Publisher. Platform builder. Convinced since approximately 2001 that the media industry has been solving the wrong problem.
I founded The Charlotte Mercury, a digital-first local news publication covering Charlotte and the surrounding region, and Mercury Local — the AI-native publishing platform that powers it. The publications are the mission. The platform is the proof.
Where It Started
The year was 2001. Napster was everywhere — in the news, in the courts, and in every dorm room with a broadband connection. Most people filed it under music thing. In Robert Langran's Mass Media Law seminar at Villanova University, we were required to think harder than that.
The conclusion, once you stripped away the legal theater, was that charging for content was a losing proposition. Not a pricing error. A structural failure. The moment information could be copied and distributed for free, the economics of selling it changed permanently. The record labels were fighting the last war. The news industry would eventually fight the same one.
That seminar lodged itself somewhere I couldn't shake.
After Villanova, I went to New York Law School — which sharpened the analytical toolkit considerably: how platforms work, where liability concentrates, how information economies are built and broken. That foundation turned out to be precisely the right preparation for what came next, even if what came next wasn't yet visible.
What I'm Building
Mercury Local is a local news platform built from scratch — Next.js, Supabase, Vercel — with AI integrated into the editorial workflow at every level. Real beats. Real reporters. Real fact-checking. The infrastructure behind it is something different: an AI-assisted content pipeline that lets a small team publish with the consistency and volume that used to require a staff ten times the size.
This is not AI replacing journalists. It is AI extending what journalists can do.
The Charlotte Mercury launched in this model. The Farmington Mercury — covering Farmington, Connecticut — followed. Strolling Ballantyne serves a Charlotte neighborhood with the same discipline and a gentler pace.
Every article published is evidence that local journalism can be rebuilt — not as a charity case, not as a legacy brand on life support, but as a viable, technology-forward business. Langran's seminar got the diagnosis right twenty-odd years ago. This is my answer to it.
On AI, LLMs, and Keeping It Honest
I have been working with large language models — LLMs — since before most people had heard the term. Not as a hobbyist. As an operator who needed them to perform to editorial standards, in production, on deadline.
The gap between what AI does in a demo and what it does in the real world is vast. Closing it requires what I call agentic control: systems where the model operates within defined constraints, is checked against reality at each step, and is never the final word on anything that matters.
Token efficiency matters, too. Every token is compute. Every unnecessary word in a prompt is waste. I think about AI workflows the way a good editor thinks about a sentence: if it works but it's carrying extra weight, you're not done.
I write about all of this at petercellino.com/notes.
The Publications
- The Charlotte Mercury — Charlotte, NC. Sports, government, business, culture.
- The Farmington Mercury — Farmington, Connecticut. Government, police, development, education.
- Strolling Ballantyne — Ballantyne neighborhood, Charlotte. Community, dining, local life.
Get in Touch
If you want to talk about AI in media, local news economics, platform architecture, or working together — contact me or find me at pcellino@gmail.com.
If you're a reader of one of the Mercury publications: thank you. This is for you.