Anthropic Launches Tools to Show Companies Claude's Cost and Return

Anthropic has launched a set of reporting and analytics tools that let companies measure how much they spend on Claude and weigh that cost against the work their teams get done. The aim is to help organizations judge the return on their AI investments instead of treating raw usage as a sign of success.

The tools let managers review team activity over defined periods and see which tasks were completed, what they cost and where workflows stalled. The analysis draws on internal analytics and employee surveys, and it comes with recommendations for improving results and cutting wasted spending. By combining usage, cost and work output, the tools show how the resources consumed by AI relate to what teams deliver.

Anthropic also added shared budgets for groups. Teams can use them to manage spending on tokens, the text units used to calculate the cost of running models, rather than letting expenses grow without a single view of consumption. The company says it does not use data from these reports to train its models by default, a point that matters to firms handling information about their operations and staff.

Spending on coding agents is climbing

The launch comes as enterprise spending on AI coding tools rises quickly. A study by Jellyfish, covering agentic coding tool use among 276,000 of its customers, found that spending on these tools has grown tenfold since the start of the year. It now equals about 2% of total engineering costs at the organizations covered.

The same study found that Claude Code and other tools built on Anthropic's models accounted for roughly a third of the code changes proposed for review.

The new tools target a problem many companies share: it is hard to tell whether heavier AI use actually improves productivity, especially as running costs rise and results have to be set against spending.