Anthropic regularly publishes its Economic Index — a look at what people actually use Claude for in daily life. The new edition, out June 26, is called “Cadences,” and the name says it all: instead of the old seven-day snapshots, Anthropic now looks at usage hour by hour. The result is a surprisingly human picture.
The day has a rhythm — and Claude knows it
The data reads like a daily schedule. People ask for news at 7 a.m. Recipe requests are 2.3 times more common at 6 p.m. than average. Sleep advice clusters in the hours before dawn, around 5 a.m. And around the US tax deadline on April 15, tax-related questions spiked to eight times a normal day.
On weekends the whole balance flips: the share of personal conversations jumps from about 35% on weekdays to just under 50%. Business correspondence and slide decks give way to emotional support, medical questions, and investment advice. Interesting for us coders: backend architecture and API debugging drop on weekends — while AI agent design, quant trading, and gaming rise. The weekend is clearly the playground.
More autonomy lives in the tool, not the model
One detail stuck with me. Anthropic measures how much decision-making Claude gets in a conversation — and it’s noticeably higher in Claude Code than in chat or Cowork. The best example: a blog post takes a median of 13 rounds of back-and-forth in chat. In Claude Code, a single prompt does it.
The striking part: the gap holds even for the same underlying model. So it’s the surface, more than the model, that shapes how much we delegate. As an aside, Claude tends to answer about one year of education above the level it was asked at — most of all when you ask it to build something.
The survey is the real headline
For the first time, Anthropic links a survey (around 9,700 respondents) to actual usage data. Over a third expect AI to be able to do most or nearly all of their work tasks within twelve months. 10% rate their own job loss as likely.
And here’s the twist: the people who delegate the most to Claude are the most optimistic — about pay, job security, and meaning. If you hand entire tasks to AI, you apparently see more of what it can do, and that makes you feel safer, not more threatened. Asked what they dream of in ten years, more than half gave the same answer: collaborating with AI on meaningful work, automating away the drudgery — and seeing the gains shared broadly.
My take
Reports like this aren’t marketing, they’re self-observation at scale. What fascinates me is how mundanely human the patterns are — recipes in the evening, sleep advice at dawn. And the finding that heavy delegators are more optimistic matches my own experience: people who actually use AI, rather than fearing it from a distance, get a more realistic sense of what’s possible. That’s exactly the stance I try to take here.
Sources: Anthropic Economic Index report: Cadences, The Anthropic Economic Index