You know the email I'm talking about. "I hope this message finds you well." Three paragraphs of nothing. A closing line about "circling back" that makes you want to throw your phone across the room. You've gotten a hundred of them, and the second you see one, you stop reading.
Here's the problem: if you're using AI to write your emails and proposals without thinking about it, you're probably sending that exact email right now. And your customers can smell it just as fast as you can.
That doesn't mean AI is bad at this. It means most people use it wrong. They type "write me a follow-up email" and hit send on whatever comes back. That's like asking someone who's never met your customer to write a letter on your behalf and signing your name to it without reading it first. Of course it sounds off. It doesn't know you.
The fix isn't to avoid AI for this stuff. It's to use it the way you'd use a sharp employee who's fast at drafting but doesn't know your voice yet — you give them more to work with, and you edit before it goes out. Let me show you exactly how.
Why AI writing sounds like AI in the first place
AI models are trained to be safe, polite, and complete. That's great for a legal disclaimer. It's terrible for a text-message-era world where people read fast and expect you to talk like a human. The tells are always the same: overly formal openers, unnecessary padding sentences, everything wrapped up in a neat little bow at the end, and a tone that's professional in the way a hostage note is professional — technically fine, but nobody actually talks like that.
The other issue is that AI defaults to generic because it doesn't know your specifics. It doesn't know that you and this client already talked on the phone last Tuesday. It doesn't know you're the guy who always throws in a joke about the weather before getting to business. It doesn't know your actual writing rhythm — short sentences, or long ones, whether you use exclamation points or never touch them.
Once you understand that the flatness comes from a lack of context and a lack of editing, the fix is obvious. Give it more context. Edit what it gives you. Every time.
Start with what you'd actually say, not what sounds "professional"
The single biggest shift that changes everything: stop asking AI to write a professional email. Ask it to write the email the way you'd say it if you were standing in front of the person.
If you'd normally say "Hey Mike, following up on that quote — you still thinking it over, or did you go a different direction?" then that's the tone you want back. Not "I am writing to follow up regarding the estimate provided on [date]." Nobody talks like that, and your customers know it.
Try this prompt: Write a follow-up email to a customer who hasn't responded to a quote I sent 5 days ago. Keep it short, casual, and direct — like I'm texting a friend, not writing a formal letter. No "I hope this finds you well." No corporate language. End with a real question, not "please let me know if you have any questions."
Notice what that prompt does. It doesn't just ask for an email — it tells the AI what to avoid, what tone to hit, and what a good ending looks like. The more specific you are about what you don't want, the less generic the output becomes.
Feed it your actual voice, not a blank slate
Here's a trick almost nobody uses: paste in a few examples of emails you've actually written before you ask for a new one. It doesn't have to be anything fancy — just two or three real messages you sent to past clients, even messy ones with typos. That gives the AI something to match instead of guessing.
Try this prompt: Here are three emails I've written to clients in the past. [paste them] Study how I write — my sentence length, my tone, whether I use humor, how formal I am. Now write a new email to a client following up on a proposal I sent last week, using that same voice. Keep it under 100 words.
This works because AI is genuinely good at pattern-matching once you give it a pattern to match. Without examples, it defaults to its training data, which sounds like every other AI email on the planet. With examples, it starts sounding like you on a good day.
Proposals: cut the fluff, keep the structure
Proposals are where people either lose the job before it starts or seal it. A proposal stuffed with AI-generated filler — "we are committed to excellence," "our team brings unparalleled expertise" — reads like it was written for nobody in particular, which means it wasn't written for the person reading it.
The move here is to use AI for structure and speed, but load it with specifics about this exact job, this exact customer, and this exact problem. Generic proposals get generic results. Specific proposals get called back.
Instead of asking for "a professional proposal for a kitchen remodel," tell it what you actually know: the customer mentioned they're tired of their cabinets falling apart, they've got two kids and want durability over looks, they mentioned a budget range, and they want it done before a family event in a few months. Feed all of that in and ask AI to write a proposal that speaks directly to those concerns, not a template that could apply to anyone's kitchen.
Try this prompt: Write a proposal for a kitchen remodel job. The customer's main concerns are durability (they have two young kids), staying under $18,000, and finishing before a family reunion in 10 weeks. Address those three things directly in the opening paragraph before getting into scope of work. Keep the tone confident but not salesy — like I'm explaining the plan to them in person, not pitching them.
That's the difference between a proposal that gets skimmed and deleted and one that makes the customer feel like you actually listened to them. AI can write it fast, but only if you're the one supplying what matters.
The edit pass is not optional
Here's the part everyone skips because it takes an extra two minutes: read it out loud before you send it. If you stumble over a sentence, or it doesn't sound like something you'd actually say, change it. This single habit will do more for your emails than any prompt trick.
AI has a habit of adding phrases nobody uses in real conversation — "I trust this finds you well," "please do not hesitate to reach out," "moving forward." Cut them. Say what you mean instead: "Let me know if anything's unclear," or "Text me if it's easier." Simple, direct, human.
Also watch for length. AI tends to over-explain because it's trained to be thorough. Most emails don't need three paragraphs of context. If you can say it in four sentences, say it in four sentences. Busy people — which is everyone you're emailing — will thank you for it, even if they never say so.
Follow-ups: the place most people give up too early
Follow-up emails are where AI genuinely earns its keep, because most business owners hate writing them and just... don't. You quote a job, you hear nothing back, and instead of following up, you move on and lose work you would've won with one more email. That's not a writing problem — it's a consistency problem, and AI solves consistency problems really well.
Set yourself a rule: every proposal gets a follow-up at day 3, day 7, and day 14 if you haven't heard back. Instead of dreading writing three separate emails, ask AI to draft all three at once, each with a slightly different angle so you're not just saying "just checking in" three times in a row, which is the fastest way to sound like you're begging.
Try this prompt: Write three short follow-up emails for a proposal I sent that hasn't gotten a response. Email 1 goes out 3 days later — casual check-in. Email 2 goes out at 7 days — offer to answer questions or adjust the quote. Email 3 goes out at 14 days — direct, low-pressure, basically asking if they've gone another direction so I can close the loop. Keep all three under 60 words and make sure none of them sound desperate.
That prompt alone will save you the mental energy of writing follow-ups from scratch every single time, which is usually the real reason people stop following up at all.
What to do today
Pick one email you've been putting off — a follow-up you've been avoiding, a proposal you haven't finished, whatever's sitting in your drafts folder making you feel guilty. Paste in two or three old emails you've actually sent, use one of the prompts above, and write the new one using your own voice as the model. Read it out loud once before you send it. That's it. Do that one time today, and you'll never go back to asking AI for "a professional email" again.