AI Agent Examples is your all-in-one Python starter kit for building, customizing, and deploying real-time AI voice agents using VideoSDK. Leverage the power of OpenAI, Gemini, AWS NovaSonic, and more to create interactive, intelligent assistants for meetings, support, education, and beyond.
To get started, make sure you have:
- A VideoSDK authentication token
- Sign up or log in at VideoSDK Dashboard
- Go to API Keys and generate a new token
- Copy the token and add it to your
.env
file asVIDEOSDK_AUTH_TOKEN
- API keys for LLM providers and Google integrations
- OpenAI: Add your OpenAI API key as
OPENAI_API_KEY
in.env
- Google: Add your Google API key as
GOOGLE_API_KEY
in.env
(required for Gemini and Google function tools) - AWS: Add your AWS credentials as
AWS_ACCESS_KEY_ID
,AWS_SECRET_ACCESS_KEY
, andAWS_DEFAULT_REGION
in.env
(for AWS NovaSonic agents)
- OpenAI: Add your OpenAI API key as
- A VideoSDK meeting ID
- Simply run the provided script:
This will print a new meeting ID in your terminal, ready to use with any agent!
cd PythonSDK python main.py
- Simply run the provided script:
- 🎤 Real-time Voice & Media: Agents can listen, speak, and interact live in meetings.
- 🤖 Multi-Model Support: Integrate with OpenAI, Gemini, AWS NovaSonic, and more.
- 🧩 Pluggable Agent Types: Includes recruiter, tutor, doctor, storyteller, and more.
- 🛠️ Function Tools: Extend agent capabilities with event scheduling, expense tracking, and more (see
fuctionTools/
). - 🔌 Easy Integration: Built on VideoSDK's robust Python SDK.
- 🏗️ Extensible: Add your own agents and tools easily.
- Clone the repository:
git clone https://github.com/videosdk-community/ai-agent-examples.git cd ai-agent-examples
- Create your environment file and set up all API keys:
cp .env.sample .env
- Set up a Python environment and install dependencies:
python3 -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate pip install -r requirements.txt
- Generate a meeting ID:
cd PythonSDK python main.py # Copy the printed meeting ID
- Update the meeting ID in any agent script (e.g.,
basicAgents/recruiter.py
) - Run your chosen agent:
python "basicAgents/recruiter.py" # Or try openai.py, gemini.py, aws.py, etc.
- Join the same meeting from a VideoSDK client app (Web, Mobile, etc.) to interact with your agent in real time.
Agent Type | Description | File |
---|---|---|
🎙️ Basic Voice Agent | Simple voice assistant with different LLM options | openai.py, gemini.py, aws.py |
👩💼 Recruiter | Conducts screening interviews, evaluates communication skills | recruiter.py |
👨🏫 Tutor | Helps students understand academic concepts | tutor.py |
👨⚕️ Doctor | Provides general medical guidance (not medical advice) | doctor.py |
📚 Storyteller | Creates stories from user keywords in real-time | storyteller.py |
👯 Companion | Friendly, empathetic AI for casual chat | companion.py |
🌟 Celebrity | Role-plays as famous personalities | celebrity.py |
🧘 Confession | Nonjudgmental listener for reflection | confession.py |
🛠️ Agents with Function Tools | Agents that can schedule events, track expenses, or take notes using fuctionTools | See fuctionTools/README.md |
- basicAgents: Ready-to-use AI agents for recruiter, tutor, doctor, storyteller, companion, celebrity, and more. Each agent is a Python script you can run directly.
- pythonSDK: Core VideoSDK Python integration, including meeting and participant event handlers. Start here to understand the SDK basics and generate meeting IDs.
- fuctionTools: Utility tools for agents, such as brain dump, event scheduler, and expense tracker. Integrate these with your agents for advanced workflows.
- aws.py, openai.py, gemini.py: Example scripts for running a basic agent with AWS, OpenAI, or Gemini models.
- Production-Ready: Built on a scalable, real-time platform trusted by thousands of developers.
- Flexible: Mix and match LLMs, agent types, and tools for your unique use case.
Pull requests and new agent ideas are welcome!