Anthropic launched Personal Finance in Claude on Wednesday, a new beta that connects a user's bank and credit card accounts directly to the chatbot. The feature starts rolling out this week to Pro, Max and Ultra subscribers in the United States, and positions Claude to answer money questions using actual transaction records rather than generic advice.
The connection runs through Plaid, the same data pipeline many banking apps use. Users sign in to their bank themselves and approve Claude's access during setup; Anthropic says the assistant never sees the banking password. Once linked, Claude can read the bank's name, each account's name and type, the last four digits of the account number, balances, credit limits and up to two years of transaction history.
The boundary is deliberately narrow: access is read-only. Claude cannot transfer funds, pay bills, manage accounts or make investment decisions, and Anthropic warns that its answers can still contain mistakes. The company suggests users verify important figures with their bank before acting on them.
In practice, the feature is built around questions like "where did my money go this month?" Claude sorts spending across linked accounts into categories such as rent and utilities, dining and shopping, compares the totals with previous months, and can surface duplicate charges, unexpected fees or bills that have quietly increased. Users can also ask for a weekly spending chart, a calendar of upcoming payments or a plain-English explanation of terms like APR, complete with a calculation of the interest they paid over the past year.
Anthropic says conversations inside the Personal Finance hub are not used to train its models, and users can unlink an account at any time. The company also stresses that the feature carries no advertising or sponsored recommendations — no credit card pitches, no loan offers — a positioning aimed at keeping the tool neutral.
The launch follows OpenAI, which introduced a similar account-connected finance feature in ChatGPT in May 2026. Anthropic is effectively conceding that conversational finance is now a category worth fighting for, and the read-only design is the safer opening move: regulators treat anything that moves money very differently from anything that merely reads it.
It is also the latest step in Anthropic's push into vertical products, after Claude for Legal and Claude for Financial Services. The company says it is exploring integrations with licensed professionals, banks and wealth managers, which would move Claude from reading statements to sitting inside regulated financial workflows.
For now the beta is limited to paying US subscribers, and Anthropic has not said when it will expand to other countries or to free users. The competitive logic, however, is already clear: the assistant that can ground its answers in two years of your real transactions has a very different answer quality than one guessing from a screenshot.
Comments (0)
Log in to join the discussion
Log InNo comments yet