The Workflow Explainer Agent explains an existing workflow in DCKAP Integrator through natural-language conversation. Instead of opening every step and configuration yourself, you ask the agent questions and get answers based on the workflow’s configuration.
You can use it to:
- Understand what a workflow does as a whole
- See how its steps work together
- Get answers about specific steps, logic or API calls
It is for anyone who needs to understand a workflow in DCKAP Integrator.
Current version: Workflow Explainer Agent v1.
Getting started
- Open the workflow you want to understand in Workflows.
- Review it and make sure no sensitive information is hardcoded in its configuration.
- Select the RhineAI icon in the bottom-left corner.
- Provide consent confirming the workflow contains no hardcoded sensitive data.
- Type your question and start the conversation.
The agent replies in natural language. You can keep asking follow-up questions to explore other parts of the workflow.
Asking questions
Ask a broad question for a summary, or ask about a specific step for detail.
| What you want to know | Example question |
| The workflow’s overall purpose | “What does this workflow do?” |
| A full walkthrough | “Explain this workflow to me.” |
| Where specific logic lives | “Where does customer creation logic exist?” |
| What a single step does | “What does Step 3 do?” |
| How a code runner works | “Explain this code runner in step 10” |
| Which APIs are used | “What are the api calls used?” |
Tip: Ask for a step-by-step explanation when you need a detailed walkthrough. The agent will format its answer step by step.
Data security and consent
Consent is required because workflow configuration may include hardcoded values that should not be sent to AI models.
Before sending workflow information to the AI model, DCKAP Integrator applies these security controls:
- Sanitization of workflow information
- Guardrail checks
- Protection against inappropriate or unsafe inputs
These controls do not replace your own review. You are responsible for checking the workflow before giving consent.
Current limitations (v1)
The agent’s answers depend on the information and configuration in the workflow. In v1, explanations of the following may be incomplete or less detailed than expected:
- Snippets
- Replace Variables
- Initialize Variables
- Detailed API definitions
- Non-API systems: FTP, SFTP, SQL
The agent explains workflows only. It does not:
- Suggest changes to a workflow
- Recommend improvements
- Modify or edit the workflow
- Automatically fix workflow issues
These capabilities are planned for upcoming releases.
Note: The agent is AI-powered and can make mistakes. Always check the actual workflow configuration before making decisions about how a workflow operates.
Troubleshooting
| Problem | Likely cause | What to do |
| You can’t start a conversation | Consent hasn’t been provided for this workflow | Review the workflow for hardcoded sensitive data, then provide consent |
| The answer is vague or incomplete | The question involves a v1 limitation (Snippets, variables, detailed API definitions, FTP, SFTP or SQL) | Check those parts directly in the workflow configuration |
| The answer is too high-level | The question was broad | Ask about a specific step, or ask for a step-by-step explanation |