AgenticAI App leveraging the new OpenAI Agents SDK and Tool usage

An agentic AI application that leverages the OpenAI Agents SDK to create specialized AI agents for different tasks.

Features

• Agent Specialization: Create specialized agents for different domains (math tutoring, history tutoring, web search, file search)
• Agent Triage: Implement a triage system to route queries to the appropriate specialized agent
• Input Guardrails: Apply guardrails to filter and validate user inputs before processing
• Knowledge Sources: Connect to both local knowledge bases and web search capabilities

Examples

The project includes two main example implementations:

1. Educational Assistant (quickstart_example.py)

This example demonstrates:

• A homework validation guardrail that checks if queries are homework-related
• Specialized tutor agents for math and history domains
• A triage agent that routes homework questions to the appropriate tutor

2. Knowledge Assistant (file_assistant.py)

This example demonstrates:

• A file search agent that queries local knowledge sources
• A web search agent that retrieves information from the internet
• A triage agent that determines whether to use local knowledge or web search

Getting Started

1. Set up your environment variables in a .env file (including the MODEL variable)
2. Run one of the example files:

   python quickstart_example.py
   

   or

   python file_assistant.py
   

Dependencies

• OpenAI Agents SDK
• Pydantic
• Python-dotenv
• AsyncIO