Model Context Protocol (MCP) lets you securely connect AI platforms, such as Claude, ChatGPT, Gemini, and more, to your Birdeye account so they can read your business data and, if you allow it, take actions on your behalf through a standard set of tools.
As an account owner or admin, you control exactly what a connected AI platform can do: which modules it can access, whether that access is read-only or read and write, and which roles on your account are even allowed to use MCP. This guide walks you through connecting an AI platform, configuring tool-level permissions, and controlling MCP access by role.
Connect an AI platform to Birdeye via MCP
This example uses Claude, but the same flow applies to any AI platform that supports MCP connectors.
In your AI platform, go to Settings > Connectors.
Find Birdeye Inc and click Connect.
You'll be redirected to the Birdeye authorization page. If you're already logged in to Birdeye, the connection completes automatically without asking you to log in again.
Once connected, the AI platform shows the tools currently enabled for it in Birdeye. By default, only read-only tools are available until write access is turned on from Birdeye (covered in the next section).
Configure MCP tool permissions in Birdeye
1. Go to Settings > Integrations.
2. Click Model Context Protocol (MCP).
The MCP settings page lists every module that supports AI platform access, including Reviews, Reports, Business, Custom Fields, Business Media, Surveys, Listings, AI Search, Social, Ticketing, Contact, Webhook, User, Google Services, Aggregation, Insights, and Competitors. Each module shows a short description of what read and write access allows.
Turn a module on or off
Click the arrow next to a module name to expand it and view its individual tools.
Use the toggle beside the module name to enable or disable the entire module at once. Disabling a module turns off every tool within it, but keeps your previous tool selections in place in case you re-enable it later.
Enable or disable individual tools
Inside an expanded module, each tool is listed with a checkbox and a Read or Write label. Check or uncheck individual tools to fine-tune exactly which actions an AI platform can perform within that module.
Use bulk actions to update permissions faster
Instead of updating modules one at a time, use the Bulk select dropdown at the top right of the page to apply a preset across every module at once:
Enable all read-only tools – gives access to every read tool across all modules, and turns off write access.
Enable all read and write tools – gives full read and write access across every module.
Disable all – turns off MCP access completely across every module.
If you manually adjust individual tools after applying a preset, the label next to Bulk select updates to reflect that your selection is now custom.
Search for a specific tool or module
Use the search field at the top of the page to quickly filter to a module or tool by name — for example, typing survey filters the list down to the Surveys module and its tools.
Save your changes
Click Save once you've made your changes.
If an AI platform is already connected, disconnect and reconnect it (or refresh its tool list) to pick up the updated permissions.
For example, after enabling all read and write tools, Claude's Connectors panel reflects the full set of available tools, split into read-only and write/delete groups:
Control MCP access by role
Tool-level permissions in Integrations control what an AI platform is allowed to do on your account overall. Separately, you can control which roles are permitted to use MCP at all. From Settings, go to Account, and click Roles.
Create a new role or edit an existing one.
On the Permissions tab, scroll to Model Context Protocol (MCP). This permission is disabled by default for every new role.
Set it to Enable to allow users with this role to access MCP tools and use any connected MCP client.
NOTE:
Owner and Admin roles automatically inherit the account's full MCP configuration, since these roles already have complete ownership of the account. When Owners or Admins change the MCP settings under Integrations, the change applies to the account as a whole.
NOTE:
Guest and Guest Plus roles always get read-only access to MCP tools, even if write access is enabled at the module level. MCP doesn't create a separate permission layer of its own — it follows whatever access level a role already has on the account, and Guest / Guest Plus roles are read-only by design.
NOTE:
If MCP is disabled for the entire account (using Disable all under Bulk select), the Model Context Protocol (MCP) row won't appear at all when creating or editing a custom role.