AI Assistant Node
The cornerstone of Delaverse's intelligent automation, leveraging advanced machine learning models to perform tasks like data analysis, content generation, conversational responses, and predictive modeling.
Versatile for both non-technical users and developers, the AI Assistant Node processes inputs and delivers tailored outputs based on user-defined instructions, acting as a digital brain for your business needs.
Video Tutorial
Watch this comprehensive tutorial to see the AI Assistant Node in action.
Core Functionality
Data Analysis
Processes structured (e.g., Google Sheets) or unstructured (e.g., Google Docs) data for insights like RFM lead scoring, sales trend analysis, inventory forecasting, or customer behavior insights.
Content Generation
Creates text outputs like SMS campaigns, social media captions, blog articles, or product descriptions, adhering to brand tone.
Conversational Outputs
Generates responses for interactive channels like Frame Chat or Telegram for customer/team communication or automated Q&A.
Predictive Modeling
Forecasts outcomes like sales performance, inventory needs, customer churn, or demand planning based on historical data.
Configuration Guide
Adding the Node
Select Assistant Node from the Playground menu, click to drop into the canvas, and assign a title (e.g., Lead Analyzer or Content Generator).
Select AI Provider
After entering the node title, select the AI Provider. Currently available options are OpenAI, Grok (xAI), and Claude (Anthropic).
Choose Model
Next, choose a Model based on the selected provider. OpenAI offers gpt5, gpt4.1, gpt4o, o3, and o1. Claude provides Sonnet, Opus, and Haiku. Grok includes code fast, fast reasoning, and fast non reasoning.
Configure Parameters
Depending on the selected model, configure parameters including Temperature, Top P, and Reasoning Level (Fast, Medium, or Deep).
View Connected Functions
Below the parameters, view the list of functions connected to the Assistant. If any functions are linked (e.g., Google Sheets, Google Docs, SEO Analyzer, or Web Crawler), they will be displayed here.
Enable Capabilities
Enable Python Code Interpreter for executing and analyzing Python code, suitable for lightweight statistical analysis. Enable Delaverse Interpreter for advanced statistical analysis on big data. Turn on Execution Capability for direct execution without connecting to a Telegram node (Company ID is activated when enabled). Enable Web Search to provide access to up-to-date information from the internet.
Select Instruction Template
Choose from a dropdown of pre-built instruction templates to quickly set up common use cases.
Manage Conversation Thread
Use the Clear Thread Conversations button when you want the Assistant to no longer access previous conversations for its tasks, effectively clearing the thread.
Enter Custom Instructions
Use the Instructions Box to enter custom prompts tailored to your specific needs.
Save or Delete
Use the Save Changes button to store the configured settings, or the Delete Node button to remove the node entirely.
Instruction Components
When crafting instructions for the AI Assistant, consider including these key elements for optimal results:
Data Sources
Specify inputs (e.g., Google Sheet ID, Request API).
Task Definition
Describe goal (e.g., analyze customer data for RFM, generate SMS).
Output Format
Define structure (e.g., JSON for SMS, text for Telegram).
Tone and Language
Specify language and tone (e.g., Persian with a professional or friendly tone).
Pro Tips
- Start with pre-built templates and customize as needed.
- Provide detailed instructions for better outputs.
- Test with sample data before production.
- Use structured formats (e.g., JSON) for easy integration.