The problems are the same. The budgets are not.

The Enterprise Version

Brian Alvey is the CTO of WordPress VIP, the enterprise division of Automattic. His team — roughly 200 people — hosts websites for News Corp, NBC, the White House, and NASA. In a recent conversation on The AI Report, he described the challenge keeping all of them up at night: brand governance at production speed.

WordPress VIP built a governance plugin that constrains infinite design choices — fonts, colors, layouts — down to what the brand permits. The problem it solves is drift. A thousand people touch the website every day. Without the system, somebody puts an orange headline in Comic Sans.

The Local Version

We run Mercury Local — three publications, five domains, one publisher. The Charlotte Mercury, The Farmington Mercury, Strolling Ballantyne. Each has a distinct editorial voice, defined down to the words each author uses and the words they never use.

If you've ever read a local news article and couldn't tell whether it was written for your city or any other city in America — same generic lede, same sourceless claims, same non-voice — you've seen editorial drift without infrastructure. At enterprise scale, drift puts an orange headline on the homepage. At editorial scale, it puts a story on your site that nobody trusts enough to finish reading.

Same Architecture, Different Magnitude

Alvey frames the solution as documenting "the hundreds of steps" at both the operation level and the individual story level until you arrive at what he calls "a machine that is your business." At WordPress VIP, that machine is governance plugins enforced across enterprise newsrooms. At Mercury Local, it's an Obsidian vault with editorial rules, beat research files, and publication records that make the work auditable after the fact.

The principle is identical: encode what you know into something the system can enforce.

Why Small Operations Need It More

The conversation about AI in publishing is dominated by scale. Enterprise teams deploy AI across hundreds of writers. Startups wonder whether they can afford the tools. That framing lets small operations off the hook — if you only have three reporters, the thinking goes, you don't need governance.

Exactly backwards.

The smaller the operation, the more the system matters. A 200-person newsroom can afford an editor who reads every draft, a copy desk that checks every fact, a brand team that reviews every headline. A one-person newsroom builds those functions into the infrastructure or they don't exist.

Structure Over Speed

Alvey described his enterprise customers this way: "Every single one of them feels like they're doing just enough AI to stay behind."

The answer isn't more AI. It's more structure. The organizations that will use AI well — at any size — are the ones that did the unglamorous work of defining what good looks like and building a system that enforces it.

That work doesn't require 200 people. It requires the discipline to do it at all.

Scale is a spectrum, not a category.