Giving admins more visibility and control over Claude spend

· Source: Claude Blog · Field: Business & Management — Operations & Process Management, Project & Product Management, Corporate Strategy & Leadership · Depth: Intermediate, short

Summary

Anthropic has introduced enhanced analytics and cost controls for Claude Enterprise, addressing the complex usage and cost patterns of agentic AI work. New features include a richer admin analytics dashboard displaying usage and cost by group and user, alongside output like artifacts created and skills used. Claude Code specifically gains insights into active developers, session counts, top commands, and estimated productivity lift. Admins can leverage Analytics Chat for plain-language queries and access usage data programmatically via the Analytics API, integrating with tools like Datadog Cloud Cost Management. Additionally, model-level entitlements allow setting default models and restricting access, while spend-threshold alerts notify admins and users at 75%, 90%, and 95% of limits, preventing unexpected cutoffs. The Admin API further enables automated cost-control workflows.

Key takeaway

For IT professionals or finance managers overseeing AI expenditures, understanding and controlling Claude Enterprise costs is now more robust. You should utilize the new analytics dashboard to track usage by group and user, configure model entitlements to prevent overspending on routine tasks, and set up spend-threshold alerts to proactively manage budgets. Integrating the Analytics API with your existing cloud cost management tools will provide a unified view of AI spend, enabling better ROI analysis and preventing unexpected service interruptions.

Key insights

Granular visibility and control over AI spend are crucial for managing complex agentic AI usage.

Principles

Method

Admins can configure model defaults and access entitlements, set organizational spend limits, and receive automated alerts at 75% and 90% thresholds. They can also use the Analytics API to integrate Claude usage data into existing finance and IT systems.

In practice

Topics

Best for: CTO, VP of Engineering/Data, Director of AI/ML, IT Professional, Executive

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Claude Blog.