AI Skills for Communicators: Write Faster Without Losing Your Voice
AI drafts fast and defaults to generic. For anyone whose job is being heard, that's a real risk. Here's how to use it for speed without your writing converging on everyone else's.
By Patin Team · Examples are illustrative composites
If your job involves being heard rather than just being clear, AI presents a specific hazard. It drafts quickly, and its default register is the statistical middle of everything it has read. That register is competent, balanced, faintly corporate, and completely forgettable — which is fine for a policy summary and fatal for anything meant to land.
The risk isn't that AI writes badly. It's that it writes acceptably, and acceptable is easy to ship.
The convergence problem
Everyone using the same tools with the same light-touch prompting produces work that sounds increasingly alike. If your differentiator is a distinctive voice, that's your differentiator eroding — slowly, through a series of individually reasonable decisions to accept a good-enough draft.
The counter isn't avoiding AI. It's changing what you use it for.
Use it for the parts that aren't voice
There are stages of writing where voice doesn't live, and those are the ones worth delegating:
Structure. What are the three things this needs to cover and in what order? Genuinely useful, and the answer is rarely voice-dependent.
Compression. Take 600 words to 200. Mechanical, tedious, and AI is good at it.
Variants. The same message for three audiences at three lengths. High-volume, low-creativity.
Interrogation. What's the weakest claim here? What would a sceptical reader push back on? This is the highest-value use and the most under-used — you're asking for critique, not prose, so nothing of yours gets overwritten.
The stuck part. Not to write it, but to give you five bad options so you can see what you don't want. Reacting is easier than generating.
Notice what's missing: the first draft of anything where the phrasing is the work.
Show it your voice rather than describing it
"Write in a friendly, professional tone" means nothing — every organisation on earth believes that describes them. Tone instructions are almost useless in the abstract and very effective as examples.
Paste in three things you've written that sounded right. Ask what's consistent about them. You'll usually get a more precise description of your own voice than you could have written yourself — sentence length, whether you use questions, how you open, whether you hedge. That description is reusable, because it's specific enough to check against.
Edit against your own standard, not the model's
The last pass has to be yours, and it's not a proofread. It's a check against a standard you hold: does this sound like a person, does it say the difficult thing, would I have written this sentence.
The practical test: read it aloud. Generic writing survives silent reading and falls apart when spoken.
Priya — the newsletter that stopped working
Priya writes a customer newsletter for a design software company. Open rates had been strong for two years. Over six months they drifted down, with no obvious cause — same cadence, same topics, same list.
She read a year of back issues in order. The older ones had jokes that didn't quite land, opinions she'd have softened later, sentences that ran on. The recent ones were tighter, cleaner, and read like every other product newsletter in the category.
She'd started drafting with AI and editing lightly. Every individual edit had improved the sentence. Collectively they'd sanded off everything that made it hers.
Her change: AI now does the structure and the product details. The opening and closing paragraphs — the parts people actually remember — she writes cold, before opening any tool.
Sam — the sentence that made it real
Sam is a communications manager at a hospital trust. He was drafting a message to staff about a rota change that was going to be unpopular, and the AI draft was flawless: acknowledged the difficulty, explained the reasoning, thanked everyone for their patience.
It was also completely hollow, in the specific way that institutional writing is hollow when it's trying not to say something. Anyone who'd been on the receiving end of that kind of message before would recognise the shape immediately.
What was missing was one sentence naming what was actually hard: that the change meant more weekend cover for the same pay, and that leadership knew it. He wrote it himself and put it in the second paragraph.
The response was still negative. It was negative about the rota rather than about being managed, which was the outcome he wanted.
The one thing
Use AI for the scaffolding, not the voice. Structure, compression, variants, and critique are all safe to hand over. The opening line, the difficult sentence, and the ending are the parts people remember, and they're the parts worth writing cold.
If you only keep one habit: read the final version aloud. Generic survives the eye and dies in the mouth.
Put this into practice
Reading is a start — but skill comes from doing. Try these drills now.
Reading about it only gets you so far
Patin turns this into five-minute drills that score what you write and tell you why. It's in closed beta — join the waitlist and we'll email you when your cohort opens.
Just want the writing? .
Keep reading on this
AI Skills for Writers: Draft, Edit, and Publish with AI
The useful question isn't whether writers should use AI. It's which stage of writing to hand over — because handing over the wrong one costs you the thing people are paying for.
5 min readSubstack Now Lets Any Reader Score Your Writing for AI. That Changes How You Use It.
Substack launched AI-detection scanning on July 21, powered by Pangram. Any reader can now check any post over 100 words for human vs AI content — one click, no subscription required. Content authenticity is now a visible, reader-facing metric. Here's what to change before you publish again.
5 min readYour AI First Draft Is Pulling You Toward Generic. Here's How to Avoid It.
Figma's CEO Dylan Field told Stratechery that AI output 'draws from the middle of the distribution' — and that teams become viscerally attached to their first concept. The iterate-and-refine skill is what separates competent AI output from distinctive work.
4 min read