Claude Enterprise Now Shows What Your $20 Seat Actually Buys

Anthropic is selling enterprises a receipt for their token bill. On October 28, 2026, the company made three Claude Enterprise features generally available: Smart reports, in-app surveys, and pooled group budgets. Together they answer the question finance teams have been asking since AI budgets started climbing: what did we actually get for the money, and who spent it?

Quick answer: Smart reports analyze a sample of team transcripts and price the work at list rates, with 10 reports a month free per organization. In-app surveys ask employees what Claude changed about their work. Pooled group budgets give each team one shared monthly spending pool on top of individual limits. Enterprise still costs $20 per seat per month plus usage at API rates, billed annually, so these tools track the bill rather than cutting it.

What launched, and who gets it

All three features are for Claude Enterprise, Anthropic's plan for large organizations. They are not add-ons for Pro or Max accounts, and Anthropic has published no separate price for them: they arrive as part of the Enterprise plan. Coverage of the launch puts the three features in beta since September 2026, with Smart reports in beta since September 10, before the October 28 general availability date. Anthropic's own announcement dates the general availability to October 28 and notes one availability caveat up front: exactly what an organization gets varies by configuration, including for organizations with a HIPAA configuration or customer-managed encryption keys. The Help Center pages carry the detailed eligibility rules, so an admin should check those before promising a rollout date internally.

That Enterprise-only scope matters for the price discussion below. These are admin tools for organizations already paying the Enterprise seat price and metered API usage. They do not change what a seat costs, and they do not lower the per-token rates. What they change is visibility: an admin can now see spend by workstream, hear from the people doing the work, and stop a team's spend at a line the admin chose.

Smart reports: a sample of transcripts, priced at list rates

A Smart report starts with three choices: a team, a date range of up to the last 28 days, and the products to include, such as chat, Claude Code, Claude Cowork, or the unified chat and Cowork experience. Claude then reads a sample of that team's transcripts, groups the work into workstreams, attaches spend to each group, and writes up what it found. Every chart is interactive: clicking a workstream opens the sessions inside it, ranked by cost, each with a one-paragraph summary, the product used, the date, and the output type. Reports can be downloaded as HTML with those drilldowns intact.

The report sections read like an audit file. Workstreams shows what the group used Claude for most, by sessions and by spend, side by side. Deliverables produced groups sessions by output type, such as analysis, documentation, content drafts, and code. Cost per session by type of output shows the average spend for each kind of output, so an admin can see what is cheap or expensive to produce. Task outcomes, most common pain points, inefficiencies, complex autonomous work, and most expensive sessions round out the picture. Pain points are categorized, for example a connector that was never set up, output that did not match the ask, approval or sign-in gating, tool failures, or rework loops.

Two details deserve attention. First, the costs in a Smart report are list-price costs. Anthropic's support guide says reports show what work costs "at list price," which means the numbers will not match a discounted invoice. Treat the report as a way to compare workstreams against each other, not as a billing statement. Second, a report analyzes a sample of transcripts, not every session. That is a reasonable design for a tool that takes a few hours to generate, but it means a report is an estimate of where work and spend concentrate, not a census.

The recommendations section is where the report tries to earn its keep. It brings together recurring workflows worth turning into skills, examples of complex work from across the team, and rules to add to a code repository. Findings can be shared as an artifact with the people who can act on them. The report also flags people whose work is being held back by a spend limit, and an admin with billing permission can raise that limit from the report itself. Admins can also save custom questions and custom insights as templates to reuse in future reports, and owners can let team leads who are not admins run reports for their own groups.

On price and limits, the support guide is specific: each organization can run up to 10 reports per month at no charge. Reports that are queued or in progress count toward that limit; reports that fail do not. When an organization hits the limit, it cannot create new reports until the limit resets on the first day of the next calendar month. If it needs more reports before then, the guide points to the account team, which means there is no published self-serve price for report number 11. Smart reports are also opt-in: an Owner or Primary Owner enables them for the organization, and coverage describes them as disabled by default.

Privacy guardrails are built in, and Anthropic draws a hard line on one use. Smart reports are not designed for evaluating individual performance or making employment decisions, and the announcement says so plainly. Sessions involving restricted topics, including personal conversations, are used only for aggregate analysis, with no summaries, no individual sessions, and no names. Those guardrails are always on and cannot be turned off, according to the support guide. Admins can delete any report.

In-app surveys: the team's own verdict, in Claude itself

Smart reports show what happened. In-app surveys ask the people it happened to. Admins create a survey that appears as a short card after a reply in the tool employees are already working in. The default questions focus on value: the most valuable thing someone has done with Claude recently, how Claude changed the time that work took, which use of Claude a teammate should copy, and what would break or slow down first if the team lost access tomorrow. Admins can customize or remove questions, choose who receives a survey, when it runs, and where it appears.

Responses appear on the survey page as they arrive, and admins can export them to a spreadsheet. That export matters: survey data that stays locked in an admin console is easy to ignore, while a spreadsheet can sit next to the spend report in a budget review. Anthropic says surveys are treated as the customer's data, accessed only as necessary to operate the service. Two configuration notes from the announcement are worth flagging: in-app surveys can now be enabled for organizations that use customer-managed encryption keys, and for organizations that set one retention period for all data, surveys and their responses are deleted on that schedule.

The honest limitation is that surveys are self-reported. An employee saying Claude saved time is not the same as a measured time saving, and survey response rates will shape how much weight any result can carry. Used alongside Smart reports, though, the combination is stronger than either alone: the report says what the work cost, and the survey says whether the people doing it thought it was worth it.

