SI Agent News
DevelopingClaude

Claude Enterprise adds usage reports, staff surveys and shared budgets for managers

Anthropic added three tools to its business plan to help companies show what AI spending delivers, and to control it.

The short version

Why it matters

Companies are starting to ask whether AI tools pay for themselves, and the answer has been hard to show. These features give managers numbers and staff feedback to point to. They also mean your own AI use at work may be measured more closely.

Why Anthropic built these

As itmedia.co.jp reported, Anthropic added three features to Claude Enterprise, its plan for businesses. The timing fits a mood in many companies: some are seeing AI costs run past their budgets, and IT departments are being asked to justify the spend.

Anthropic says many customers already have staff using Claude in their own ways. The next step is spreading that use to every team, and the new tools are meant to give administrators what they need to do that.

Smart reports: what the work looks like

An admin picks a team, a time period and a model. Claude then reads part of that team's conversations and groups the work into workstreams, meaning recurring kinds of tasks.

For each workstream, the report shows what it cost, what was produced and what got in the way. Anthropic's examples of obstacles are failures with connectors (links to outside services) and failed tool calls.

The report also suggests next steps. For example, a task people repeat often could become a skill, a saved routine Claude can follow. Admins can share findings and add their own questions about how a model is used.

Owners can give team leaders limited access to reports for their groups. Anthropic says the reports should not be used to judge individuals or make hiring decisions.

In-app surveys: asking the people who use it

Numbers from software only tell part of the story, so this feature asks users directly. A short card appears after a Claude reply, inside the tools people already use.

The standard questions cover what work they did with Claude, how it changed the time the work took, what to use it for next, and what they would lose if Claude went away. Admins can edit or delete questions, choose who gets them and decide where they appear.

Answers can be exported to a spreadsheet. The feature also works for organizations that manage their own encryption keys. Where one retention period is set for all data, survey answers are deleted on that schedule.

Pooled group budgets: sharing the spend

With personal spending limits alone, some people hit their cap early while others leave money unused. A pooled budget gives a group one shared monthly amount instead.

Anthropic describes personal limits as a brake on sudden spikes. Each request is checked against both the group budget and the personal limit, and use stops when either is reached. A user who hits a limit can ask for more from Claude's screen, and an admin approves or declines.

When setting a budget, admins see a suggestion based on the group's past use. They also get emails as spending reaches set points.

What to know before relying on it

All three features go to Claude Enterprise customers. But organizations set up for the US health-privacy law HIPAA, or using their own encryption keys, get different access depending on their configuration. Anthropic explains the details in its help center.

What to watch

Try it

If your workplace pays for Claude, ask your admin whether usage reports or surveys are switched on. Then note one task it saved you time on.

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by itmedia.co.jp · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

More AI news