Anthropic made Claude smart reports generally available for Enterprise on October 28, following a beta launched September 10. The feature connects a sample of team conversations with spending, giving administrators a starting point for examining how Claude is used.

The big change

  • Budget owners can examine categories of work alongside spending instead of relying solely on an aggregate usage figure. The categories are generated from a sample, so the report is a lens on selected activity rather than a complete account of the organization’s work.
  • A costly workstream can direct attention to its sessions and reported output. The value of that output still depends on what happened after the session, which calls for evidence beyond the report.
  • Organizations need to decide who should be able to inspect reports and understand the controls before sharing them.

What a smart report can show

The October 28 announcement says the feature is generally available; release notes record the earlier beta. Anthropic says Claude groups sampled transcripts into workstreams and outputs, with spending and common friction alongside them.

A report's categories and summaries are Claude's interpretation of the selected conversations. They can point an administrator toward a pattern worth checking, but a pattern in the sample does not establish its frequency across every session or the entire organization. The guide does not specify a sampling rate or method.

A generated output is evidence of activity, not proof that a document was accepted, code shipped or time saved. Those outcomes require separate evidence. A comparison between reports also needs consistent scope: changing the team, products or dates changes the pool from which the sample is drawn, so a difference between reports cannot by itself establish a change in behavior.

The smart reports guide says the chosen period can cover up to 28 days and spending is shown at list price. List price need not equal the customer's negotiated bill.

Who can see the detail

Reports are off by default. Owners, admins and custom roles with analytics view access can create and view them. Owners can delegate scoped access to team leads without making them admins. Named attribution is also off by default; even when enabled, names and session IDs remain hidden until a viewer explicitly reveals them, and reveals are logged. Sessions identified as restricted are included only in aggregate, without summaries. Anthropic says reports should not be used for individual performance evaluation or employment decisions.

The guide excludes Enterprise organizations using customer-managed encryption keys, HIPAA configurations or hybrid inference in their own cloud. Claude Code is excluded when its zero data retention setting is enabled. Organizations should verify their configuration and terms before planning around the feature.

The reporting controls concern visibility within the organization. Anthropic's separate commercial data policy says customer chats and coding sessions are not used to train models unless the customer opts in through described mechanisms, including explicit feedback. That policy addresses a different use of the data from report access.

BIG CHANGE reviewed Anthropic's announcement and documentation; it did not generate a report or test the controls.