Perk MCP is Perk's Model Context Protocol server. It lets you connect your AI assistant to Perk and ask questions about your company's travel, spend, travel invoice, and event data in natural language. Use this article to understand what Perk MCP is, what you can do with it, and how your access works.
In short: you connect your AI assistant to Perk once, then ask it questions about trips, expenses, travel invoices, reporting, policy, events, and cards, and it reads the answers from Perk for you. Instead of logging into Perk and running a report, you get answers inside the assistant you already use — up to 4x faster than running a report manually — removing the manual admin, the work behind the work, that sits between you and a straightforward answer. Perk MCP is for every Perk user — travelers, finance teams, and travel managers alike, not just admins.
Perk MCP is available to all customers on all plans. What you see depends on your existing Perk role and your company’s active plan.
About Model Context Protocol (MCP)
The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data sources, a bit like a plugin system. Once an assistant is connected to a platform through MCP, it can read data from that platform and use it to answer your questions.
Perk MCP is Perk's implementation of this standard. It gives your assistant a secure connection to your Perk data so you can ask about travel, spend, invoices, and events directly in the assistant you already use.
AI assistants Perk MCP works with
Perk MCP works with all major AI assistants. This includes Claude and ChatGPT, plus other assistants and developer tools such as Gemini, Cursor, and VS Code. You connect your assistant using a single server URL and your existing Perk credentials.
To connect your assistant, see Connect your AI assistant to Perk.
What you can do with Perk MCP
Perk MCP lets you query a range of travel and spend data in Perk without opening the app. At a glance, you can:
- Look up trips, including travelers, bookings, dates, and cost.
- Search and filter expenses by date, status, merchant, amount, or trip.
- Find, open, and download travel invoices and drill into invoice line items.
- Generate travel reports and query spend data such as top cost objects and categories.
- Check travel and spend policy rules, such as hotel budgets and cabin class restrictions.
- Look up user profiles and, for admins, search all users in the account.
- List your Perk Events — upcoming, past, or by your role (organizing, attending, or invited) — and, for events you organize, see who's attending.
- See your company cards and the pending card transactions you still need to match to an expense.
You can't book, change, or cancel trips, create, edit, or send expenses, or approve anything through your assistant. Write actions like booking a trip, creating and sending expenses, approvals and more are coming soon in a future release.
For the full list of capabilities and example prompts by role, see What you can do with Perk MCP.
How access and security work
Perk MCP acts within your existing Perk permissions. You authenticate with your existing Perk credentials, and the assistant can't access any data you couldn't already see in Perk yourself.
Your access mirrors your Perk role and your company's active plan. For example, a traveler sees only their own trips, an admin can search across the organization, expense insights need expense management, and reporting is limited to users with reporting access. Because of this, two people can ask the same assistant the same question and see different results, based on what each is allowed to see in Perk.
How Perk MCP differs from the Perk REST APIs
Perk MCP and the Perk REST APIs both reach your Perk data, but they're for different people. The Perk REST APIs need developers to build integrations in code. Perk MCP lets non-technical users query Perk in natural language from an AI assistant they already use, with no code.
Where to go next
To connect your assistant and ask your first question, see Connect your AI assistant to Perk and Ask Perk your first question.
Tip: If you want more technical details about the Perk MCP, see Perk Developers.