Fusion AI

Documentation

What is Fusion AI?Key Benefits How NeuroSwitch Works Data Flow

Tool Calling Framework

Extend AI capabilities beyond text generation with a powerful, extensible tool system that enables real-world interactions and complex task execution.

Extended Capabilities

Enable AI models to perform actions beyond text generation - file creation, code execution, web interactions, and more.

Secure Execution

All tools run in controlled environments with proper validation and security measures to ensure safe operation.

Fully Extensible

Build custom tools for your specific needs using our standardized framework and integration patterns.

How Tool Calling Works

Tool Request

AI model identifies need for external functionality

LLM analyzes the user request and determines which tool is needed

Parameter Extraction

Model specifies tool name and required parameters

AI formats the tool call with proper input parameters

Tool Execution

Fusion backend validates and executes the tool

System runs the tool with provided parameters in secure environment

Result Integration

Tool output is returned to the model for final response

AI incorporates tool results into comprehensive user response

Example Tool Call Flow

User: "Create a Python script that analyzes this CSV data"

AI: Identifies need for file creation and code generation

Tool Call: filecreatortool with Python code content

Result: File created successfully with data analysis script

Response: "I've created a Python script for your CSV analysis..."

Built-in Tools Library

File Creator

filecreatortool

Available

Generate and save files with AI-generated content

Capabilities

Create text files

Generate code files

Save content to disk

Example Use Case

Create a Python script for data analysis

Code Interpreter

e2bcodetool

Available

Execute Python code in a secure sandboxed environment

Capabilities

Capabilities

Code analysis

Style checking

Error detection

Example Use Case

Validate Python code quality

Implementation Example

API Request with Tools

const response = await fetch('/api/chat', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'Authorization': 'Bearer your-api-key'
  },
  body: JSON.stringify({
    message: "Create a Python script to analyze sales data",
    model: "claude",
    enable_tools: true,
    tools: [
      "filecreatortool",
      "e2bcodetool"
    ]
  })
});

Tool Call Response

{
  "response": "I'll create a Python script for sales analysis...",
  "tool_calls": [
    {
      "tool_name": "filecreatortool",
      "parameters": {
        "filename": "sales_analysis.py",
        "content": "import pandas as pd\n..."
      },
      "result": "File created successfully"
    }
  ],
  "model": "claude"
}

Building Custom Tools

Tool Structure

Inherit from BaseTool class
Define tool name and description
Specify input parameters schema
Implement execution logic
Handle errors and validation
Return structured results

Best Practices

🔒 Security

Always validate inputs, use secure execution environments, and limit access to sensitive resources.

⚡ Performance

Keep tools lightweight, implement timeouts, and provide clear error messages for better UX.

📖 Documentation

Document parameters clearly, provide usage examples, and specify expected output formats.

Related Features

Streaming Support

See tool execution results in real-time with streaming responses

Web Search Integration

Access real-time information through integrated search tools

Image & PDF Input

Process visual content with specialized analysis tools

Start Using Tools Today

Enable powerful tool calling in your AI applications with just a single parameter. Extend capabilities beyond text generation.

Get Started API Documentation