What I Learned Using AI to Reply to Every Work Email for Two Weeks

Monday morning, 47 unread emails, and I'm staring at my inbox thinking about how much of my life I've spent typing "Thanks for the update, I'll circle back on this." So I did something stupid: I committed to letting ChatGPT draft every single work email response for the next two weeks. Every. Single. One.

I thought I'd either become incredibly efficient or get fired. Neither happened. But something weirder did.

The Setup Wasn't As Simple As I Expected

First problem: you can't just paste an email into ChatGPT and say "reply to this." Well, you can, but you'll get something that sounds like a customer service bot wrote it. I spent the first day refining my prompt until I had something that actually worked.

Here's what I landed on: I'd paste the original email, add two sentences about my actual position on the topic, and tell it my relationship to the sender (boss, client, annoying vendor, etc.). That context made all the difference. Without it, every response came out at the same weirdly formal level.

The bigger issue was time. Copy email, switch to ChatGPT, paste, add context, generate, read it, fix the parts that didn't work, copy back. For a simple "sounds good, let's do Thursday" type email, this took longer than just typing it myself. I almost quit on day two.

But I kept going because the longer, more complicated emails — the ones where you have to be diplomatic, or explain something technical, or gently say no — those were different. Those actually saved time. About 40 minutes total on day three, when I had to write a detailed pushback on a project timeline without making my client feel attacked.

The Kick: AI Made Me Realize How Badly I Communicate

Here's what you won't read in any productivity article about this. After about four days of watching ChatGPT rewrite my thoughts, I noticed a pattern. My instinct is to over-explain. I'll write three paragraphs when one sentence works better.

ChatGPT kept cutting things. And almost every time, the shorter version was clearer. I'd feed it my rambling context about why I couldn't take on a project, and it would spit back two sentences that said exactly the same thing without the defensive tone I didn't realize I was using.

The AI wasn't just writing for me. It was showing me what I actually meant to say, stripped of all the anxious padding I add to avoid conflict. That was uncomfortable to see. I genuinely thought I was a concise communicator. I was wrong.

By week two, I started noticing my instincts changing. Before I'd even paste something into ChatGPT, I'd think: what's the actual point here? Can I say this in one sentence? Do I need that apology at the beginning? The tool was training me to edit myself before I even wrote anything.

What Went Wrong (And It Went Wrong a Lot)

I should be clear: this experiment nearly backfired multiple times.

Day five, I sent an AI-drafted response to a coworker who'd emailed about a frustrating situation with another team. The draft was perfectly professional. Too professional. It came across cold. She followed up asking if she'd done something to upset me. I had to explain I was experimenting with AI drafts, which was an awkward conversation I don't recommend.

The AI has no idea when someone needs warmth instead of solutions. It defaults to helpful and efficient when sometimes you need to just say "ugh, that sucks, I'm sorry you're dealing with that." I missed those cues more than once because the AI-generated version looked correct on paper.

Also: humor. ChatGPT is terrible at casual workplace humor. It either ignores your jokes entirely or responds with something that reads like a bot trying to be relatable. I had to manually add every "lol" and every self-deprecating aside. Which, fine, but those are often the parts of emails that actually build relationships.

And one genuinely scary moment: I almost sent a reply to the wrong thread because I was moving so fast between tabs. The AI had drafted a perfectly reasonable response to an email about budget concerns, but I nearly pasted it into a thread about someone's medical leave. That would have been bad. The speed of the process made me sloppy in ways that felt dangerous.

What Actually Stuck After the Experiment

I stopped using AI for every email immediately after the two weeks ended. The overhead isn't worth it for quick replies. But I still use it for three specific types:

• Anything where I need to say no without burning a bridge
• Technical explanations for non-technical people
• Responses where my first instinct is defensive or annoyed

That last one is the biggest win. When I'm irritated, I write irritated emails. Now I dump my frustrated thoughts into ChatGPT and let it translate them into something professional. It's like having a filter between my emotions and my sent folder.

The strangest part is what happened to my writing outside of email. I'm editing this blog post differently than I would have three weeks ago. Tighter. Less hedging. I keep hearing that AI voice asking "what do you actually mean here?" and cutting the rest.

I went into this thinking I'd learn about AI tools. Instead I learned that I've been hiding behind words for years, and a robot called me out on it. Still not sure how I feel about that.

Heads up: Some links in this post may be affiliate links. I only recommend tools I've personally tested. Opinions are entirely my own.