Every weekday morning, the Farmington Police Department in Farmington, Connecticut — a town of about 26,000 people outside Hartford — posts an arrest log. It's a PDF. Names, charges, dates, bond amounts. It's public record, free to anyone.
Two publications cover it. Farmington Patch, which is part of a national network that operates in thousands of towns. And The Farmington Mercury, which is one of ours.
They start with the same document. What they do with it is not the same.
Patch publishes the arrests once a week. You get a name, a hometown, the charges, the time of arrest, and a court date. That's it. The article is about 250 words long, and it's surrounded by ads — a banner at the top, pop-up subscription prompts that interrupt you mid-read, a sidebar packed with paid listings, and tracking software that follows you around the internet afterward.
The Farmington Mercury publishes every weekday — usually by the next morning. You get everything Patch gives you, plus the full street address, the specific law that was allegedly broken (with the statute number so you can look it up), the arresting officer's name, the bond amount and whether it was posted, and the police incident number in case you want to file a public records request.
Then we do something Patch doesn't do at all: we actually write about it.
When an address keeps showing up in the log, we tell you. When an officer appears for the fourth time in a week, we note it. When the charges are serious, we explain what the law says and why a high bond amount matters. Every entry links to the one before it, so if you follow along over time, you start to see the patterns — which streets, which hours, which charges keep coming back.
You might be wondering how a small publication can do all of that when a national network with far more resources does less. The answer involves artificial intelligence, and it's worth being honest about it.
A lot of local news right now is being quietly generated by AI. Not assisted by it — generated by it. The same article template applied to thousands of towns, with a human somewhere pressing publish. That's how you get a police log that reads exactly the same in Farmington as it does in Avon, West Hartford, or any other Connecticut town. Nothing about it is specific to your community. It's data processing dressed up as news. The content isn't the product — the coverage is.
We use AI too. We use it to process the same police department PDFs, to speed up research, to handle the mechanical parts of production that would otherwise require a much bigger team. But after the AI does its part, a human editorial process takes over. The context, the pattern recognition, the explanations, the links between entries, the judgment about what matters and what's routine — that's the work that turns data into something worth reading.
The technology is the same. The difference is whether anyone on the other end is paying attention to your town.
We're doing this in Farmington. We're doing it in Charlotte, North Carolina. We'll be doing it in more places. The model is simple: take the same public records everyone else has access to, and do more with them. More data per entry. More context. More frequency. No ads cluttering the page. No software tracking your reading habits.
Local news doesn't have to be thin, late, and buried in ads. It just has been — because for a long time, nobody offered anything better.
We do.
Peter Cellino is the publisher of The Charlotte Mercury and the founder of Mercury Local LLC, which operates hyper-local publications in North Carolina and Connecticut. He writes from Charlotte.