What happened
Anthropic announced on October 28, 2026 that three Claude Enterprise features are now generally available: smart reports, in-app surveys, and pooled group budgets. The stated aim is to help organizations expand useful Claude work beyond the individual employees who first discover it. Availability can vary by configuration, including HIPAA configurations and customer-managed encryption keys.
Smart reports are intended to show what a selected team does in Claude and what that work costs. An administrator chooses a team, date range, and products to include. Claude then analyzes a sample of the team’s transcripts and organizes the work into workstreams. The resulting report shows spending by workstream, deliverables produced, and common friction points, which Anthropic says can include a missing connector or a failed tool call. The reports are based on transcript samples; Anthropic describes them as a tool for adoption and investment planning, not for individual performance evaluation or employment decisions.
Each report produces a consolidated recommendation list. Recommendations can include recurring workflows to turn into skills and rules to add to a code repository. Findings can be shared as artifacts, while custom questions about model use and cost can be saved as templates for later reports. Billing-permitted administrators can raise a spend limit from a report when that limit is blocking a user. Enterprise owners can also allow non-admin team leads to run reports for their own groups, moving some action closer to the people managing the work.
In-app surveys add a direct feedback channel. Default questions ask about a user’s most valuable recent Claude task, how Claude changed the time required, which use a teammate should copy, and what would break or slow down if access disappeared. Administrators can change or remove questions and decide who receives a survey, when it appears, and where it runs. Members see a short card after a reply in the Claude tool they are using; administrators can view responses as they arrive and export them to a spreadsheet. Surveys can be enabled for organizations using customer-managed encryption keys. Where an organization has one retention period for all data, survey responses are deleted on that schedule.
Pooled group budgets establish one shared monthly budget for a group. The design is meant to reduce the chance that heavy users hit an individual limit while capacity assigned to lighter users remains unused. Individual member limits still apply as a guardrail. Every request counts against both the group and member limits, and access stops when either is exhausted. Members can request more capacity in Claude for administrator review. Setup suggests an amount using recent group usage, and billing administrators receive email alerts at 50%, 75%, 95%, and 100% of the budget.
What the archive adds
This is a new fact in the continuing Claude Enterprise story rather than a restatement. Earlier coverage recorded the general availability of a unified Cowork-and-chat experience for Claude Enterprise. It also covered Claude’s move toward generated, editable business artifacts through Dashboards and Motion. The new release adds an administrative layer for observing and scaling that work.
Trend signal
This supports AI-native organizations move toward review and orchestration. Smart reports turn usage traces into workstreams, friction points, and recommended skills; surveys collect judgments that logs cannot provide; pooled budgets give administrators a mechanism to redirect capacity. The product emphasis is not merely on making each employee faster, but on specifying what should be repeated and assigning people to review and act on evidence.
Why it matters
Enterprise AI programs often have a visibility problem: aggregate spend shows consumption, while isolated chat examples do not show which workflows are reliable enough to standardize. Claude’s new controls combine behavioral evidence from a transcript sample, explicit feedback from users, and budget administration in one product surface. That can help a team distinguish a promising repeated workflow from a one-off success, identify operational blockers such as missing connectors, and decide whether a skill, repository rule, or training intervention is warranted.
The features do not remove the need for governance. They do, however, provide a structured route from observed use to an operational response. For teams already on Claude Enterprise, that may be more actionable than treating adoption reporting as a separate analytics project.
Limits and uncertainty
Anthropic’s announcement describes the features and their intended uses, but it does not provide independent evidence that report recommendations improve outcomes or that transcript sampling accurately represents a whole organization’s work. It also does not specify sampling methodology, the quality threshold for a recommended workflow, pricing implications, or detailed eligibility across every Enterprise configuration. Smart reports are explicitly not designed for individual performance assessment. Organizations will still need their own policy for access, review, retention, and deciding which reported patterns deserve to become formal skills or rules.
Skill to build now
Create a small review loop for one AI-enabled team this week. Define a monthly question set: which repeatable task produced useful output, which step failed or required escalation, and what capability or connector would remove that friction. Assign an owner to review answers alongside spend, then require one concrete disposition for each candidate workflow: document it, turn it into a skill or repository rule, fix the blocker, or reject it. This mirrors the operating model the new Enterprise features are designed to support.
Reader takeaway
Claude Enterprise is adding a management layer around AI use: reports identify repeatable work and friction, surveys capture employee judgment, and pooled budgets move capacity to active groups. For builders, the useful question is not only how much AI is being used, but which observed workflows deserve to become governed, reusable capabilities.
Open question
Will organizations use these reports to create durable, reviewed skills and rules, or will the recommendations remain another administrative dashboard without an owner or follow-through?