Claude can now analyze your spending with its new Personal Finance tool

Anthropic announced on Thursday that its Claude chatbot now includes a tool called Personal Finance, which lets users link bank and credit card accounts and ask questions about their money. Connections are made through Plaid, the widely used account-linking service that many budgeting and fintech apps rely on. Anthropic says access is read-only, so the tool cannot move, manage or invest a user's funds. The feature begins rolling out in beta this week to subscribers on the Pro, Max and Ultra plans in the United States.
The design goal, as Anthropic describes it, is to turn Claude into a financial companion that works from a user's actual balances and transactions rather than offering generic budgeting advice. In a blog post the company said Claude answers in everyday language, drawing on balances and transactions, and explains unfamiliar terms. It also said the tool carries no ads or sponsored recommendations and does not sell credit cards, loans or savings accounts of its own, a positioning aimed at distinguishing it from financial products that profit from steering customers toward particular offers.
A typical use is a month-end question such as where the money went. Claude totals spending across the connected accounts and groups it into categories like rent and utilities, food and drink, and shopping. From there users can push further with follow-ups, asking whether a given month is higher than usual or whether anything has been overlooked. Because the model can hold a conversation, the same session can move from a summary into narrower questions without the user rebuilding context each time, which is the main structural advantage of a chat interface over a fixed dashboard.
Anthropic says the tool does more than summarize. It scans transactions across linked accounts to surface possible problems such as duplicate charges, unexpected fees and bills that keep creeping up. It can also handle broader questions, for example whether a person is spending more than they earn, and it can explain concepts like APR while showing how much interest the user has paid over the past year. Those calculations are the part worth watching in practice: a language model that misreads a transaction or double-counts a transfer could produce a confidently wrong number. The value of the feature will depend heavily on how reliably it aggregates accounts, which is a data-plumbing question as much as a model one.
On privacy, Anthropic has set explicit bounds. Claude's access is limited to each connected bank's name, the account name and type, the last four digits of the account number, balances, credit limits, and up to two years of transaction history. The company also says conversations held inside the Personal Finance hub are not used to train its models. Those limits matter because the sensitivity of the data is unusually high: a transaction history can reveal location, health, relationships and employment. Limiting scope and excluding those chats from training are sensible defaults, though users will still be trusting a third-party aggregator and a model provider with a detailed picture of their finances.
This is not the first move in the category. OpenAI's ChatGPT introduced comparable personal finance tools in May, also letting users connect accounts and ask about spending patterns and longer-term planning, so Claude is arriving second. Anthropic has nonetheless been pushing Claude into vertical markets, having launched Claude for Legal and Claude for Financial Services earlier this year. Personal Finance is a consumer-facing counterpart to those enterprise offerings, and it fits a broader pattern of chatbots evolving from answer engines into interfaces layered over personal data.
A likely trade-off is depth versus trust. A read-only assistant cannot help a user act, such as moving money or setting up an automated transfer, which caps how useful it can be as a full financial manager. But read-only also avoids the regulatory and liability weight of executing transactions, and it lowers the bar for a beta launch. Expect the competitive ground to be accuracy of categorization, quality of anomaly detection and clarity about data handling rather than raw model capability, since the underlying language models are broadly comparable.
For developers and product teams, the launch is a signal about where assistant platforms are heading: connectors to sensitive third-party data, narrow permission scopes, and clear statements about training exclusions are becoming table stakes for consumer trust. Teams building budgeting, expense or PFM products should assume users will increasingly ask a general assistant these questions first, which raises the bar on integrating with or differentiating from that experience. It is also an early test of whether consumers will accept an AI chatbot as a place to keep a live view of their bank accounts, and how the answer shapes expectations for every other category that involves personal data.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
With Personal Finance in Claude, users can connect their bank and card accounts through Plaid and ask questions about their finances.