AI Agents

AI Agents That Users Actually Use

Build intelligent agents that solve real problems - from simple chat bots to complex business automation. Deploy to the marketplace where users discover and install your AI solutions.

Build Your First Agent Agent Documentation
2 Types Simple & Advanced
Marketplace Ready
Real Users Install & Use

Agent Types

Choose the right agent architecture for your use case

Custom Python Agents

Build sophisticated agents with full programmatic control and advanced business logic.

from fiberwise_sdk import FiberAgent

class ContentGenerator(FiberAgent):
    def run_agent(self, input_data: dict, 
                   llm_provider: LLMProviderService):
        topic = input_data.get('topic', '')
        result = llm_provider.complete(
            f"Write content about {topic}"
        )
        return {
            'status': 'success',
            'content': result.text,
            'word_count': len(result.text.split())
        }
  • Full programmatic control
  • Custom business logic
  • Service injection support
  • Advanced error handling

LLM Agents (Configuration-Based)

Rapidly deploy AI agents using simple YAML configuration without custom code.

agents:
  - name: Smart Assistant
    type: llm
    provider: openai
    model: gpt-4
    system_prompt: |
      You are a helpful AI assistant that 
      provides accurate, concise responses.
    temperature: 0.7
    max_tokens: 500
  • No-code deployment
  • YAML configuration
  • Multiple LLM providers
  • Quick prototyping
  • Complex reasoning pipelines with simple agents

Modern Service Injection Architecture

Automatic dependency resolution with type-hint based injection

advanced_agent.py
Python
def run_agent(self, input_data: dict, 
               llm_provider: LLMProviderService,
               email_service: EmailService, 
               fiber_app: FiberApp):
    """Services are automatically injected by type hints"""
    
    # Process with LLM
    response = llm_provider.complete(input_data['prompt'])
    
    # Send notification
    email_service.send({
        'to': input_data['user_email'],
        'subject': 'Task Complete',
        'body': response.text
    })
    
    # Store in app data
    fiber.data.create_item({
        'user_id': input_data['user_id'],
        'response': response.text,
        'timestamp': self.get_current_time()
    })
    
    return {'status': 'success', 'response': response.text}

Benefits of Service Injection

Automatic Resolution

Services are injected based on type hints - no manual wiring required

Testable Architecture

Easy to mock and test individual components in isolation

Modular Design

Clean separation of concerns with reusable service components

Performance Optimized

Efficient pattern detection and service lifecycle management

Agent Activation System

Powerful execution engine with intelligent lifecycle management

Pattern Detection

Analyzes agent code for service dependencies automatically

→

Service Injection

Provides required services based on detected patterns

→

Lifecycle Management

Handles execution, error handling, and cleanup automatically

→

Performance Monitoring

Tracks execution time and resource usage in real-time

Agent Activation API
Python
# Agent activation via API
result = await fiber.agents.activate('content-generator', {
    'topic': 'AI automation',
    'type': 'blog_post',
    'tone': 'professional',
    'target_audience': 'developers'
})

# Response includes execution metadata
{
    "status": "success",
    "content": "Generated blog post content...",
    "metadata": {
        "execution_time": "2.3s",
        "tokens_used": 1250,
        "services_injected": ["llm_provider", "content_service"],
        "activation_id": "act_123456789"
    }
}

Core Capabilities

Multi-Language SDKs

Build agents in Python or Node.js with consistent APIs and seamless integration across languages.

  • Python SDK with async support
  • Node.js SDK with TypeScript
  • Consistent API patterns
  • Built-in error handling

Dependency Injection

Powerful dependency injection system for complex services and seamless agent coordination.

  • Service container management
  • Automatic dependency resolution
  • Testable architecture
  • Configuration-driven setup

Real-Time Tracking

Monitor agent performance, track activations, and analyze execution patterns in real-time.

  • Activation history and metrics
  • Performance monitoring
  • Error tracking and debugging
  • Usage analytics

Multi-Agent Coordination

Orchestrate multiple agents working together with conversation, chain, and parallel modes.

  • Conversation coordination
  • Sequential chain execution
  • Parallel processing
  • Workflow templates

Agent Development Examples

chat_agent.py
Python
class ChatAgent:
    async def run_agent(self, input_data, fiber, llm_service):
        """AI agent with dependency injection"""
        
        # Get message from input
        message = input_data.get('message', '')
        chat_id = input_data.get('chat_id')
        
        # Generate AI response
        response = await llm_service.generate_completion(
            prompt=message,
            temperature=0.7,
            max_tokens=1000
        )
        
        # Store conversation history
        await fiber.storage.store_chat_message(
            chat_id=chat_id,
            message=message,
            response=response.get("text")
        )
        
        return {
            "status": "success",
            "response": response.get("text"),
            "metadata": {
                "model": response.get("model"),
                "provider": response.get("provider"),
                "tokens_used": response.get("usage", {}).get("total_tokens", 0)
            }
        }

Quick Start Guide

1

Create Agent

Build your agent with the Fiberwise SDK

fiber agent create my-chat-agent
2

Test Locally

Test your agent with sample inputs

fiber agent test my-chat-agent --input "Hello world"
3

Deploy Agent

Deploy to Fiberwise platform

fiber agent deploy my-chat-agent
4

Activate & Use

Activate and start using your agent

fiber activate my-chat-agent