Savi MCP

Bring your money to the conversation

Savi MCP connects your financial data to the AI tools you already use, helping turn everyday money questions into clear visual answers and practical next steps.

Accounts, spending, and subscriptions connect through Savi MCP to an AI Assistant.
Accounts
Spending
Subscriptions
Secure contextSavi MCP
Clear answersAI Assistant
Savi Household Spending Flow displayed in an AI conversation on laptop and mobile screens

Ask where spending changed, understand what drove the change, and get recommendations directly inside the conversation.

Connected capabilities

More than a spending summary

Savi MCP exposes financial tools to an authorized AI client. The assistant can use the relevant account context and supported actions to answer a question or help update a record.

  • Understand accounts and spending

    Review accounts, balances, transactions, alerts, categories, and spending analytics in response to a money question.

  • Plan upcoming changes

    Work with savings goals, financial plans, and income or expense projections as your priorities change.

  • Keep records organized

    Create and manage categories, tags, recurring rules, and saved filters to make financial records easier to review.

  • Coordinate shared work

    Manage shared accounts and work with supported invoice and time-tracking records from the same assistant connection.

Review access before connecting

Savi uses OAuth sign-in for MCP clients. You can approve read-only access, and a client that asks to change data needs a separate opt-in for write access. Check the requested access on Savi's consent screen before approving it.

Read the authorization guide

How Savi MCP works

From a money question to a useful next step

Savi MCP brings the right financial context into your AI-assisted workflow, helping you move from raw data to clear explanations.

  1. Step 1

    Ask a financial question

    Start with an everyday question about your household spending, savings, or subscriptions.

  2. Step 2

    Savi MCP retrieves context

    Savi MCP securely brings the relevant financial context into the conversation.

  3. Step 3

    AI analyzes the context

    Your AI assistant uses that context to identify changes, patterns, and useful insights.

  4. Step 4

    Get a clear next step

    Receive an understandable explanation and a practical action you can take next.

Illustrative example

Savi data, ready for AI questions

Secure financial context

User asks

“Which subscriptions should I review before next month?”

$220

Monthly subscriptions

7

Recurring services

2

Possible overlaps

Savi response

Review streaming, storage, and productivity services first. Two tools appear to serve similar use cases, and canceling one could reduce recurring spend before the next billing cycle.

What the connection includes

Your financial data stays attached to your Savi account

MCP gives a compatible assistant a structured way to request information from Savi. The assistant does not receive a separate copy of your bank login; it uses the access you approve for your Savi account.

Ask about real records

Ask an assistant to summarize accounts, transactions, spending, recurring services, or goals that are available to your Savi account. Give it a time period or category when you want a focused answer.

Choose the access level

Sign in through Savi and review the requested access before approving it. You can untick “Allow changes” to connect with read-only access. The MCP guide explains the consent screen and the first read-only account check.

Keep the result grounded

An assistant interprets the records returned by Savi. Check important totals against the underlying transactions, and remember that a missing account or delayed bank feed can affect what the assistant sees.

Get better answers with a clearer prompt

A useful request says what to compare and what kind of explanation you want. For example, ask which spending categories changed between two months, request the transactions behind the largest change, or ask how a savings goal is progressing. If the first answer is too broad, narrow the date range or ask the assistant to show the transactions that support its summary.

MCP is the connection protocol; it does not make an assistant a financial adviser. Use its summaries to explore your own records, then verify calculations and decide what action fits your circumstances.

Common questions about Savi MCP

Do I need to create an API token?

No. An MCP client uses Savi’s OAuth sign-in flow. You approve access in your browser; you do not copy an API token into the assistant.

Can I connect with read-only access?

Yes. On the consent screen, untick “Allow changes” before you approve the connection. The developer guide shows how to verify the connection with read-only calls.

Built for financial clarity

Ask better questions about your money

Connect Savi to AI-powered conversations and turn financial context into answers that are easier to understand, act on, and trust.

Claude and ChatGPT open example prompts; connect Savi in those clients first. Cursor opens a prefilled Savi MCP install prompt. Muse support is coming soon.

Savi MCP questions

What can an AI assistant do with Savi MCP?

Depending on the access you grant, an assistant can read accounts, transactions, goals, and analytics, and use tools to create or update supported financial records. The developer guide lists each available tool and its required scope.

Can an assistant change my Savi data?

Some MCP tools can create, update, or delete records. During authorization, you can grant read-only access; write access requires a separate opt-in when the client requests it.

Do the Claude, ChatGPT, and Cursor buttons connect Savi MCP?

Claude and ChatGPT open an example question. Cursor opens a prefilled Savi MCP install prompt. Sign in through Savi to authorize access.