Claude opens up personal finances and courts the banks: AI wants to be the advisor, with a professional behind it

By PYMNTS · November 5, 2026.

Anthropic has introduced Personal Finance, a Claude feature that will let users connect bank accounts and cards and ask about their own money in natural language. The company says it is rolling out gradually this week to Pro, Max and Ultra plans in the United States.

Its scope is limited by design. According to Anthropic's announcement, the feature is read-only: it cannot move, manage or invest money. It also includes no ads or sponsored recommendations, and the company does not offer its own cards, loans or savings accounts. Nor is it meant to give investment advice.

What it does do is answer using the user's actual balances and transactions. Anthropic cites questions such as “Where did my money go this month?” and “What is an APR (annual percentage rate) and how much has mine cost me this year?” It also offers a weekly spending chart, a calendar of due dates and a dashboard of figures chosen by the user.

Connections are made through Plaid, an intermediary that links apps to bank accounts. According to Anthropic, they are established only when the user logs in to their bank through Plaid, and they can be unlinked at any time. Will Robinson, Plaid's chief technology officer, says most consumers share personal data with Claude “when they see a clear benefit.” That is Plaid's claim, not independent data.

This is not uncharted territory. According to PYMNTS, OpenAI announced in May that it had begun letting select ChatGPT users connect their financial accounts, also through Plaid, and receive answers grounded in their financial context.

**Our take.** The spending dashboard is the least interesting part of the announcement. What matters is Anthropic's closing line: for bigger financial decisions, it says, people are better served by Claude working alongside licensed professionals and regulated firms, which is why it is inviting banks, wealth managers and financial companies to help shape it.

It is a sensible strategic position. In finance, the risk is not that the model cannot explain an APR, but that users hand off high-stakes decisions to a fluent but wrong answer. Putting a regulated professional in the loop spreads responsibility and fits what we already see in other fields: the value lies in orchestrating AI with human judgment, not in replacing it. It also helps banks, which gain a distribution channel instead of losing the customer relationship. It remains to be seen whether institutions will agree to be partners or prefer to build their own.

There is also a warning. Read-only access is a good safeguard, but transaction data is among the most sensitive data there is, and trust will depend on how retention, use and deletion are handled. We will hold those policies to the same standard we apply to any agent: what the company promises versus what can be audited.

In the short term, this mainly changes everyday financial literacy: people who never read their statements will be able to understand them. In the long term, a competent advisor available to everyone, not just those who can pay for one, is a concrete part of the abundance we advocate. Getting there means doing it with clear limits and real oversight.