Everyone's talking about AI like it's either going to save marketing or kill it. The reality? It's just a tool. And like any tool, it's only useful if you know when to pick it up and when to put it down.
I've been testing AI in my email marketing workflows for the past year. Some experiments worked brilliantly. Others produced content so generic I'd have been embarrassed to send it. Here's what I've learned about using AI to work faster without turning your emails into soulless robots.
Where AI Actually Helps in Email Marketing
Let's start with what AI does well. It's excellent at the boring stuff that eats your time but doesn't require your unique perspective.
Subject line variations. I'll write one subject line that captures what I want to say, then ask AI to give me ten variations. Most will be rubbish, but two or three will spark ideas I wouldn't have had. It's brainstorming without booking a meeting.
First draft structures. When I'm staring at a blank page for a product launch email, AI can give me a skeleton to react against. I rarely keep more than 20% of what it produces, but that's not the point. It gets me moving.
Repurposing content. Turning a blog post into an email sequence, or breaking down a long newsletter into shorter segments. This is grunt work that AI handles well because the thinking has already been done.
AI should handle the tasks you'd delegate to an intern, not the ones that require your judgment.
Data analysis. Spotting patterns in open rates, identifying best send times, segmenting lists based on behaviour. AI processes this faster than any human could, and it doesn't get distracted halfway through a spreadsheet.
Where AI Falls Flat
Here's where people get into trouble. They hand over the parts that actually need a human brain.
Voice and personality. AI can mimic tone if you give it examples, but it can't replicate the weird specific way you see the world. That slightly odd analogy you'd use. The reference only your audience would get. The opinion that might ruffle feathers.
Strategy decisions. AI can tell you your Tuesday emails perform better than Friday ones. It can't tell you whether that matters for your specific business goals, or whether you should ignore the data because you're building a different kind of relationship with your list.
Common mistake
Using AI to write entire emails from scratch, then sending them without heavy editing. Your subscribers signed up for you, not a language model. They can tell the difference, even if they can't articulate why something feels off.
Emotional nuance. When you're writing about something sensitive, or responding to a crisis, or sharing genuinely vulnerable content, AI just can't get there. It'll produce something that looks right but feels hollow.
My Actual Workflow
Here's how I balance AI productivity with keeping emails human. This isn't theory - it's what I do every week.
Planning phase: I decide what I want to say and why. This is pure human thinking. No AI involved. What does my audience need right now? What's the one thing I want them to take away?
Drafting phase: I'll often talk through my ideas out loud, then use AI to help structure those rambling thoughts into a coherent outline. But the ideas themselves come from me.
Writing phase: I write the first draft myself. Yes, it takes longer than asking AI to do it. But this is where my voice comes through. The draft is usually messy. That's fine.
Editing phase: Here's where AI earns its keep. I'll ask it to check for clarity, suggest tighter phrasing, spot repetition I've missed. It's like having a copyeditor on call at 11pm.
Try this
After writing your email, paste it into ChatGPT and ask: "What's unclear here? Where do I repeat myself? Which sentences could be tighter?" Take the useful suggestions, ignore the rest. You're the editor, not the AI.
Subject lines: I write three options myself, then ask AI for ten more. I usually end up with a hybrid - my core idea, refined by a variation AI suggested.
The Productivity Gains Are Real
I'm not going to pretend AI hasn't made me faster. It has. What used to take me two hours now takes about 75 minutes. That's significant when you're producing multiple emails per week.
But the speed comes from eliminating friction, not from outsourcing thinking. AI handles the mechanical parts so I can focus energy on the parts that matter.
The emails that perform best are still the ones where I had something genuine to say and said it in my own voice. AI helped me say it more clearly and get it out the door faster. That's the sweet spot.
The goal isn't to produce more emails. It's to produce better emails with less wasted effort.
What I'd Tell Someone Just Starting
Start small. Pick one part of your email workflow that feels tedious and experiment with AI there. Maybe it's writing alt text for images. Maybe it's creating preview text. Maybe it's summarising your email for a mobile-first version.
Don't start by asking AI to write your welcome sequence or your big product launch emails. Start with low-stakes tasks where you can learn what works without risking your relationship with your list.
And read everything out loud before you send it. If it doesn't sound like something you'd actually say to a real person, rewrite it. That's still your job. Probably always will be.
The takeaway
Use AI for speed and polish, but keep your hands on strategy and voice. The emails that convert are the ones that feel like they came from a human who gives a damn. No AI can fake that.