Enhanced GPT Actions with Retaining and Utilizing History Feature

Presenting History for GPT Actions

We are thrilled to unveil a substantial augmentation to our GPT action – the inclusion of a retention and utilization feature for their history. This feature allows GPT actions to generate more pertinent and tailored responses based on previous interactions.

Types of Histories

The history can be categorized into 5 types:

• This Sub Account: Records all GPT conversations across all workflows under the sub account where the History type is set as “This Sub Account”.
• This Workflow: Stores all GPT conversations within the workflow, relevant when the History type is “This Workflow”.
• Per Execution: Logs all GPT conversations within the workflow for a single execution, tagged under the History type “Per Execution”.
• This Step: Retains all GPT conversations for the distinct action across multiple executions classified as “This Step” under History type.
• Custom: Provides the user with the authority to generate a unique History type to be deployed within the same or multiple workflows.

Usage Instructions

Follow these simple steps:

1. Activate the feature in Labs.
2. Select the “GPT Powered By OpenAI” action.
3. Toggle the “Enable history” switch.
4. Pick out the “History for” from the dropdown menu.
5. Include system instructions.

Additional Options

Additional options include System Instructions and Exclusions from History:

• System Instructions: The user has the flexibility to incorporate instructions to provide additional details to the GPT action for creating a more precise and desirable outcome.
• Exclude instructions from history: Useful when the user prefers to keep the System Instructions out of the “history”.
• Exclude responses from history: If toggled ‘on’, the responses generated by this action will not be recorded in the history. Particularly beneficial when the feedback is sentiment-based (positive or negative) or if any analytics are received as a result.

Important Considerations

Please note the following:

• If Enable History is turned ‘on’, the default preference will be GPT 4 models.
• History will be maintained individually for each contact point.