Anthropic has rolled out 'smart reports' for Claude Enterprise users, a feature designed to give managers granular visibility into how their teams interact with the AI. Announced on Wednesday, the tool generates visual graphs detailing usage patterns, output types, and associated costs. Available immediately on paid enterprise plans, the feature requires admins to enable it within organization settings, signaling a shift toward data-driven management of AI adoption in the workplace.

How Smart Reports Analyze Team Activity

Managers can configure reports by selecting specific teams, defining date ranges up to the last 28 days, and choosing which Claude products to include. Supported tools include chat, Claude Code, Claude Cowork, and the unified chat and Cowork experience, though the recently announced Claude Duet agent is currently excluded. The system ingests conversation transcripts, automatically grouping them into workstreams and categorizing outputs such as 'vibe code' or technical documents. It also aggregates spending data, allowing admins with billing permissions to adjust limits directly from the report interface.

Custom Queries and Interactive Insights

Beyond standard metrics, managers can ask custom questions about usage and cost, which can be saved for future reports. Smart reports also provide recommendations, such as suggesting new rules for code repositories. The interactive charts allow users to click through to see session breakdowns, which can be filtered and ranked by cost, product, date, or output type. This level of granularity helps organizations understand not just what is being produced, but where obstacles occur and how often they arise.

The Rise of AI Surveillance Concerns

The reports do more than just track productivity; they actively flag 'inefficiencies,' a category that includes sessions producing no usable output and those classified as 'personal or off-topic.' This capability arrives amid growing anxiety about AI in human resources. Recent surveys indicate that one in four U.S. managers use AI for layoff decisions 'often or all the time.' Furthermore, former Meta employees alleged in July that the tech giant used improperly tested AI tools to rank workers for mass layoffs, though Meta denied that AI made the final decisions.

Anthropic’s Disclaimer and Usage Limits

Aware of the potential backlash, Anthropic explicitly states that smart reports 'are not designed and should not be used for evaluating individual performance or making employment decisions.' The tool is intended to help organizations understand adoption rates and plan investments. To prevent discriminatory misuse, the system restricts queries related to race, gender identity, religion, political affiliation, immigration status, health, union activity, harassment complaints, or workers' personal lives. Organizations can generate up to 10 smart reports per month for free, with additional reports incurring a fee.

Key Takeaways

  • Smart reports analyze transcripts to categorize workstreams and flag 'off-topic' or personal sessions.
  • Anthropic explicitly disclaims using the tool for individual performance evaluations or employment decisions.
  • The feature supports chat, Claude Code, and Cowork but excludes the new Claude Duet agent.
  • Free usage is capped at 10 reports per month, aligning with Anthropic's broader push for enterprise monetization.

The Bottom Line

While Anthropic insists these tools are for adoption planning rather than surveillance, the ability to flag 'off-topic' sessions creates an undeniable chilling effect on workplace autonomy. Companies must tread carefully to ensure these insights improve workflow rather than erode trust.