Why do AI-written newsletter drafts need so much rewriting, and how do you fix it?

A writer on my team was getting AI drafts for each section and rewriting almost every single one, nearly 100% of the text. That defeats the whole purpose.

Why were the drafts so bad?

We had the AI investigate its own failures against her rewrites. The split was about even:

• Half were facts. The writer AI only saw a headline and a short summary, so it wrote thin, wrong or outdated details. One draft said 197 hotel rooms. The real number was 297.
• Half were voice. The section prompt told it to write like a premium food writer for affluent readers. Our readers wanted a neighbor.

The same thing shows up in business profiles. One AI draft I reviewed was not horrendous, just awkward. It added stuff I never gave it, like what was on the menu. My rules after that one:

• If there aren't enough facts, cut the element. Drop the data card instead of padding it.
• Never invent anything. It made up a card saying the business wasn't claimed. Gone.
• Use every real asset you were given. I had supplied at least five photos and it used one.
• Pull real material. I have it pull up to 10 real reviews with Apify and add them to the page.

Does the AI model matter?

More than people think. On one of our newsletter tools, rewrites kept coming out in a bad voice. The reason: the system was still on a three-year-old model we had picked only to save money at the start.

Picking the cheapest option early didn't mean staying on it forever. We moved news, restaurants and businesses to Claude Sonnet. Events stayed on automatic rewriting, because events are always the same style and there's nothing to edit.

How do you make AI sound like a real local writer?

I had the AI reverse-engineer better prompts from two issues she liked. The voice target, in my words:

"Personal voice, sounds like an actual neighbor, not AI. Honest but positive."

1. Give it the full article, not a headline and summary.
2. Show it two or three real examples per section of writing you'd actually publish.
3. Add a notes box where the human types quick details AI can't fake, like what it was like to visit the restaurant.
4. Ban the tells. No AI language, no emojis, no em dash in every line.

Testing the new prompts cost about 39 cents.

What can a local add that AI never will?

The resident's take. My advice to other operators: let a helper or AI do the first draft, then edit it to be more fun and more relatable, in a Morning Brew or Hustle style.

Here's my favorite example. In my area they announced a new Chick-fil-A. Everybody knows Chick-fil-A drive-thrus back up. This one was going next to a McDonald's at an intersection that's already weird and already jammed. I could comment on that combination. AI will never know that intersection is a mess.

That kind of insight is what helps people who actually live there win when there are tons of AI newsletters nobody believes are real.

What did that look like in my own newsletter?

In early 2025 our weekly process was simple. A VA found all the events. My wife and I found the news on the local news channels and ran it through ChatGPT to get the basics. Then she went through and personalized it. New restaurant? "Oh yeah, we took our family there last week." Add that spice.

What if the owner is too busy to add local flavor?

Then someone else who lives there has to. My partner newsletters get great feedback, and I think they're quite good. But the owners are busy business people, so they're missing a little of that true local flavor.

My answer for the partner model: we provide the base layer, and somebody living in that town customizes it. It doesn't have to be the owner. An assistant is fine. It just has to be a local.

What does a good section prompt look like?

Here's the brief I gave for restaurant and business blurbs:

Headline: 4-8 words. Do NOT use the business name.
Body: two short paragraphs in the voice of a friend recommending it.
Call out one or two interesting, very direct specifics.
No AI language. No emojis.

Should each kind of story get its own prompt?

Yes. City council meetings have a certain style. School board has a certain style. Permits have a certain style. Each type gets its own plan for how to write the article, and everything should stay easy to edit.

Don't cap it too low either. From a city council meeting transcript I'd want something like 5 to 10 articles, not 1 to 3, with the publisher setting the budget.

Is it OK for AI to write the story if a person checks it?

I think it's close to the ideal version. A journalist gathers the information, AI drafts the story, then the journalist goes back through it: make sure it's correct, make sure it sounds right, then publish. Some people push back on that. The content is still true, custom and unique.

How long until the drafts are good?

Plan on rounds. On one of my agents, feedback took it from about 70% good to 90%. The next target is 95 to 98%. You won't get there on the first prompt.

This is the work lightbreak does: section prompts, real sources and an editor in the loop. For the wider workflow, see whether AI can run a local newsletter.