The most interesting thing about Claude’s new money feature is what it refuses to do. It can’t move funds, can’t invest, and won’t pitch you a credit card. Anthropic announced Personal Finance in Claude this week, and the pitch is simple: connect your accounts, ask questions in plain English, get straight answers.
It’s a beta, limited to US users on the Pro, Max, and Ultra plans. Free users are out for now.
What it actually does
You link bank and card accounts, then ask things like “Where did my money go this month?” Claude totals spending across accounts and sorts it by category. In Anthropic’s example, that comes to $4,309.37, with $1,024.51 going to food and drink. Follow-ups work too. Ask whether that’s more than usual and Claude checks earlier months, flags a membership renewal, then lists four payments hitting checking before payday and what you’ll have left.
It also hunts for oddities across accounts: duplicate charges, surprise fees, a bill that has crept up all year. And it explains jargon. Ask what an APR is and Claude defines it, then adds up the interest you’ve paid this year. You can request a weekly spending chart, a calendar of due dates, or a custom dashboard and keep tweaking it.
Privacy and limits
Access runs through Plaid. You sign in to your bank yourself, approve read-only access, and Claude never sees your password. For the accounts you pick, Claude sees the bank name, account name, type, last four digits, balances, credit limits, and up to two years of transactions.
Conversations inside the Personal Finance hub aren’t used to train models. Outside the hub, your normal privacy settings apply. You can unlink an account anytime. Anthropic also admits Claude can get numbers wrong, and tells users to ask which transactions sit behind a figure and verify anything important with the bank.
How it stacks up
This isn’t a new idea. Mint, which Intuit shut down, trained a generation on automatic categorization, and apps like Monarch, Copilot, and YNAB still own the budgeting niche. Banks have also bolted assistants onto their apps, like Bank of America’s Erica. What those tools share is an incentive to steer you toward something: a card, a loan, a premium tier.
Anthropic is betting that a no-ads, no-products stance is the differentiator. That’s smart positioning, though trust will be earned slowly. The real advantage is flexibility. A chat model handles odd questions that fixed dashboards can’t, and it builds the dashboard you want on demand. The weakness is accuracy. Budgeting apps use deterministic math. An LLM summing transactions can slip, and in finance a small slip erodes confidence fast.
OpenAI and Google will watch this closely. Both have the models and the user bases to ship something similar, and neither is as clean on the ads question.
Why it matters
Personal finance is one of the most obvious everyday uses for a chatbot with real data access, and it’s also one of the riskiest. Anthropic is starting with the safe slice: understanding, not advice. The closing note is telling. The company says bigger decisions are best handled with licensed professionals and regulated firms, and it’s inviting banks, wealth managers, and financial services companies to help shape that. Read that as a partnership play, and probably a future route into regulated advice.
For now, this is a useful beta with a sensible scope. If you already pay for Claude, connecting one account and asking a single question costs you nothing. If you’re a budgeting-app founder, the clock just started.
