Expensify Connects Its Expense Data to Claude, ChatGPT, and Cursor
"As software shifts from menus and forms toward conversational interfaces, businesses will expect every product to work seamlessly with AI agents."
Expensify announced on June 8, 2026 that its expense data is now accessible through the Model Context Protocol, connecting the platform to ChatGPT, Claude, Cursor, and other MCP-compatible AI clients through a secure OAuth 2.1 integration.
This replaces the manual export and integration workflows that currently sit between expense management systems and the AI tools finance and operations teams use daily, according to the press release.
Finance teams and managers who want to query expense data, to find missing receipts, review approval queues, or summarize team spending by category, currently have to export data from Expensify, format it, and bring it into whatever tool they are using for analysis.
The MCP eliminates that friction entirely, according to the company. The AI client connects directly to the live Expensify account and returns real-time answers to natural language questions.
"As software shifts from menus and forms toward conversational interfaces, businesses will expect every product to work seamlessly with AI agents," said David Barrett, founder and CEO of Expensify. "MCP is quickly becoming the standard for how AI systems interact with external services, and we want Expensify to be accessible wherever our customers are working."
The MCP handles a broad range of expense queries that previously required navigating Expensify's interface directly, according to the press release.
Users can search for specific expense categories, "Show me all my meal expenses over $20" or filter by vendor and date, "Find Marriott receipts from the last 90 days."
Approval workflows are accessible through natural language. "Which expense reports need my approval?" returns a live queue rather than requiring the user to open Expensify separately.
The compliance use cases are the most operationally significant. "Find all expenses over $75 that are missing receipts" surfaces a compliance gap that would otherwise require a manual audit of the expense ledger.
"Summarize my team's spend by category for the last quarter" produces a summary that previously required a data export and a pivot table. Both are now a single sentence inside Claude, ChatGPT, or Cursor.
Client-level spend tracking is also supported. "How much did my client spend on Uber last month?" gives account managers and consultants immediate visibility into client expense patterns without leaving their AI client.
The Expensify MCP launch is part of the broader enterprise shift toward MCP as the connectivity layer between AI agents and the business systems they need to act on.
DocuSign launched its MCP connector in May 2026 and described the inbound interest as generating thousands of beta signups, "an unprecedented level of engagement."
Mastercard also launched its Agent Suite for merchant payments last month, using MCP as the integration architecture.
Expensify's launch extends that pattern into expense management, one of the most data-rich and most manually intensive workflows in enterprise finance operations.
The MCP is available now to any Expensify user with a validated account. Expensify serves 15 million members worldwide.
Key Takeaways
- Expensify integrates expense data with AI tools like ChatGPT and Claude through Model Context Protocol.
- The new integration eliminates manual export processes for finance teams, streamlining data access.
- Users can query real-time expense data using natural language, enhancing efficiency in financial management.
- Expensify aims to align with the shift towards conversational interfaces in software applications.
- MCP is becoming the standard for AI interaction with external services, boosting accessibility for users.