AI Agent Overview

Configure Large Language Model

Choose the right model for your needs. Momen provides GPT family, Gemini family, or you can integrate your own models. You can use 100,000 AI points gifted monthly with Momen hosted models.

Momen Hosted Models

• GPT family
• Gemini family

Bring Your Own Model

Supports OpenAI official platform and self-hosted models (like DeepSeek, Gemini, etc.). Configuration requires:

• Model name and type
• API key or server URL
• Authentication details

Configure Input Parameters

Set up parameters required for AI requests.

• Text-based inputs can be used as variables in prompt template and contexts
• Image and Video inputs will be automatically appended to the end of the prompt.

Write Prompts

Prompts define the AI’s role, task requirements, and constraints. Momen AI provides a structured framework:

Prompt Source Options

1. Direct Input: For fixed tasks
2. Parameter Reference: For dynamic content, more flexible
3. Hybrid Approach: Fixed roles + dynamic content
4. Contexts Integration: Retrieve data from database, APIs, or files

Structured Framework

• Task Role: Define AI’s expertise and identity
• Task Details: Specify how AI should perform
• Task Constraints: Set output format, style, and limitations

Configure Contexts

Contexts provide AI with relevant external information to improve response accuracy.

Data Sources

• Database: Select tables and fields from Momen database
• API: Connect to third-party data sources

Configure Tools

Tools enable AI to perform specific actions like API calls, workflow execution, etc.

Available Tool Types

• Action Flow: Execute complex business logic
• API Calls: Connect to external services
• AI Chaining: Invoke other AI models
• User Interaction: Request additional information

Configure Output

Choose how AI results are formatted:

Output Types

• Text: AI automatically determines best format,Streaming is supported
• Structured: Define structured output schema

Test and Debug

Use the right panel debug interface to test your AI configuration:

• Input test parameters
• Verify tool integrations
• Check output formats
• Monitor performance metrics

Run AI Agent

Momen AI provides 4 core actions:

1. Start Conversation: Initiate new AI conversations or single requests.

2. Continue Conversation: Continue existing conversations with context preservation. You need to bind previous conversation ID to the action.

3. Delete Conversation: Remove conversation records and related data.

4. Stop Response: Interrupt ongoing AI responses.

You can run AI agent with 2 ways:

1. In frontend: Add AI agent action to a target component to start a conversation.
2. In backend: Add AI agent node to an Actionflow to run in a flow.

Result Handling

Text mode

• Create a TEXT type page variable to store the result.
• Set the variable to the result data/content on success.

Structured mode

• Create a OBJECT type page variable to store the result.
• Set the variable to the result data/content on success.