Anthropic's Economic Index is now a Claude connector — ask it anything about AI and work

Anthropic has added a new connector to Claude.ai that lets you query the Anthropic Economic Index directly from a conversation. No downloads, no data exports, no pivot tables. You ask a question in plain English and get an answer grounded in real usage data.

If you’ve heard of the Economic Index but never dug into it, here’s the short version: since early 2025, Anthropic has been using a privacy-preserving analysis tool called Clio to study how Claude is actually being used across the economy. Not hypothetically, not based on surveys, but from anonymised patterns across roughly a million conversations, mapped onto around 20,000 work-related tasks from the U.S. Department of Labor’s O*NET framework. The results have been released in a series of public reports and open datasets, and they paint a surprisingly specific picture of which jobs, tasks, and regions are seeing meaningful AI adoption right now.

Until now, exploring that data meant downloading CSVs from Hugging Face or reading through research reports. The connector changes that.

How to turn it on

Open claude.ai, go to the connectors menu, find the Anthropic Economic Index in the directory, and enable it. That’s genuinely it. It works in any conversation with any Claude model, and there’s nothing to install or configure. The whole process takes under a minute.

What you can actually ask

Once it’s enabled, you can ask the kinds of questions that would previously have required someone to pull and analyse the data for you:

• “Which occupations use AI the most?”
• “What are the most common ways people in Colorado use Claude?”
• “What sorts of tasks do teachers use Claude for?”
• “What kinds of tasks are people automating with AI?”

The connector grounds Claude’s responses directly in the Index data, so you’re not getting speculative answers. You’re getting answers derived from observed usage patterns, with Claude able to point you to the underlying source data as you go.

A good approach is to start broad and then drill down. Ask something like “What does the Index say about how AI is being used in healthcare?” and then follow up with more specific questions about particular roles or task types. Claude will tell you when it’s reached the limits of what the data can show.

What the data actually covers

It’s worth being clear about what you’re looking at and what you’re not. The Index reflects how Claude is being used, not AI usage across the industry as a whole. That’s a meaningful distinction. The data comes from claude.ai conversations (primarily consumers) and the first-party API (primarily businesses). Other AI products aren’t represented.

There are also some known skews in the data. Because Claude is often marketed as a capable coding assistant, coding tasks may be overrepresented relative to other use cases. Creative tasks involving image generation don’t show up because Claude doesn’t generate images natively. Anthropic flags these limitations clearly, and the connector will surface them when relevant.

What the data does show is genuinely interesting. One of the Index’s headline findings is that early AI adoption is uneven. Claude is used more intensely in higher-income countries, and within the US, more heavily in areas with a higher concentration of knowledge workers. Usage concentrates around a relatively small set of specialised tasks and occupations. That’s not a surprise exactly, but having actual data behind it matters, particularly for anyone thinking about workforce policy or skills investment.

The datasets are released under a CC-BY licence, with code under MIT, so anyone can work with the raw data directly. The connector just makes it accessible to people who don’t want to do that themselves.

What this means for you

If you work in HR, policy, education, or any field where questions about AI and jobs keep coming up, this gives you a quick, conversational way to check what the data actually says rather than relying on generalised takes. You can look at specific occupations, specific task types, and specific geographies.

For researchers and journalists, the connector is a faster on-ramp to understanding what the Index covers before deciding whether to go deeper into the raw datasets.

For everyone else, it’s a reasonably honest answer to the question “what is AI actually being used for?” Most of the discourse around AI and work operates at a level of abstraction that makes it hard to reason about. The Economic Index is one of the few attempts to ground that conversation in observed behaviour, and the connector makes it easier to explore.

The full datasets remain freely available if you want to work with them directly. The connector is available now in claude.ai for free.