Chat Completion

The Chat Completion Step is used to generate a Chat Response and is available for use across several AI Modules. The step enables Users to input Model and AI Prompt information to output a Chat Response. To utilize this step, Users must have AI Common, and the desired AI Module installed, as these Modules will be used to populate the Chat Response. Since this step is available across several AI Modules, Users can quickly switch between models without the need for an additional step.

Prerequisites 

• This step requires the AI Common Module to be installed before it will be available in the toolbox. 
• Users must create a Project Dependency for the AI Common Module and any other AI Modules that will use the Chat Completion step. 
• Users need to be on v9 to access this step.

Properties

Inputs

Settings

Outputs

Common Errors

Using a Bad API Key

When utilizing the Chat Completion Step with the Open AI Module, the step will not return desired results unless the API Key is working properly. If an incorrect API Key is entered on the Open AI Settings Page, the Step will error out in the Debugger. 

To correct this, ensure the API Key utilized is accurate and there are no issues with an associated Open AI account. 

Exception Message:

Exception Stack Trace: DecisionsFramework.Design.Flow.ErrorRunningFlowStep: Error running step Chat Completion 2[CreateChatCompletionStep] in flow [Using an AI Prompt in a Flow]: HTTP 429 (insufficient_quota: insufficient_quota) You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors.
 ---> System.ClientModel.ClientResultException: HTTP 429 (insufficient_quota: insufficient_quota) You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors. at OpenAI.ClientPipelineExtensions.ProcessMessage(ClientPipeline pipeline, PipelineMessage message, RequestOptions options)
 at OpenAI.Chat.ChatClient.CompleteChat(BinaryContent content, RequestOptions options)
 at OpenAI.Chat.ChatClient.CompleteChat(IEnumerable`1 messages, ChatCompletionOptions options, CancellationToken cancellationToken)
 at OpenAI.Chat.ChatClient.CompleteChat(ChatMessage[] messages)
 at Decisions.AI.OpenAI.Steps.Completions.CreateChatCompletionStep.CreateChatCompletion(CreateChatCompletionRequest request)
 at Decisions.AI.OpenAI.Steps.Completions.CreateChatCompletionStep.Run(StepStartData data)
 at DecisionsFramework.Design.Flow.FlowStep.RunStepInternal(String flowTrackingID, String stepTrackingID, KeyValuePairDataStructure[] stepRunDataValues, AbstractFlowTrackingData trackingData)
 at DecisionsFramework.Design.Flow.FlowStep.Start(String flowTrackingID, String stepTrackingID, FlowStateData data, AbstractFlowTrackingData trackingData, RunningStepData currentStepData)
   --- End of inner exception stack trace ---

Using the Chat Completion Step in a Flow

The following example demonstrates utilizing the Chat Completion step with a Model from Google Gemini. To set up the Step in Decisions:

1.  Drag the Chat Completion step onto the workspace and connect it to the Start and End steps. 
2. Navigate to the Properties Panel and customize the settings according to the desired Model. In this example, Gemini -1.0-pro was used as the model, and a prompt was selected from the AI Manager. 
3.  For the Inputs, enter desired variables into the Find and Replace fields. Ensure these variables correspond to the chosen prompt. 

After the desired settings and Inputs have been configured in the Properties Panel, select Debug. The Flow should generate a Chat Response that represents something similar to the screenshot below.