Real Story
A national park found 7,000 days of staff time in one afternoon
Stanford's Jeremy Utley tells this story: a facilities and operations employee at Glen Canyon National Park, not a technical person, sat through one two-hour AI training session. Afterward, on his own time, he spent about forty-five minutes building a tool for the paperwork he dreaded most, the kind of work that used to eat two or three days every time it came around.
Other parks picked up what he built. By the Park Service's own estimate, that one afternoon of tinkering is expected to save roughly 7,000 days of staff time this year (per Jeremy Utley, Stanford, EO interview, 2025, as he tells the story).
No coding. No IT department. Two hours of training and forty-five minutes of curiosity, from someone whose job had nothing to do with technology.
Real Data
The gap that matters isn't rich versus poor. It's practiced versus new.
Anthropic looked at more than a million real conversations people had with its own AI product. The people who had been using it for six months or more had roughly a 10% higher success rate (per Anthropic's own Economic Index) than people who were newer to it.
Here's the part worth sitting with: that gap was not explained by which country someone lived in, what kind of task they were doing, or which AI they happened to use (per Anthropic's Economic Index). It came down to practice. People who stayed with it long enough to get a feel for it did better than people who tried it once and put it down.
That's genuinely good news for a small business owner. The gap that matters isn't about money or size. It's about getting your hands on it and staying with it a little while. The only way to start closing that gap is to start.
We'll say this plainly every time we cite it: Anthropic is measuring its own product here, and this is a correlation, not a guarantee.