Create Grounded Conexões de modelo with Google Vertex AI RAG capability

Use the Google Vertex AI RAG (Retrieval Augmented Generation) capability to create Fundamentado em armazenamento de dados Conexões de modelo to generate accurate and contextually relevant information referenced from Google Data Source.

Nota:

Google Vertex AI Fundamentado em armazenamento de dados Conexões de modelo is now available in Automation 360 v34 release on Nuvem as well. You can use this feature in Nuvem and No local.

Google Data Store with document chunking is now supported to ensure optimal results in automation executions. You can enable document chunking in the Google Data Store for using Google Vertex AI Grounded models in AI Agent Studio.

We are now offering you the option to create Fundamentado em armazenamento de dados Conexões de modelo using the Google Agent Builder service. A search query on the Google Data Store retrieves relevant content from large-data-sets and feeds it to the model to generate accurate response.

During an automation execution, using Conexões de modelo created with Fundamentado em armazenamento de dados retrieves the response by referencing the Google Data Source in Google Data Store. This ensures an optimized response from the relevant content with higher accuracy. Grounding is an important aspect of using a foundational model as it returns grounded responses against organizational data and prevents response inaccuracies and model hallucinations.
Nota: To use the Google Vertex AI RAG capability in AI Agent Studio, you would first create a data source in Google Agent Builder. Then create a Conexão de modelo using the Fundamentado em armazenamento de dados option.

See: Data stores in Google Vertex AI.

Pré-requisitos

The Automation Admin requires these roles and permissions to create and manage Conexões de modelo for their business organization.
  • Role: AAE_Basic, Automation Admin custom role
  • Permission: Attended Bot Runner
  • Settings: AI Data Management must be enabled by the Automation Admin and the check box selected for Allow users to disable logs on AI Skills option.

See Funções e permissões para Ferramentas de IA for the Automation Admin custom role permissions.

Other requirements:
  • As mentioned earlier, you would first create a Google Data Source to create a Fundamentado em armazenamento de dados Conexão de modelo and use it successfully in an Habilidade de IA.
  • If you want to store authentication details in a credential vault, have that information handy. See Armazenamento seguro de credenciais no Credential Vault.
  • To test a Conexão de modelo, you must be connected to a Agente de bot 22.60.10 and later. As part of the test, you would have to run the bot on your desktop. Hence ensure the Agente de bot is configured to your user. For this task, if you have to switch connection to a different Control Room, see: Alternar o registro do dispositivo entre instâncias da Control Room.
  • You would need access to the Gravador package and the Habilidades de IA package to test the connection successfully. A test Prompt would be executed to test the Conexão de modelo.

Procedimento

  1. In your Control Room environment, navigate to AI > Model connections > Create model connection.
  2. In the Create model connection screen you would configure these Connection settings:
    Create Grounded Conexões de modelo with Google Vertex AI RAG capability
    You can manually enter the model name in the Choose a model or create a custom one field. The name you enter will be used for creating the Conexão de modelo.
    1. Model connection name: Provide a name for easy identification of the Conexão de modelo.
    2. Description (optional): Add a meaningful short description defining the connection.
    3. Choose a vendor: Choose a foundational model vendor from the supported list of vendors. For creating a Fundamentado em armazenamento de dados Conexão de modelo with Google Vertex AI, select Google Vertex AI from the drop-down list.
    4. Choose a type: Choose Fundamentado em armazenamento de dados for using the RAG capability for Google Vertex AI.
    5. Choose a model or create a custom one: Choose a model from the drop-down list.
      Additionally, we also support other models available from Google Data Store, which are not available in the drop-down list. If you want to add a model from the Google Data Store, you would have to enter the model name and version with the complete URI of the model. For example: If the model is Gemini 1.5 Flash 001, the format would be gemini-1.5-001/answer_gen/v1.
      For a complete list of supported models for each foundational model vendor, see Perguntas frequentes sobre AI Agent Studio.
    6. Click Next to proceed to the Authentication details section.
  3. In the Authentication details section, configure these settings:
    1. Project Name: This is the Google Cloud account project.
    2. Region: Select a region from the drop-down list to connect for authenticating the Conexão de modelo. You can also add your own region that you configured when you created a data source in Google Agent Builder.
      See Create a data store.
    3. Control Room OAuth Connection: Create an OAuth 2.0 Client ID. A client ID is used to identify a single application to Google's OAuth Servers.
    4. After setting up the authentication details, confirm and click Next to proceed to the Test connection section to test the Conexão de modelo.
    Nota: For details on setting up the Google Cloud Project and Control Room OAuth Connection for Google Vertex AI, see Vertex AI: Ação Conectar
  4. Click Test connection to make sure all connection details have been defined correctly and check if the connection is working.
    This is a desktop operation using a Agente de bot. Use Agente de bot 22.60.10 and later for successful testing.
    • If the connection works as expected, the system will process the request and you will get a system generated success message.
    • If the connection does not work as expected, you will get a system-generated message stating the reason for the connection failure. For example, if you have not downloaded the supported foundational model package to your workspace, you would get an error message. You would have to download the package and then retest the Conexão de modelo.
    • If the testing a Conexão de modelo is unsuccessful or if you leave the task incomplete, the Conexão de modelo will not get saved and you will have to restart the process of creating the Conexão de modelo.
  5. Click Next to proceed to the Invite roles section to begin assigning custom roles to users.
    The Automation Admin would create custom roles and assign the Conexões de modelo to the role, which can then be assigned to users. Only users assigned to this custom role can use this Conexão de modelo.
  6. Assign the Conexão de modelo to the custom role (using RBAC) so users assigned to the role can access it.
  7. Click Create model connection to complete creating the Conexão de modelo.
    After successfully creating the Conexão de modelo, the Pro Developer would use it to create an Habilidade de IA.

    See: Create Habilidades de IA with Fundamentado em armazenamento de dados Conexões de modelo

    .

Próximas etapas

After creating and testing the Conexão de modelo, you would assign it to the Pro Developers, who would use this connection to create Habilidades de IA.

As the next step in your sequence of tasks, go to Create Habilidades de IA with Fundamentado em armazenamento de dados Conexões de modelo and create an Habilidade de IA and connect to a Fundamentado em armazenamento de dados Conexão de modelo to use it in an automation.