Pooled group budgets: one pool per team, individual limits intact

Pooled group budgets solve a mundane but real problem. Under individual limits alone, a heavy user can hit a cap early in the month while a colleague's budget sits unused. A pooled budget gives the group one shared monthly budget, while member limits stay in place as a guardrail against unusually high individual use. Every request counts against both limits, and a member stops at whichever runs out first. When someone hits a limit, they can request more from inside Claude, and the request goes to admins for review.

Setup is deliberately low-friction: the setup flow suggests an amount based on the group's recent usage, so an admin is not guessing from zero. Admins with billing permissions receive email alerts when spending reaches 50%, 75%, 95%, and 100% of a budget. Coverage of the launch adds that users can belong to multiple pooled budgets, which administrators can rank by priority, a detail worth confirming in the Help Center before designing a budget structure around it.

What pooled budgets do not do is reduce token costs. They cap spend and redistribute it within a team. A team that burns through its pool still faces the same choice as before: request more budget, wait for the month to reset, or do less. The value is control and predictability, not savings.

What it costs, honestly

Anthropic's pricing page puts the Enterprise plan at $20 per seat per month plus usage billed at API rates, with Enterprise billed annually. Prices shown do not include applicable tax. That fine print is the whole pricing story for these features: the seat price buys access, the API rates decide the variable bill, and the new tools help an organization watch both. For comparison, Pro is $20 billed monthly or $17 per month with an annual subscription ($200 billed up front), and Max starts at $100 per month. Those plans do not include these admin tools.

The one genuinely free quantity in this launch is the 10 Smart reports per month per organization. Everything beyond that is unpriced in public: no per-report price, no published overage rate, just a pointer to the account team. And because Smart report costs are list-price figures, an organization with negotiated discounts should expect its real invoice to differ from its report totals. None of that makes the tools useless, but it does mean the launch is a visibility upgrade, not a discount.

Pros and cons

Pros: spend broken down by workstream and output type instead of one lump sum; friction points named, such as missing connectors and failed tool calls; recommendations that point at reusable skills and repository rules; employee feedback collected inside Claude and exportable; pooled budgets that stop one heavy user from being capped while teammates leave budget unused; clear alert thresholds; and privacy guardrails that keep personal conversations out of drilldowns.

Cons: Enterprise only, so smaller teams on Pro or Max get nothing here; report costs are list prices, not invoice prices; reports analyze a sample, not every session, and take hours to generate; only 10 free reports a month per organization with no published price beyond that; surveys are self-reported; budgets cap spend without lowering token rates; and availability varies for HIPAA and customer-managed-encryption-key configurations, which Anthropic does not spell out in the announcement itself.

FAQ

Are Smart reports, surveys, and pooled budgets included in the Enterprise price?

Anthropic presents them as generally available for Claude Enterprise and publishes no separate price for them. The seat price remains $20 per seat per month plus usage at API rates, billed annually. The only free quantity stated is 10 Smart reports per organization per month; Anthropic publishes no price for additional reports.

Do Smart reports read every transcript a team produces?

No. A report analyzes a sample of transcripts within the chosen team, date range of up to 28 days, and products. Costs are shown at list price, and reports usually take a few hours to generate.

Can managers use Smart reports to evaluate individual employees?

Anthropic says no. The announcement states Smart reports are not for evaluating individual performance or making employment decisions, and restricted topics such as personal conversations appear only in aggregate, without summaries or names.

What happens when a pooled budget runs out?

Every request counts against both the shared pool and the member's individual limit, and the member stops at whichever runs out first. The member can request more from inside Claude, and admins with billing permissions get email alerts at 50%, 75%, 95%, and 100% of a budget.

Do these tools work for every Enterprise configuration?

Not necessarily. Anthropic says availability varies by configuration, including for organizations with a HIPAA configuration or customer-managed encryption keys, and points admins to the Help Center for eligibility and setup steps. Surveys, specifically, can now be enabled for customer-managed-encryption-key organizations.

Key takeaways

• Anthropic made Smart reports, in-app surveys, and pooled group budgets generally available for Claude Enterprise on October 28, 2026, after a beta period starting in September.
• Smart reports price a sample of team transcripts at list rates, cover up to 28 days, take hours to generate, and are free for 10 reports per organization per month, with no published price beyond that.
• In-app surveys collect self-reported value and time-saved feedback inside Claude, with customizable questions and spreadsheet export, but they measure opinion, not output.
• Pooled budgets pair one shared monthly team pool with individual limits, count every request against both, and alert billing admins at 50%, 75%, 95%, and 100%.
• Nothing here lowers the bill: Enterprise still costs $20 per seat per month plus API-rate usage, billed annually, so treat this launch as a control panel, not a discount.

Read the launch in context and it looks less like a feature drop and more like a response to a procurement conversation. Enterprises that adopted Claude at speed now have finance teams asking for evidence, and Anthropic has answered with the three artifacts a budget review actually uses: a spend breakdown, a user survey, and a cap. Whether that evidence convinces anyone depends on what the reports show once real teams run them, and on October 29, 2026, that testing has barely begun.

Sources: Anthropic's official announcement on claude.com; Anthropic's Smart reports support guide; Anthropic's pricing page; Unite.AI, Daily AI Brief, and Cryptopolitan launch coverage. Internal links: our Claude Code explainer and our Claude Chat and Cowork Enterprise merge coverage.