Is it OK to let AI be your copy editor?

Gabriel Abreu @gabe_onchain
2026, September

I am an AI engineer, and I use AI for many things, such as a personal assistant, a virtual friend, a coder, and all the workflows you can imagine.

FYI, I have a personal assistant agent that even calls me and reminds me to take my meds.

But there's one thing that I barely use it for. Writing.

AI isn't good at writing just yet

In my opinion, AI is not there yet. And no matter what skills you give it, the output is no more than decent, but nowhere close to excellence.

However, as a native PT/BR speaker, ofc there's 'Portuguese grammar' hidden in my English drafts.

Sometimes the idea comes first in PT/BR, and only then does the English part develop, somehow more slowly.

That's why I use the famous AI app Grammarly to correct my grammar, and nothing more.

And that's fine, I guess. Right? Haven't authors always had a copy editor working on their behalf?

Bullish, btw

And folks, I'm not against AI at all - on the contrary. In the past year, I've been studying Applied AI Engineering at UniPDS, with over 480 hours of classes with BIG names like Erick Wendel and Hugo Santos, just to name a few of the legends teaching behind this MBA +++ it's certified by MEC (Brazil's Ministry of Education), and I'll soon graduate there.

A few more reasons why I'm bullish on AI:

• LinkSquares' new Workflow Builder Agent lets teams build contract workflows with AI, but nothing goes live until a human signs off.
• Lightcast data shows FDE job posts up 1,000%+ this year, vs 13% for tech jobs overall.
• Perplexity says hundreds of agents ran for two months on an internal database that could save them up to $100M a year.

These are just a few recent examples, but there're many more reasons like them I could list here.

What if we just flip the chair

I still wouldn't recommend using it for writing.

But what if we use it to interview us, extract our own ideas, tone & stories, and help us improve our craft? Something like this:

Interview me about [topic]. Ask one question at a time. Push me for a real story, a number, or a moment I remember. Don't write the piece. When we're done, give me back my own answers, in my words, grouped by idea.

Then you write the draft from your own answers. Imho, that will produce WAYYY better outputs for us.

Our readers can tell

Sorry, not sorry, but your readers can tell & I'm with them to criticize it because I know how annoying it is to watch a video or read a piece and start wondering, "Whose thinking am I actually encountering?"

When you're here reading my posts, you're not looking for the same type of answer you'd get from your bots or agents.

You're looking for a different point of view, whether you agree with it or not.

And imho, writing is an expression of my thoughts.

So, no, I refuse to delegate this to AI.

If you wanna use it anyway, here's a gift

Sure, you can disagree, and you can use it. There's no problem with that.

Btw, here's a quick gift for you that I hope will be valuable and make your time reading this piece worth it.

Add this prompt to your AI's custom instructions:

Write like a human. Every sentence should earn its place. If a sentence doesn't add new information or move the piece forward, cut it. No antithesis. No corrective negation. No paragraph pinning. No parataxis. No summary beats. No rhetorical crutches. No negative parallelisms. No negative anaphoras. No contrasting pairs. No rule of three. No em dashes. No throat-clearing openers. No landing sentences. No setup/payoff constructions. No parallel sentence structures within a paragraph. Vary sentence length unpredictably. No stacked noun phrases. No filler intensifiers (genuinely, really, truly, actually). No corporate-register verbs (leverage, underscore, reflect). No nominalization. No hedging qualifiers. Write for the spoken voice. No performed enthusiasm.

It will make the output slightly cleaner.

It still won't give away to your readers your own 'sauce', and why they should be following/reading you, though.

That part is on you, and I can PROMISE you'll always get mine here by following me on X at @gabe_onchain or reading my blog posts right here.