This project demonstrates how to create a simple AI agent using GPT models and speech recognition. The agent listens to user input via microphone, transcribes the speech to text, and generates a response using an LLM (Large Language Model).

• Voice Input: Prompts the user for information using the microphone.
• Speech Recognition: Uses Google's speech recognition API to transcribe spoken input.
• LLM Integration: Sends transcribed text to a GPT model (via LiteLLM) and returns a generated response.
• Logging: Logs function calls, return values, and exceptions to a rotating log file.
• Environment Configuration: Uses a .env file for API keys and log level.

git clone https://github.com/yourusername/agentic-ai.git
cd agentic-ai

Make sure you have Homebrew installed.

brew install python@3.11
brew install portaudio
python3.11 -m venv venv
. venv/bin/activate
pip install -r requirements.txt

Copy the sample file and adjust it to match the LLM you want to use:

cp .env.example .env

Key variables:

• LOGLEVEL – default logging level.
• LLM_NAME – LiteLLM model identifier (e.g., openai/gpt-4o-mini, gemini/gemini-1.5-pro, ollama/mistral:7b).
• LLM_PROVIDER – provider hint (openai, gemini, ollama, etc.).
• LLM_API_KEY – API key or token (use ollama for local Ollama).
• LLM_API_BASE – override base URL (set to http://localhost:11434 for Ollama, blank for OpenAI).
• LLM_TEMPERATURE, LLM_TIMEOUT – default generation settings.
• OPENAI_API_KEY – optional compatibility key for tooling that still expects it.

Any of these can be overridden at runtime with CLI flags such as --model, --provider, --api-key, --api-base, --temperature, --who, --question, or --stream.

./agent.py --model ollama/mistral:7b --provider ollama --stream

If you already populated .env, flags are optional. Use ./agent.py --help to see the full list of overrides.

venv/bin/python -m unittest discover -s tests
venv/bin/python -m pyright

• agent.py — CLI entry point that wires modules together.
• cli.py — Argument parser definition.
• settings.py — Environment/CLI configuration loader.
• llm_client.py — LiteLLM wrapper responsible for routing and streaming output.
• speech_service.py — Microphone capture and transcription logic.
• utils.py — Utility functions and decorators (e.g., logging).
• logging_config.py — Centralized logging configuration.
• graph.py — Example DFS helper used by the exercises/tests.
• .env — Environment variables (not committed to git).
• log/ — Directory for log files (ignored by git).
• requirements.txt — Python dependencies.

• Python 3.11+
• PortAudio (for microphone input)
• See requirements.txt for Python packages.

• Microphone not working: Ensure your microphone is connected and accessible. You may need to grant microphone permissions.
• No response from LLM: Check your .env file for a valid OPENAI_API_KEY.
• Logs not appearing: Ensure the log/ directory exists and is writable.

• API Keys: Never commit your .env file or API keys to version control.
• Logs: Log files may contain sensitive information. Handle them appropriately.

Further improvements and features will be added. Contributions are welcome!