David,

It was great talking with you at the event — I've been thinking about your social media situation and I think there's a pretty straightforward path forward. No pressure at all, but if you'd want to hop on a quick call sometime in the next week or two, I'd love to share a couple ideas. Either way, hope the season's off to a strong start.

— [Your Name]

That's a real email. It sounds like a person. You might change a word or two, but you'd send it.

Same tool. Four minutes apart.


What changed

The second prompt included:

  • A name (David — makes it feel like a real email, not a template)
  • Context (landscaping, social media — gives it something specific to reference)
  • The desired outcome (offer a call, low-pressure)
  • A tone direction (friendly, not salesy)
  • A length constraint (under five sentences)

None of that takes long to think through. You already know all of it — you just have to type it out.


The part skeptics get wrong

The concern is usually: "It won't sound like me."

And that's fair — the first attempt didn't sound like anyone. But the second one? With a little context and a tone direction, it's close. Close enough that most people edit one or two phrases and hit send.

AI doesn't write for you. It writes from you. The more you put in — your voice, your audience, your situation — the more it reflects back something that actually sounds human.

The robotic output isn't a feature of the tool. It's a symptom of a vague prompt.


Try it yourself

Next time you have an email you're putting off, open ChatGPT and include:

  • Who you're writing to (name, role, or relationship)
  • What you talked about or why you're reaching out
  • What you want them to do after reading it
  • The tone you want (friendly, professional, direct, casual)
  • How long it should be

See what you get. If it's not quite right, reply with "make it warmer" or "cut it in half" or "lead with the request, not the background." It'll revise on the spot.

The email you've been putting off might take two minutes once you stop trying to write it from scratch.


If you want to get comfortable doing this across every kind of writing — emails, proposals, social posts, reports — that's exactly what the Clearly, AI course teaches. It's designed for people who aren't technical and don't want to be. Just practical, real-world AI use that saves you time. Plans start at $15/mo — see what's included.

Ready to go further?

The full Clearly, AI course goes deep on everything in this post — with hands-on exercises, real prompts, and new modules launching regularly.

See plans — from $15/mo