Skip to content
English
  • There are no suggestions because the search field is empty.

[Gemini Enterprise] Configure the Counta AI Connector

This is available for Beta participants only.

Overview

The Counta AI Connector, built on MCP, utilizes individual user OAuth 2.0 credentials to enforce strict data access controls. This ensures that when an employee interacts with Gemini, the AI model can only access and surface the specific business metrics, client invoices, and accounting layouts that the individual user has explicit system permissions to view.

 

Configure the Counta Connection Settings

1. Open your Google Cloud Console and go to the Gemini Enterprise dashboard.

2. In the left-hand navigation sidebar, click Data stores, then click the ➕ Create data store button.

3. Locate the search box, type Custom MCP Server, and select the Custom MCP Server card.

4. Complete the configuration form using your target environment's specific parameters. For example, to deploy a Staging setup, input the following values:

▪ MCP Server URL: https://api-staging.accountabilityoneeu.com/v1/mcp (or your production equivalent)

▪ Authentication Type: Select OAuth 2.0 from the dropdown menu.

▪ Client ID: [Paste the Client ID generated in Step 1]

▪ Client Secret: [Paste the Client Secret generated in Step 1]

▪ Authorization URL: https://auth-staging.accountabilityoneeu.com/connect/authorize/

▪ Token URL: https://auth-staging.accountabilityoneeu.com/connect/token/

▪ Scopes: mcp:tools offline_access (Separate these two scopes with a single space if required).

5. Click Login to perform the initial administrative verification handshake, then click Continue.

Finalize and Enable User Actions

1. Provide a Server Description (Recommended): In the server description text box, enter a summary that tells Gemini's routing engine exactly what capabilities this server offers. 

 

💡 Note for Administrators:  You can customize this text to fit your organization's internal guidelines. However, to ensure that Gemini accurately routes natural language requests to your data stores, we recommend using or adapting the suggested text below. 

Suggested Description Text:

"Directs queries to the Counta MCP API to either retrieve specific reports (such as client profitability summaries) or data, or to create and update specific records within the system.'

 

 

2. Name the Connector: Set the Data Connector Name field to a clean reference that reflects the target tier (e.g., counta-api-staging or counta-api-prod).

3. Click Create.

Activating the Tools for Your End Users

1. Wait a moment on your dashboard until the status for your connector switches to Active.

2. Click on your data store link, and navigate to the Actions tab.

3. Click Reload custom actions. This triggers a secure handshake with the server to inventory available features.

4. Select the specific business actions and tools you wish to expose to your enterprise chat users.

5. Click Enable actions.