AI Chat

The AI Chat node enables interaction with a configured AI service (e.g., OpenAI) using a custom user prompt and optional context window. It returns the result of the AI response and stores it in a named output variable.

This node is useful when dynamic, intelligent responses or code generation are required based on contextual data and user input within your workflow.

---

🧠 Functionality

This node sends a request to the AI service using:

• A prompt (User Input)
• A context block (Context Window)
• A flag to return only structured code if required (Request Code Only)

It returns the AI response and optionally parses it as JSON.

---

🛠️ Inputs

Label	Type	Required	Description
User Input	text	✅	The main instruction or question for the AI (e.g., “Generate a C# class for a user object”).
Context Window	text	✅	A block of context to give the AI additional information before the prompt (e.g., system state, variables).
Request Code Only	checkbox (boolean)	✅	If true, AI is prompted to return only code. Useful for generation tasks.
ServiceAccount-OpenAI	serviceAccount	✅	References the ID of the configured OpenAI service account (used for authentication).
Save to Variable Name	text	✅	The name of the output variable to store the AI response in. The response will be parsed to JSON if Request Code Only is checked.

---

✅ Output

The AI's response is saved to the variable name defined in Save to Variable Name. If the response is JSON-compatible and Request Code Only is enabled, the response will be deserialized.

---

💡 Example Use Case

You could use the AIChat node to:

• Generate code snippets dynamically
• Provide explanations or summaries of technical content
• Return formatted JSON objects based on prompts

For example:

• User Input: Generate a class called Employee with Name, ID, and Department fields.
• Context Window: The project is written in C# and uses PascalCase naming.
• Request Code Only: true
• Save to Variable Name: generatedCode

The variable generatedCode will contain the C# class definition returned by the AI.

---

🛑 Error Handling

If the AI request fails or the service account is misconfigured:

• taskSuccess = false
• statusReturn = "Fail"
• taskMessage will include the failure message

---

⚙️ Execution Notes

• Adds current DateTime and user's country code to the context block automatically.
• Uses AISingleRequest() from the internal AIService class.
• Attempts to deserialize the AI output into an object if Request Code Only is checked.

---

🚧 Limitations / Manual Review

• Ensure that the ServiceAccount-OpenAI is correctly configured in your environment.
• ⚠️ Manual Confirmation Required: Whether other AI services besides OpenAI are supported.
• If AI returns invalid JSON, deserialization will fall back to storing as a raw string.

---