# Use BetterMenu with an AI assistant



You want your AI assistant to help you build a recipe, search for the right ingredients, and calculate a nutrition label — without switching between your AI tool and BetterMenu. BetterMenu connects directly to Claude, Claude Code, and other compatible AI assistants, giving them access to your full recipe library and every BetterMenu tool.

How does the connection work? [#how-does-the-connection-work]

BetterMenu provides a server that speaks the Model Context Protocol (MCP) — a standard way for AI assistants to call external tools directly. Once your AI tool is connected, it can use BetterMenu tools as naturally as it uses any other capability: searching the USDA ingredient database, creating and updating [recipes](/docs/guide/concepts/recipe), computing nutrition facts, configuring serving sizes, and exporting labels.

<Mermaid
  chart="`
%%{init: {&#x22;theme&#x22;: &#x22;base&#x22;, &#x22;themeVariables&#x22;: {&#x22;primaryColor&#x22;: &#x22;#ede9ff&#x22;, &#x22;primaryTextColor&#x22;: &#x22;#1e1b4b&#x22;, &#x22;primaryBorderColor&#x22;: &#x22;#7c3aed&#x22;, &#x22;lineColor&#x22;: &#x22;#7c3aed&#x22;, &#x22;secondaryColor&#x22;: &#x22;#ede9ff&#x22;, &#x22;tertiaryColor&#x22;: &#x22;#faf9f7&#x22;, &#x22;background&#x22;: &#x22;#ffffff&#x22;, &#x22;edgeLabelBackground&#x22;: &#x22;#ede9ff&#x22;, &#x22;fontFamily&#x22;: &#x22;Geist, system-ui, sans-serif&#x22;}}}%%
flowchart LR
  A[&#x22;Your AI assistant\n(Claude, Copilot, etc.)&#x22;] -->|&#x22;BetterMenu MCP&#x22;| B[&#x22;BetterMenu&#x22;]
  B --> C[&#x22;Your recipes&#x22;]
  B --> D[&#x22;USDA ingredients&#x22;]
  B --> E[&#x22;Nutrition labels&#x22;]
`"
/>

The AI assistant acts on your behalf, using the same permissions you have in BetterMenu. It can read and create recipes within your organization, but it cannot access data belonging to other organizations. An R\&D team using Claude to prototype new granola bar formulas, for example, sees only the recipes that team already has access to in BetterMenu Studio — the connection does not grant broader access. The result is that your AI assistant becomes a hands-free interface to BetterMenu: you describe what you need, and it handles each step for you.

How do I set up the connection? [#how-do-i-set-up-the-connection]

Setting up the BetterMenu MCP server takes three steps:

1. Open your AI tool's configuration settings and add the BetterMenu MCP server. The server address and any required fields are in the [API reference](/docs/api).
2. When your AI tool connects for the first time, BetterMenu will prompt you to sign in with your BetterMenu account to authorize the connection.
3. Once authorized, your AI assistant can discover all available BetterMenu tools automatically — no manual configuration of individual tools is needed.

The authorization step uses your existing BetterMenu account login. You do not need a separate API key or developer credential — the same account you use in BetterMenu Studio is all that is required. If your organization uses role-based access, each team member who wants to use AI assistant integration must authorize the connection with their own account. This ensures that every action taken through the AI assistant is tied to the correct user and appears correctly in the [audit trail](/docs/guide/concepts/audit-trail).

What can my AI assistant do? [#what-can-my-ai-assistant-do]

A connected AI assistant can run the full BetterMenu recipe workflow without any manual steps. You describe what you want in plain language, and the assistant calls the appropriate BetterMenu tools in the right order.

The table below maps common plain-language requests to the BetterMenu action the AI assistant performs behind the scenes. Each row is a complete operation — the assistant handles the sequencing, so you do not need to know which tool to call or in what order.

| What you can ask                                    | What BetterMenu does                                 |
| --------------------------------------------------- | ---------------------------------------------------- |
| "Add a recipe for my granola bar"                   | Creates a new recipe in your organization            |
| "Find the USDA entry for rolled oats"               | Searches the ingredient database and returns matches |
| "Calculate the nutrition for this recipe"           | Computes FDA-compliant nutrition facts               |
| "What serving size should I use for a 38g bar?"     | Looks up the applicable FDA reference amount         |
| "Export the label as a PDF"                         | Generates and returns the label file                 |
| "What changed between the March and June versions?" | Retrieves the version comparison                     |

Your AI assistant can also query your recipe history, read the audit trail, and retrieve past label exports — all through the same connection, using the same plain-language approach shown above. For nutrition analysis workflows, an AI agent can loop through multiple recipes automatically, comparing formulas and flagging differences without manual effort from your R\&D team. The result is faster iteration: your team describes what they need, and BetterMenu acts on it directly.

What access does the AI assistant have? [#what-access-does-the-ai-assistant-have]

The AI assistant uses your BetterMenu account's permissions. It can access every recipe and ingredient you have access to within your organization, and it respects the same role restrictions as the BetterMenu Studio interface. If you have Viewer access to a recipe, the AI assistant also has Viewer access and cannot make changes to it. Editor and Admin roles work the same way — the connection does not elevate permissions.

All actions taken by the AI assistant are recorded in the [audit trail](/docs/guide/concepts/audit-trail) under your account, with a note indicating the action was performed via an AI tool. This means your recipe history and compliance records remain complete even when an AI assistant is doing the work. For regulatory affairs teams who need a clear record of every formula change — including changes made during AI-assisted R\&D workflows — the audit trail captures the full picture. Nothing is attributed to an anonymous agent; every action traces back to the authorizing team member.
