Anthropic launches AI report card tool to track corporate costs and usage

A new lens on AI spending

Anthropic recently launched a new feature called 'smart reports' specifically for its Claude Enterprise customers. This tool is designed to help organizations understand how their teams are actually using the AI chatbot in their daily workflows. By providing a clearer picture of usage, the company hopes to help businesses manage the costs associated with large-scale AI adoption.

The tool functions as a sort of AI report card for corporate environments. It allows managers to see which types of tasks are being automated and what specific outputs the AI is generating for the company. This level of detail is becoming increasingly important as firms try to calculate the return on investment for their software subscriptions.

As Business Insider reported, the launch comes at a time when many companies are struggling to measure productivity gains from AI. Organizations are often caught between wanting to encourage innovation and needing to control the costs of high-volume AI interactions.

How the reports are generated

To create a report, administrators can select a specific team and a time range covering up to the last 28 days. They can also choose which Claude products to include in the analysis and define specific focus areas. Users can even submit custom questions for the AI to answer while it processes the data.

Once the parameters are set, the system reads a sample of the team's transcripts. It then groups these interactions into different workstreams and identifies the types of outputs produced. Finally, the tool attaches the specific spend to each of these groups to show where the money is going.

The process is automated to handle large amounts of data quickly. By grouping work into categories, it moves beyond simple usage counts to provide a more nuanced view of how the technology is being integrated into various departments.

Flagging inefficiencies and risks

One of the most notable features of the smart reports is the ability to flag 'inefficiencies' to employers. The tool can identify sessions that are deemed to have produced no work value or those that appear to be personal or off-topic. This provides a direct way for companies to monitor whether the tool is being used for its intended professional purpose.

Anthropic has included specific safeguards regarding sensitive information. Sessions involving restricted topics—such as health, legal issues, union activity, or religious views—are only used for aggregate analysis. These types of conversations do not include individual sessions or names in the final summaries.

The company emphasizes that these reports are not intended for individual performance reviews. Anthropic stated, "Smart reports help you understand adoption and plan your investment in Claude," according to the company's support documentation. They explicitly advise against using the data to make specific employment decisions.

The corporate AI landscape

The release of these reports highlights a growing trend of corporate oversight in the AI space. Many companies are currently wrestling with how broadly to allow their staff to use these tools. Some have already begun to restrict access based on specific roles or productivity concerns.

For example, reports indicate that some major firms have limited access to coding tools to engineers only. These moves often come after internal assessments of risk and actual productivity gains. Such restrictions can sometimes cause friction when other departments, like traders, find the tools useful for their own tasks.

By providing a tool that identifies 'off-topic' use, Anthropic is leaning into this trend of corporate governance. It gives companies the data they need to set boundaries while still allowing for broad team adoption of the technology.

What comes next for teams

For companies already using Claude Enterprise, the next step will likely involve regular audits of these reports. Managers will need to decide how to interpret the 'inefficiency' flags without violating the company's own stance on employee privacy. The goal is to find a balance between oversight and trust.

As more companies join the platform, the data gathered from these reports may help Anthropic refine its enterprise offerings. They may eventually offer more granular categories or deeper insights into specific workstreams as they see how different industries utilize the tool.

For the average employee, this means a higher level of visibility into their AI interactions. While the tool is meant for high-level planning, it marks a shift toward a more measurable and scrutinized era of workplace AI usage.