Ask User

The Ask User node is designed to prompt the user for input during workflow execution. It supports two distinct modes of interaction: a Simple Question directly displayed to the user or a more dynamic AI-driven data collection session that utilizes a language model to gather multiple fields.

Overview

This node is typically used when workflow execution requires human input before proceeding. The communication is facilitated via SignalR and displayed through the web-based workflow interface.

Configuration

Users configure this node via the standard modal UI. The following fields are presented depending on the selected "Ask Method":

Ask Method

• Label: Ask Method
• Input Type: picklist
• Options:
  • Simple Question
  • Use AI
• Default Value: Simple Question
• Description: Select the method for asking the user a question.
• Visible When: * (Always visible)

---

For Simple Question Method

Question

• Label: Question
• Input Type: text
• Default Value: What do you need to proceed?
• Description: Enter the question to present to the user.
• Visible When: Simple Question

---

For Use AI Method

Description

• Label: Description
• Input Type: text
• Default Value: Collecting user details
• Description: Describes the purpose of collecting the user input. Used in the system prompt sent to the AI.
• Visible When: Use AI

Required Fields

• Label: Required Fields
• Input Type: key-value list (dictionary)
• Description: A dictionary where each key is a required field name and each value describes the expected input type (e.g., "City": "Enter a city name").
• Visible When: Use AI

---

Runtime Behavior

Simple Question Mode

1. Displays the user-specified Question to the end user.
2. Waits for a response via SignalR and stores the result in:
   • data["UserAnswer"]

Use AI Mode

1. Sends a system prompt to the AI describing the required fields and context.
2. Uses the AI to generate a natural-language question.
3. Displays the AI-generated message to the user.
4. Parses the response using multiple attempts to extract valid JSON fields.
5. Re-prompts the user for missing fields if necessary (up to 3 retries).
6. Final result is stored in:
   • data["RequiredFields"] (as a dictionary of key-value pairs)

---

Outputs

• The output of the node is written to taskParameters.data:
  • "UserAnswer" (if Simple Question was used)
  • "RequiredFields" (if Use AI was used)

---

Notes

• This node requires a valid OpenAI service account connection string.
• If the connection is lost or the user does not respond within 5 minutes, the task will timeout.
• The system supports ongoing conversations via the ChatWorkflowHub and handles session-based prompts using ConnectionId and SessionId.

---

Example Use Cases

• Prompting a user to approve a workflow step before continuing.
• Collecting contact or project details as part of an onboarding flow.
• Interactively gathering structured input using a conversational AI assistant.

---

Errors and Troubleshooting

Issue	Cause	Resolution
ConnectionId not found in task data	The session context was not correctly passed to the node.	Ensure the node is triggered in a real-time session with proper SignalR setup.
Required AI Service account not found	Invalid or missing service account credentials.	Configure the appropriate OpenAI credentials in the service registry.
User did not respond within 1 minute	No input received during timeout period.	Prompt user again or allow retry logic in parent workflow.

---

Developer Notes

• Node Type: task
• Task Type: AskUser
• Hub used: ChatWorkflowHub
• AI interaction relies on: AISingleReequest, Chat methods
• Result stored under JobTask.data