Overview
Frends 6.0 introduces LLM-Assisted Task Configuration, allowing users to configure Tasks seamlessly using natural language prompts. This feature simplifies the setup of Task parameters, low-code mappings, and variable usage, reducing manual effort and improving efficiency in integration development.
How It Works
Effortless Task Configuration
Describe the Task in Natural Language – Instead of manually selecting parameters and mapping values, users can simply describe what they need the Task to do.
Automatic Parameter Setup – The AI automatically configures Task parameters based on the provided description.
Low-Code Mappings – AI suggests variable mappings, making it easy to incorporate Task results, environment variables, and process variables.
Persistent Context – The AI retains conversation history, allowing users to refine Task configurations iteratively within the same session.
Accessing AI-Assisted Task Configuration
This feature is available in the Process Editor within the Task shape. Users can access it by clicking the Frends Assistant button inside the Task configuration panel.
Context Panel: Understanding Data Sent to the AI
The Context Panel (accessible via the top-right button in the Frends Assistant modal) displays all data being sent to the LLM, including:
Task Type & Parameters – The AI considers the selected Task type and available configuration fields.
Variable References – The AI suggests using available process variables, environment variables, and previous Task results.
🔒 Security Note: Only metadata and variable references are shared—no actual data values are sent.
Security & Privacy
We prioritize security and transparency:
No Training on User Data: The AI does not retain or learn from your data.
Customizable Context: Users can review and modify what information is shared.
Audit Log for Transparency: Admins can track all LLM sessions and review data sent to the AI.
Best Practices for AI-Generated Task Configuration
Be Specific in Descriptions: Clearly describe what the Task should do, including expected inputs, outputs, and logic.
Review AI Suggestions: AI-generated configurations are a starting point—always verify parameter mappings.
Iterate & Refine: Use the persistent chat context to fine-tune configurations without starting over.