The Problem Nobody Was Talking About
Every newsletter creator I've ever spoken to has the same confession. They have opinions. They have stories. They have things worth saying. What they don't have is three hours on a Tuesday afternoon to turn a half-formed idea into something their subscribers will actually read.
That's the gap. Not inspiration. Not ideas. Time and execution.
This is why I built Aldus. Not because the market needed another email tool with a drag-and-drop editor. Not because AI writing was having a moment and I wanted in. Because I kept watching smart, knowledgeable people abandon newsletters they'd spent months building — not from lack of effort, but from the sheer grind of production.
The writing was killing them.
What I Kept Seeing in 2026
Spend any time in newsletter creator communities right now and you'll notice a pattern. Someone launches with energy. They post about their first issue, their second, maybe their tenth. Then silence. When you ask them what happened, it's never "I ran out of ideas." It's "I couldn't keep up."
That phrase deserves some scrutiny. Keep up with what, exactly? Their own schedule? Their competitors? The pace they'd set for themselves back when producing one issue felt manageable?
All of it, usually.
The average solo newsletter creator is writing somewhere between 600 and 1,500 words per issue. For a weekly send, that's the equivalent of a short article every seven days, on top of whatever their actual job is. For most people, that's not sustainable past month three. Research from the creator economy in 2026 consistently shows churn rates among newsletter writers outpacing churn among their subscribers. Creators quit before readers do. That's a problem nobody talks about loudly enough.
Why I Built Aldus (And What I Got Wrong First)
My first attempt at solving this was, frankly, embarrassing. I built a prompt library. A collection of templates people could copy into ChatGPT to generate newsletter content faster. It worked, technically. People used it. But the output felt like it was written by someone who'd never met the creator. Generic. Flat. Safe in all the wrong ways.
Subscribers noticed. Open rates on AI-assisted issues tended to drop after a few sends, because readers are smarter than we give them credit for. They can feel when the voice is off. They don't know why, necessarily, but they disengage.
So I started over with a different question. Instead of "how do I make writing faster?", I asked "what does good newsletter writing actually require?" The answer, after a lot of conversations with creators, was uncomfortable. Good newsletter writing requires consistent voice, genuine point of view, relevant context, and some kind of stake in the topic. An AI that doesn't know any of those things about a creator is going to produce content that sounds like everyone else.
That's the insight that eventually became Aldus. A platform built around learning how a specific creator thinks and sounds, not around generating generic content at speed. The speed is a side effect. The fidelity to voice is the actual product.
The Newsletters That Made Me Rethink Everything
There were three newsletters I studied obsessively while building this.
One was a financial commentary newsletter run by a former fund manager. Dense, opinionated, occasionally wrong in ways the writer would openly revisit the following week. It had 4,200 subscribers and an open rate around 51%. Not viral. Not famous. But incredibly sticky. Readers stayed because they trusted the perspective, even when they disagreed with it.
Another was a local food newsletter in a mid-sized UK city. No paid tier. No growth hacks. Just one person writing about restaurants, food culture, and the occasional rant about what was happening to the high street. Twelve hundred subscribers who forwarded almost every issue.
The third was a B2B SaaS newsletter with 40,000 subscribers and a 19% open rate. Polished, consistent, well-funded. Completely forgettable. It read like it had been written by a committee, because it had been.
The pattern was clear. Voice and specificity beat scale and polish every time. The small newsletters with a strong point of view were outperforming the big ones with production budgets. The question was whether AI could be trained to protect that voice rather than dilute it.
What "Doing the Writing" Actually Means
I want to be precise about this, because the phrase "AI that does the writing" can mean a lot of different things, most of them bad.
It doesn't mean generating content that sounds plausible but says nothing. It doesn't mean filling in a template with your name and topic. And it definitely doesn't mean replacing the creator's thinking with a machine's approximation of it.
What it actually means, at least the version worth building, is handling the production work so the creator can focus on the thinking. The first draft that gets you past the blank page. The structural edit that catches when a section isn't pulling its weight. The subject line variants tested against what your specific audience has responded to before. The consistency check that flags when your tone has drifted from what your readers signed up for.
That's the version I was trying to build. Not a ghostwriter. More like a very capable editor who's also good at typing.
The distinction matters because the failure mode of most AI writing tools is that they optimise for output volume rather than output quality. They can produce ten newsletter drafts in the time it used to take to write one. But if nine of them are wrong for your voice and one is mostly right, you've still spent significant time editing. The productivity gain disappears in revision.
Getting this right required building in ways to capture and preserve what makes a creator's writing theirs. The words they reach for. The references they make. The rhythm of their sentences. The topics they care about enough to have an actual opinion on. When a tool knows those things, it can do the writing in a way that actually sounds like you. When it doesn't, you get content that technically covers the topic and makes no one feel anything.
What I'd Tell Anyone Thinking About This
If you're a newsletter creator reading this and wondering whether AI-assisted writing is worth trying, the honest answer is: it depends on what you're trying.
If you're looking for a shortcut that removes you from the writing process entirely, the results will be obvious to your readers and they'll leave. Voice is the only moat most newsletter creators have. Anything that erodes it is a bad trade, whatever time it saves.
But if you're looking for something that handles the mechanics so you can do more of the thinking, that's a genuinely different proposition. Your job becomes having the idea, knowing what you think about it, and reviewing whether the output captures that accurately. That's still creative work. It's just less exhausting creative work.
The newsletters that will win in 2026 and beyond aren't the ones with the biggest AI budgets or the most automated pipelines. They're the ones with the sharpest points of view and the most consistent relationship with their readers. AI can help you maintain that consistency. It can't manufacture it from nothing.
That's the bet I made when I started building. I still think it's the right one.