What is Personal Finance in Claude?

It's a section of Claude, launched in beta on November 5, 2026, that lets you connect your bank accounts through Plaid and ask about your spending. It's available on Pro, Max and Ultra plans in the US and, per Anthropic, it's read-only.

Can Claude move my money with Personal Finance?

No. Per Anthropic, it "can't move, manage, or invest your money," and it isn't designed to give investment advice. The company warns that Claude can make mistakes and recommends checking important figures with your bank.

It's an Anthropic research prototype, published November 2, 2026, that aims to prove with math that a response came from a specific model. Today it covers contexts up to 1,024 tokens and costs far more to run than the model alone.

Can FlashProve tell me which model my vendor uses yet?

No. Anthropic presents it as a research prototype that today only covers contexts up to 1,024 tokens, so it isn't something your vendor can offer you today.

Can I talk to Muse by voice yet?

Per Meta, voice conversations and avatars started rolling out on November 5, 2026 in the Muse app, in the US first. For the AI glasses and Muse Charm, Meta says "soon," with no date.

Anthropic launched Personal Finance in Claude and FlashProve for model verification. Meta gave Muse a voice and avatar. AI news recap for November 2-6.

Claude can now read your bank, in the US and in beta. What happened November 2 to 6

TL;DR

Anthropic launched Personal Finance in Claude on November 5, in beta for US Pro, Max and Ultra subscribers. It reads your bank accounts read-only and, per Anthropic, can't move, manage or invest your money.

Anthropic published FlashProve on November 2, a research prototype for proving with math which model produced an answer. Today it only covers contexts up to 1,024 tokens.

Meta introduced voice conversations with Muse and expressive avatars on November 5, starting in the US.

This week's one thing: every time someone tells you about a new tool, write down what would happen to your business if it changed plan or price tomorrow.

Last week was about three models that don't know your business, and before that about what you connect and what you let an assistant decide. This week the news came from other places: a bank connection, a research lab and a voice app.

1. Claude can read your bank accounts, but it can't move your money

What happened. On November 5, Anthropic announced Personal Finance in Claude. According to its post, it starts rolling out in beta this week to Pro, Max and Ultra plans, in the US only. You connect accounts through Plaid: you sign in to your bank yourself, and Claude, Anthropic says, "doesn't ask for or see your bank password." With that data it can total spending by category, compare it with prior months, flag possible duplicate charges or fees, and build charts or bill calendars. Access is read-only: per Anthropic, Claude "can't move, manage, or invest your money." Conversations inside that section aren't used to train the models, and the post itself warns that Claude can make mistakes and tells you to check important figures with your bank. It isn't designed to give investment advice either.

2. FlashProve: Anthropic is working on proving which model answered

What happened. On November 2, Anthropic published FlashProve, an open-source prototype for what it calls "provable inference." The idea: a provider commits to its model's weights, then generates math proofs that each response came from that model, without revealing the weights. Whoever receives the answer can check it on an ordinary processor. Anthropic calls it "a research prototype" and says it's sharing "intermediate progress." The limits it admits are plain: today it only proves responses up to 1,024 tokens, which is short next to a long conversation, and in its tests producing the proof takes roughly 70 to 95 times the GPU time of running the model alone.

3. Meta gives Muse a voice and an avatar

What happened. On November 5, Meta announced voice conversations with Muse and expressive avatars inside the Muse app, on mobile, web and Mac. They're rolling out in the US "to start." You can pick voice only, the avatar beside the conversation, or full screen, choose from hundreds of voices, and interrupt Muse while it's talking. Meta says voice conversations are coming "soon" to its AI glasses and Muse Charm, with no date.

This week's one thing

This week, every time someone tells you about a new tool, write down what would happen to your business if it changed plan or price tomorrow.

One line on your phone is enough. You don't have to decide anything else. If you're already thinking about an agent, here's where to start.

Keep reading

Three AI models in the news in one week. None of them knows your business · last week's roundup
Two assistants can get into your apps. You choose which ones · what you connect and what you let them decide
Your business's first agent decides nothing: it runs your errands · where to start with an agent
Introducing Personal Finance in Claude, Anthropic, November 5, 2026 · the full announcement

Frequently asked questions

To close

This week one AI learned to read a bank and another to talk. Your Monday is still yours.

Book a free strategy call with Todd

Good Monday reading, and thanks for sticking with us.

Sources

Introducing Personal Finance in Claude, Anthropic, November 5, 2026 · https://claude.com/resources/articles/introducing-personal-finance-in-claude
FlashProve (provable inference), Anthropic, November 2, 2026 · https://www.anthropic.com/research/provable-inference
Introducing voice and expressive avatars in Muse, Meta, November 5, 2026 · https://about.fb.com/news/2026/11/introducing-muse-ai-voice-expressive-avatars/

Claude can read your bank, in the US and in beta. What happened November 2 to 6