PingOne DaVinci

Using the PingOne AI Assistant for DaVinci

The PingOne AI Assistant for DaVinci is a utility that uses artificial intelligence (AI) to help you understand and analyze DaVinci flows using natural language. It works directly within the flow canvas to provide contextual explanations, identify configuration issues, and summarize flow executions. It’s meant to accelerate common tasks, such as understanding unfamiliar flows and diagnosing issues.

This product uses generative AI features. When using this feature, you are interacting with an AI system. Outputs are provided as-is for general informational purposes and might be inaccurate or incomplete. No warranties are made regarding their accuracy or reliability. Customers are responsible for verifying AI-generated content and any decisions based on it.

The AI Assistant is an opt-in feature. To enable it:

  1. In the PingOne admin console, go to Settings > Environment Properties > Manage Opt-Ins.

  2. Select PingOne AI Assistant for DaVinci.

  3. Click Save.

The AI Assistant can help with several broad task categories:

  • Answer questions about flows and configurations: You can provide a flow name or select a portion of a flow and ask questions about the configuration.

  • Direct you to documentation: You can ask for information about a subject and receive links to relevant documentation.

  • Validate flow configurations: You can ask the assistant to identify likely problems with a flow’s configuration.

  • Troubleshoot flow execution issues: You can provide a flow execution ID and ask the assistant to help identify the cause of the failure.

  • Summarize flow executions: You can ask the assistant to summarize what happened during a specified flow execution, including the nodes that ran and the choices made by the user.

Launching the AI Assistant

  1. In the DaVinci admin console, click Flows, then select a flow.

  2. Open the assistant using one of the following methods:

    • In the upper-right corner, click AI Assistant.

    • Select a portion of the flow canvas, then right-click and select Ask AI.

Sample tasks

You can use the AI Assistant to help with a variety of tasks. This section describes some of the common tasks and provides an explanation and example prompts for each one.

Understanding an existing flow

You can use the AI Assistant to analyze an existing flow and provide an explanation of how it works, what variables and inputs it uses, and what results it returns.

This can be useful when you need to understand a flow built by someone else before using it in your environment or modifying it.

Example prompts

The following are examples of prompts you can use to analyze an existing flow:

  • What does this flow do?

  • How many subflows are associated with this flow? Provide a high-level description of what each of these subflows does.

  • What does this group of nodes do?

  • Help me understand the script embedded in this flow.

  • List the variables configured in this flow.

Expected output

The AI Assistant explains what the specified flow does in plain language. For example, if you ask What does this flow do?, the response could be:

The flow handles user authentication, password recovery, and user registration processes. It starts by prompting the user for credentials, then attempts to look up the user and validate their password. Based on the outcome, it either creates a session, initiates a password reset, or directs the user to register. The flow also manages error conditions and email verification for new registrations.

Troubleshooting a failed flow execution

You can use the AI Assistant to review a specific flow execution and, if the flow execution failed, diagnose the reasons for the failure.

This can be useful if you’re trying to determine the cause of persistent flow failures.

Example prompts

The following are examples of prompts you can use to troubleshoot flow executions:

  • Can you briefly explain what happened in the flow execution id <execution ID>?

  • Summarize the sequence of events in the flow execution id <execution ID>.

  • Why was the user <email address> not able to log in?

Expected output

The AI Assistant provides an analysis of the flow execution and the errors that occurred. For example, if you ask Can you briefly explain what happened in the flow execution id <execution ID>?, the response could be:

Flow Name: User Enrollment Flow

Execution ID: <execution ID>

Overall Outcome: Error encountered

Execution Path:

The flow attempted to execute a single node:

Node Title: Keyless Enrollment

Connector Name: Keylessv2Connector

Capability Used: enrollment

Error Details:

The execution encountered an error during the enrollment capability of the Keylessv2Connector. The specific details of the error are not provided in the input, but it indicates that the keyless enrollment process failed.

Identifying configuration problems

You can use the AI Assistant to critique a flow, identifying errors or potential problems with the flow’s current configuration.

This can be useful as a routine check when you are building a flow, helping you to identify and fix issues early.

Example prompts

The following are examples of prompts you can use to identify configuration issues:

  • Are there any configuration issues in this flow?

  • How can I fix the above misconfigurations in this flow?

  • What configuration errors are present in this flow and how can I fix them?

Expected output

If the AI Assistant finds configuration issues, it describes the issues and provides recommendations for fixing them. For example, if you ask Are there any configuration issues in this flow, the response could be:

"Yes, there is a configuration issue in this flow.

Issue:

Unused Variable: The variable {{global.flow.variables.UnusedVar}} is defined but not used anywhere in the flow.

Recommendation:

Remove the unused variable: Since {{global.flow.variables.UnusedVar}} is not being utilized, it’s best to remove it to keep the flow clean and prevent potential confusion."

Best practices and validation

  • Validate AI responses before applying any changes to your flows.

  • Use follow-up prompts to refine your query and gather additional information.

  • When troubleshooting flow executions, compare the AI Assistant results with analytics and log data.

  • Use the AI Assistant as a guide, not a source of truth.

Limitations

  • AI might not always interpret complex flows accurately.

  • Flow summaries are limited to the data made available by the flow’s log level, so some information might not be available.

  • The AI Assistant can only summarize flow executions that took place within the past 15 minutes.

Troubleshooting

Issue Solution

Error starting AI Assistant displays on first use

The AI Assistant can take up to five minutes to start on first use. Wait five minutes, then try again.

Error is displayed after using the AI Assistant for an hour or more: I’m sorry, but I’m unable to assist with your request at the moment. Please try again later.

Close the current AI Assistant panel, then open a new one.