Documentation Center

Setting up and configuring AI Assistant

Learn how to set up and connect to a Large Language Model from AI Assistant.

Before you begin

You need a paid Open AI or Azure Open AI subscription as described here.

About this task

In this task, we will explain how to set up AI Assistant to connect to the Large Language Model (LLM) of your choice.

Procedure

  1. In Trados Studio, open any file in the Editor so you can access the AI Assistant.
  2. From the View ribbon, select AI Assistant. The AI Assistant view part is displayed.
  3. On the AI Assistant view part, click the cog wheel icon to open the settings.
    The Settings dialog box is displayed.
  4. The General tab on the Settings dialog box displays the current configuration as follows:
    • Connections - lets you choose any of the existing LLM connections from the Selected Connection: drop-down list and allows you to add or edit any connections. This list is initially empty.
    • Search Options - lets you refine the search settings as appropriate.
    • Search Results- lets you enter a value in the Assign Match value box so you can later:
      • choose and identify the best translation results
      • overwrite existing TM matches with the ones returned by your preferred LLM connection
    We are now ready to add and test our first connection. Follow these steps: We are now ready to add and test our first connection. Follow these steps:
  5. In the Connections group box, click Add.
    The Add Model Connection dialog box is displayed.
  6. Provide a name for your connection. This can be any name, such as My first OpenAI LLM connection. We recommend using a meaningful name that will work intuitively for you.
  7. Next, choose a Provider. In this example, we will use an OpenAI connection, so we'll use the OpenAI setting as is.
  8. The Endpoint is already preconfigured to use the recommended connection so we will accept it as is.
  9. Under API Key, enter the API Key that you generated when you first set up your OpenAI account. We recommend copying/pasting it from your OpenAI account.
  10. Finally, under Model choose the choose the preferred model you wish to use. For the purposes of this example, we will use gpt-4o, which is the latest model at the time of writing this documentation.
    • gpt-4o is the recommended option as it provides the best balance between performance, accuracy and cost.
    • For lower cost, pick any of the gpt 3.5 models available from the list. gpt 3.5 turbo is recommended as it provides the best performance.
  11. We can keep the other existing settings and are now ready to test the connection. To achieve this, click Test Connection.
    If the connection can be established successfully, AI Assistant will display a confirmation message. If it is not successful, we recommend that you check the settings and try again.
  12. Click OK to confirm.
    The new connection is displayed in the Connections drop-down list and is ready for use.
  13. Repeat this procedure to add any further connections or edit the existing ones. In the rest of this tutorial, we will use the gpt-4o model for the examples.