Anthropic introduces AI usage analysis tool for Enterprise clients

The company unveiled “smart reports” for Claude Enterprise users last week. This interactive tool analyzes transcripts from a sample of a team’s AI sessions, categorizing tasks, outputs, and costs.

The reports can identify areas of inefficiency, including sessions deemed “personal or off-topic” or producing “no work value,” according to Anthropic’s support documentation.

Anthropic emphasizes that these reports are intended for understanding AI adoption and investment planning, not for individual performance evaluations or employment decisions.

To generate a report, managers can specify a team, a timeframe of up to 28 days, relevant Claude products, analysis focuses, and custom questions for Claude to address during transcript analysis.

Claude then analyzes a sample of transcripts, categorizes them, assigns costs, and summarizes findings.

Sessions involving sensitive topics like race, ethnicity, health, legal matters, union activity, and political or religious views are used only for aggregated analysis, excluding individual session details and names.

The launch of smart reports aligns with broader efforts by companies to control AI costs and measure productivity gains. Goldman Sachs recently restricted access to certain AI tools, including Anthropic’s Claude Code, to engineers only, highlighting the challenges of managing AI adoption.