Function Calling with Ollama and Web Search

A smart search assistant that combines Ollama's LLM capabilities with web search functionality using function calling. The assistant can decide when to search the web for additional information to answer questions accurately.

Read the article for more details: Tool use and function calling in LLMs

Watch it in Action

Features

• Function calling with Ollama
• Web search integration via SearchAPI
• Smart decision-making on when to use web search
• Automatic summarization of search results

Setup

1. Install dependencies:

pip install ollama python-dotenv requests

2. Create .env file in project root:

SEARCH_API_KEY=your_api_key_here

3. Ensure Ollama is running locally

Project Structure

.
├── .env
├── search_tool.py  # Search functionality and utilities
└── main.py        # Main application logic

Usage

Run the assistant:

python main.py

Example interaction:

What would you like to know? What's the latest research on quantum computing?
Processing your question...
Searching the web...
Answer: [Ollama's response with current information]

Requirements

• Python 3.12+
• Ollama installed and running
• SearchAPI key (get from searchapi.io)