COMSOL 6.4 - The Chatbot Window

From the menu on the toolbar, select to open a window for communicating with an OpenAI, Azure OpenAI , or OpenAI API-compatible model. This feature assists with generating and correcting COMSOL API for Java® code directly within COMSOL Multiphysics, as well as answering modeling questions related to the software. Using the window can help with both programming tasks using the COMSOL API for Java and general modeling tasks within the COMSOL Desktop. The window is available in both the Model Builder and Application Builder workspaces.

The window is only available with the Windows version of the COMSOL Desktop. The COMSOL software must also be installed with the CHATBOT feature included in the installation and enabled using the Enable checkbox on the page in the window.

The window requires authentication details for your selected provider, for example, an API key or connection settings for OpenAI, Azure OpenAI, or another OpenAI API–compatible service.

The first time that you use the window you need to configure a chatbot provider. In the window, click , which opens the window. Choose a and and enter your key in the field and click . If you are using Azure OpenAI, also enter an endpoint URL in the field and a model deployment name in the field. For the OpenAI API-compatible option, instead enter a and .

If enabled, the option allows the to browse the COMSOL documentation for information.

Once configured, you can start a chatbot conversation. Note that obtaining an API key typically requires signing up for an API key payment plan. For more information, see the Reference Manual.

The window functionality connects to an external AI system. A subscription agreement with the AI system host is needed to obtain an API key. COMSOL is not a party in this agreement, and use of the external AI system may be subject to additional fees.

It is your responsibility to ensure that your use of the AI system is in accordance with laws and policies that may apply. It is also your responsibility to review the suitability of any suggestions made by the AI system.

The following window examples use various models from OpenAI. The output may vary depending on the selected provider and model, and due to the nature of large language models (LLMs), some random variation between responses is expected.

The window includes a set of toolbar buttons and menus, as shown in the figure below. In addition, information on the number of tokens used is displayed in the window’s top-right corner, next to used:

From the list in the top-left corner of the window, you can select a subject: , , and , as shown in the figure below.

The selected subject influences how the chatbot interprets your prompts. The chatbot understands Java syntax but not all aspects of the COMSOL API. However, you can guide the chatbot to help you write code by using the functionality of the Method Editor.

For example, let us use the window to add annotations to the Pacemaker Electrode tutorial model which is available in the Application Libraries under with the filename pacemaker_electrode.mph.

If we open this model and record the creation of one annotation feature for the plot, then the following code is produced:

Say that you would like to create 10 such annotations along a line. While having the subject selected, the following prompt will then guide the chatbot to write functional code:

The figure below shows the resulting output from the window, which can be used to automate this task by entering the code in a method or in the window.

To run the code in the window, right-click in the output area and select . Alternatively, you can click the button in the toolbar.

Then, in the window, click . The result is shown in the figure below.

For more details about the window, see the Reference Manual.

You can attach the entire model history, represented as a Model File for Java, to a prompt:

This is shown in the figure below.

When prompted to compact the model history, choose to remove redundant steps, unless you specifically want to include them.

You can also attach a specific model tree node:

To attach images or screenshots: