Intel / article

Analyze your Stripe data directly with your agent

Source: stripe.com
Date: Published Nov 5, 2026

Context

Every Stripe user can now connect an such as Anthropic's Claude or Cursor to their Stripe analytics data and ask questions in plain language, either through Stripe's server or by calling its Queries and Reports APIs directly (Stripe). Stripe's examples include reconciling each payout against the payments, refunds, disputes and fees behind it, building a monthly recurring revenue dashboard, and finding the payment segments with the highest chargeback rates.

The harder part, as Stripe frames it, is that an agent has to know which tables to use and how Stripe defines its metrics. The Reports API carries a modeled schema of more than 300 tables plus built-in guidance on how an agent should use them, which Stripe says yields more accurate analysis. Because these results can feed business decisions and investor reports, each analysis exposes the SQL query behind it, so a person can confirm how a number was calculated and edit the query. The agent can also combine Stripe data with other connected tools, such as website analytics from PostHog in Claude.

The Queries API is free until January 31, 2027, after which it is billed per credit of usage with no monthly subscription; Stripe has not given the credit price in this post. The Reports API stays free.

Key takeaways · AI-distilled

Stripe's Queries API is free for all users through January 31, 2027, after which it is billed per credit of usage with no monthly subscription; the Reports API stays free.
Stripe says the Reports API exposes a modeled schema of more than 300 tables plus built-in guidance on how the agent should use them, and can route the agent to attribute MRR changes to upgrades, downgrades and churn.
For verifiability, each agent analysis run through these APIs exposes the underlying SQL, so users can check how a number was calculated and edit the query.
Because the agent can combine Stripe data with other connectors, Stripe gives the example of joining PostHog website activity with subscription data in Claude to find customers who clicked a button but later churned.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.

Read the source stripe.com