Claude Enterprise Gets Smart Reports That Flag Off-Topic Use
Anthropic on Oct. 28 made three administration features generally available for Claude Enterprise: smart reports, in-app surveys, and pooled group budgets, according to Unite.AI and AIPolix. All three had been in beta since September, reported TheTechStreetNow, summarizing ZDNET. PivotNews, also citing ZDNET's framing, described the release as an answer to enterprise buyers who want clearer evidence that AI spending matches productivity gains. The features are Enterprise-only and do not extend to Team or Pro plans, according to Adaptd.
To run a smart report, an admin picks a team, a date range, and the products to include, and can attach documents to guide the analysis. Claude then reads a sample of transcripts in that scope, groups the work into workstreams and output types, attaches spend at list price to each group, and writes up the findings, according to Unite.AI. Each report usually takes a few hours to generate and can be downloaded as HTML with drilldowns intact. Charts are interactive: managers can click a workstream to open the sessions inside it, ranked by cost, each with a one-paragraph summary, the product used, the date, and the output type. Admins can also ask custom questions about usage and cost, save them for reuse, and share findings as artifacts, while reports recommend recurring workflows worth turning into skills and rules for code repositories.
The most talked-about part is what Anthropic calls "inefficiencies." Mashable reported that smart reports show obstacles encountered and how often they cropped up, alongside sessions that produced no usable output and ones classified as "personal or off-topic." TheOutpost reported the same classification. There are also hard limits on custom questions: queries about race, gender identity, religion, political affiliation, immigration status, health, union activity, harassment complaints, or workers' personal lives are not accepted, according to Mashable. Reports are off by default, Mezha noted in its summary of the ZDNET reporting.
Pooled group budgets replace strictly identical per-person caps with one shared monthly budget per team, while individual limits remain as ceilings and every request counts against both, according to AIPolix and PivotNews. Anyone who runs out can request more from inside Claude. Unite.AI, citing the help documentation, reported the sizing rule of thumb: set the shared budget at about 3 times the group's prior-month spend, and the per-member cap at about 3 times the highest member's spend, with a worked example of 1,450 dollars in group spend yielding a 4,350-dollar pool and a 310-dollar top spender yielding a cap of about 950 dollars. Everyone with billing access gets email alerts at 50, 75, 95, and 100 percent of usage, the usage bar turns amber at 75 percent and red at 90 percent, and the docs suggest raising a budget by 50 percent if 75 percent is already used before the 20th of the month.
Anthropic's documentation states that smart reports aren't designed for and should not be used for evaluating individual performance or making employment decisions, Mezha and Adaptd both noted. Delegated access lets team leads run reports on chosen groups without becoming full admins, with an optional attributed view for shared readers. Availability varies by configuration, including exclusions tied to HIPAA and customer-managed encryption keys. Unite.AI reported statements from Monday, Slalom, Wix, Deloitte, DoorDash, and Docusign in the announcement, including Wix engineer Ran Yitzhaki's comment that each smart report is based on tens of thousands of sessions. For surveys, AIPolix reported that admins can rewrite or drop questions, choose who sees them and when, and export answers to a spreadsheet, with respondents getting a brief prompt card after Claude answers. What comes next is whether companies use the new visibility to fix friction or to watch workers more closely.