Skip to main content

Morning Overview

Claude is testing a feature that plugs an AI assistant straight into your bank accounts

A new “Money” tab has appeared in the Claude mobile app’s navigation bar, sitting alongside Chats, Code, Artifacts, Dispatch and Cowork. Anthropic has not announced it, but the interface itself was spotted first by Alexey Shabanov at TestingCatalog on September 14, then confirmed independently two days later by BleepingComputer’s Mayank Parmar. Neither the feature nor its underlying data provider has been made public by Anthropic.

The unreleased screen carries its own pitch, inviting a user to “understand” personal finances with Claude and to link a bank account so Claude can be asked about spending, plans and more. A “Get started” button sits underneath, prompting a user to connect an account, though neither reporter who tested the app could get the flow to complete — the feature is visible but not yet functional for most people who have it.

The onboarding screen promises more than a chat window

TestingCatalog’s reporting frames Money as an attempt to give Claude standing access to a person’s financial picture rather than requiring a manually uploaded bank statement or spreadsheet export for every question. Shabanov’s writeup lists the likely uses as analyzing spending, flagging recurring payments, summarizing balances, charting trends and answering budgeting questions, eventually extending to helping plan financial decisions. The exact account types the feature will support, which data provider moves the transactions behind the scenes, and which specific actions Claude will be allowed to take all remain unconfirmed.

No equivalent Money tab has surfaced yet on Claude’s desktop or web interface, only on the iOS build, which is itself one signal the feature is still in early internal testing rather than close to a public release. Both outlets that reported on it noted development had progressed far enough to produce a dedicated navigation entry and a full onboarding screen, which is further along than a stray mention buried in app code, but neither treated that as evidence of an imminent launch date.

OpenAI already ships a version of this through Plaid

The closest existing comparison is ChatGPT’s Finances feature, which OpenAI has already made available to eligible Plus and Pro subscribers in the United States on web, iOS and Android. It connects bank accounts, credit cards and brokerages through the financial data platform Plaid and answers questions about spending, bills, subscriptions, net worth and investments. OpenAI has said it supports more than 12,000 financial institutions in the United States and does not train its models on the personal financial data it collects through the feature.

That timeline matters for how the Claude Money leak should be read. OpenAI’s Finances feature had already been live long enough to draw an explicit comparison from reporters covering Anthropic’s interface, meaning Claude would be following an established consumer pattern rather than pioneering the idea of a chatbot with standing access to a checking account. The novelty in the Anthropic case is not the concept — it is which company is building it next, and how closely the execution tracks what OpenAI’s subscribers already use.

Money would make Claude the second major chatbot to reach directly into a user’s bank accounts rather than waiting for uploaded statements, and reporters covering both companies expect Anthropic to lean on a similar account-linking partner rather than building bank connections from scratch. That would put Claude in competition with a feature OpenAI has already normalized among its paying subscriber base, rather than introducing the idea of chatbot-linked banking to the market first.

The move fits a broader push into finance, not a one-off feature

Money would not be Anthropic’s first step into financial services. In May, the company released ten ready-made agent templates aimed at financial professionals, covering tasks such as pitchbook building, earnings-transcript review, month-end closing and KYC compliance screening, deployed through Claude Cowork, Claude Code and Claude’s managed-agent platform. Those templates target analysts and back-office teams inside banks and investment firms; Money, by contrast, would be the first Anthropic product built for an individual managing a personal checking account rather than a firm managing client portfolios.

Claude already integrates with a handful of third-party personal-finance tools, including Era Context and PocketSmith, according to TestingCatalog’s reporting, so a native bank-linking feature would consolidate functionality that currently requires a separate connector rather than introduce an entirely new category of use to the assistant.

What still has to be answered before this ships

Both outlets that broke the story flagged the same unresolved questions: which banks Money will support, exactly how the connection will work, and whether the feature launches everywhere Claude operates or only in the United States. An initial United States-only rollout is the likelier path, since bank-linking services typically require country-specific data providers and compliance work before they can operate legally, and expansion into the European Union would need additional regulatory approval on top of that.

Until Anthropic makes an official announcement, the “Get started” button in the Claude app remains the only public evidence the feature exists at all — a screen a small number of testers can see, without a confirmed timeline for when, or whether, everyone else will see it too.

The gap between the two products is notable on its own terms. Anthropic has spent much of 2026 selling Claude to banks, hedge funds and insurers as a tool for institutional finance work, where the agent templates operate on a firm’s own ledgers under compliance review built for professional users. A native bank-linking feature for individual consumers is a different kind of product, tied to one person’s own transaction history rather than a firm’s books, which is one plausible reason the interface reached testers before Anthropic was ready to confirm or explain it publicly.


More from Morning Overview

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.