OpenAI is an artificial intelligence research laboratory which includes the for-profit OpenAI LP and its non-profit parent company, OpenAI Inc. OpenAI conducts research in artificial intelligence and develops advanced AI systems.
They are known for creating models like GPT-3 (Generative Pre-trained Transformer 3), a state-of-the-art language model capable of understanding and generating human-like text based on the input it receives.
The OpenAI plug-in allows users to receive a model-generated message from the OpenAI assistant by submitting a message using OpenAI’s chat completion capability. This plug-in is available for download on Automation Hub to enhance your Workload Automation environment.
After downloading the plug-in, log in to the Dynamic Workload Console and go to Workload Designer. Create a new job and select “OpenAI” in the Other section.
Figure 1: Job definition page
The Action Tab
In the Action tab you must enter all information required both for carrying out a test connection (which is usually advisable) and for using the plug-in:
- OpenAI API Key: API key which you can generate on your OpenAI personal profile
- Model: The model used to generate the output message
- User message: The message you want to send to the assistant
There are also two optional fields:
- Username: This is linked to the OpenAI API key field so that, if necessary, you can create a User object inside your environment, choosing as password the OpenAI API key. You can then insert the name of the newly created User in the username field and choose the User option for the OpenAI API key field.
- System message: This is a message that can be used to help guide the assistant’s actions. If this message is specified, it will be the first message in the conversation, appearing directly before the user message which you will specify. An example of a system message can be found at this link.
Figure 2: Action page
Submitting Your Job
You can submit the job according to the current plan. After selecting the Submit icon, click OK. A confirmation message will appear. You can then move to the Monitoring and Reporting tab to check status.
Figure 3: Submitting job
Figure 4: Monitor page
The plug-in manages cases in which you explicitly request a JSON object as a result. For example, you could ask: “Give me only the JSON; don’t provide any additional text” at the end of the user message you specify.
Figure 5: Job log > first example
Instead of asking the assistant for detailed observations about the previous JSON, you could also receive something like the following job log.
Figure 6: Job log > second example
Figure 7: Output properties > first example
Figure 8: Output properties > second example
You can download the OpenAI plug-in here.