After configuring an AI Agent in a workflow, whether you create a new agent or use an existing one, you must define how data flows between the application and the agent. This is done using Input Mapping and Output Mapping.
Input Mapping defines what application data is passed to the AI Agent for evaluation.
Output Mapping defines where the agent’s result or decision is written back so the workflow can continue correctly.
¶ 1. Input Mapping
Input Mapping specifies the application data that the AI Agent evaluates during workflow execution.
- Switch to the Input Mapping tab.
- For each input property, select the corresponding application form field from the App Elements dropdown.
- These mapped values are passed to the AI Agent at runtime. This ensures the agent evaluates the correct data when making a decision.
- After completing the input mapping, click Save to apply the changes.
¶ 2. Output Mapping
Output Mapping defines how the AI Agent’s decision or result is used by the workflow.
- Switch to the Output Mapping tab.
- Map the agent’s output properties to the appropriate application form fields or workflow variables.
- This allows the workflow to use the agent’s output in subsequent steps such as routing, approvals, or conditional actions.
- After completing the output mapping, click Save to apply the changes.
¶ 3. Why Are Input and Output Mapping Required?
AI Agents do not automatically know which application data to evaluate or where to store their results. Input Mapping ensures the agent receives the correct data for evaluation. Output Mapping ensures the agent’s decision is available for the workflow to act on. Both mappings are required to ensure accurate decision-making and predictable workflow behavior.