Claude Enterprise adds smart reports, surveys and shared budgets
Anthropic made smart reports, in-app surveys and pooled group budgets generally available for Claude Enterprise on October 28. Together, they address three different questions: what teams are doing, what users think is useful and where shared spending capacity should sit. General availability does not mean every deployment configuration supports all three.
Smart reports inspect workflows, not employee performance
The smart reports guide says reports sample transcripts, group work into workstreams and identify costs at list price and recurring obstacles. They are off by default. Owners must enable them before authorized analytics users or scoped delegates can run reports.
An organization gets up to ten reports monthly at no charge. Queued and running reports count, while failed reports do not. A report can cover up to the previous 28 days. Generation usually takes hours, not an instant dashboard refresh.
Smart reports exclude organizations using customer-managed encryption keys, HIPAA configurations or hybrid inference on their own cloud. Claude Code zero-data-retention configurations omit Code from reports while allowing chat and Cowork coverage.
Individual attribution is a separate opt-in setting and names remain hidden initially. Revealing names is logged. Anthropic explicitly says reports should not be used for individual performance evaluation or employment decisions. The feature’s stated purpose is adoption and investment planning.
Surveys are feedback, not anonymous polling
In-app surveys appear after a response or between tasks in chat, Cowork or Claude Code. Admins choose the audience, schedule, surfaces and questions. Dismissing the card declines participation, and the survey does not interrupt a running task.
Responses go to the organization’s admins. The downloadable CSV includes the respondent’s email, group, submission time and product surface, so readers should not assume anonymity. Surveys support customer-managed encryption keys but remain unavailable for HIPAA configurations. The guide notes a transition caveat: data from roughly the first day after adding a customer-managed key remains encrypted with Anthropic’s key.
For a useful pilot, ask about one repeatable task and the obstacle that prevented completion. Compare that feedback with workflow-level findings rather than turning a response rate or session count into a worker ranking. This is an editorial evaluation suggestion, not a measured outcome from deploying these tools.
A shared budget does not replace individual limits
The pooled-budget documentation covers usage-based Enterprise plans and Standard or Premium seats with usage credits enabled. On seat plans, the pool applies only to usage credits, not the usage included with the seat.
Requests count against the member’s own monthly limit and an available shared pool. If a person belongs to several pools, the largest pays first by default, with the next eligible pool taking over after depletion. Admins can choose a custom priority. That order is separate from the setting that decides which group’s member limit applies.
| Illustrative single-pool case | Result |
|---|---|
| Member limit reached, pool remains | Member cannot keep spending against the pool |
| Pool exhausted, member limit remains | Usage credits stop without another eligible pool |
| Seat-included usage remains | That usage is outside the pooled credit budget |
Budget editing requires billing-management permission, which applies organization-wide rather than to one selected group. Pooled budgets are managed in Organization settings and are not yet available through the Admin API.
Check eligibility before designing the rollout
The practical first step is a feature-by-feature check, not one Enterprise checkbox. An organization may qualify for surveys but not smart reports. A team may have a shared pool but still encounter a member limit. Decide who can read reports, who can see survey exports and who may change spending controls before inviting a pilot group.
ByteForward reviewed the announcement and documentation but has not enabled these features or tested them. Their value should be assessed through completed workflows, unresolved obstacles and actual spending, not an assumption that more visibility automatically produces savings.
