Agent quickstart

This quickstart guide walks you through setting up and running the TEN Agent to experiment with different AI models, speech services, and communication tools. Using the interactive TEN Agent playground you can test and configure AI agents to:

• Experiment with different AI models, TTS/STT systems, and real-time communication
• Test agent behavior in a controlled environment
• Configure extensions and modules through a visual interface

Before starting, ensure you have:

• Agora credentials from Agora Console

  • App ID
  • App Certificate

• API keys for your chosen services

  • LLM service like OpenAI or Google Gemini
  • STT service such as Azure
  • TTS service like ElevenLabs or Azure TTS

• Development environment
  Install one of the following:

  • Docker and Docker Compose (recommended)
  • Node.js(LTS) v18

• System requirements

  • CPU: Dual-core processor or higher
  • Memory: 4 GB RAM minimum

info

If you are using an Apple Silicon Mac, uncheck the Use Rosetta for x86/amd64 emulation checkbox in Docker Desktop settings. While this may slow ARM builds, it ensures proper performance on x64 deployment servers.

Start by getting the code and configuring the basic environment with your credentials.

_1

git clone https://github.com/TEN-framework/ten-framework.git

1. Switch to the project directory and create a .env file from the template:

   
   

   _2

   cd ten-framework/ai_agents

   _2

   cp ./.env.example ./.env

   
   

2. Open .env and add your Agora credentials:

   
   

   _2

   AGORA_APP_ID=your_app_id_here

   _2

   AGORA_APP_CERTIFICATE=your_certificate_here

   
   

3. Add API keys and other configuration parameters to the .env file for the LLM, STT, and TTS extensions you want to test. Following are some commonly used parameters:

   
   

   _23

   # Extension: gemini_llm

   _23

   # Gemini API key

   _23

   GEMINI_API_KEY=

   _23

   _23

   # Extension: openai_chatgpt

   _23

   # OpenAI API key

   _23

   OPENAI_API_BASE=https://api.openai.com/v1

   _23

   OPENAI_API_KEY=

   _23

   OPENAI_MODEL=gpt-4o

   _23

   _23

   # Extension: agora_rtc

   _23

   # Azure STT key and region

   _23

   AZURE_STT_KEY=

   _23

   AZURE_STT_REGION=

   _23

   _23

   # Extension: azure_tts

   _23

   # Azure TTS key and region

   _23

   AZURE_TTS_KEY=

   _23

   AZURE_TTS_REGION=

   _23

   _23

   # Extension: elevenlabs_tts

   _23

   # ElevenLabs TTS key

   _23

   ELEVENLABS_TTS_KEY=

   
   
   info

   You can also configure these API keys later through the playground interface.

Run the docker command to compose containers:

_1

docker compose up -d

Follow these steps to build and run TEN Agent:

1. Enter the development container

   
   

   _1

   docker exec -it ten_agent_dev bash

   
   

2. Build the agent

   Use the following command to build the agent:

   
   

   _1

   task use

   
   

3. Start the web server

   
   

   _1

   task run

   
   

The playground interface allows you to test different configurations and see results in real-time. Each graph type supports specific extensions and capabilities - Voice Agent for speech interactions, Realtime Agent for low-latency communication.

To get started:

1. Open the playground at http://localhost:3000

2. Configure your agent:

   1. Select a graph type (Voice Agent or Realtime Agent) from the dropdown.

   2. Click the module configuration button. Select extension modules such as STT, TTS, and LLM from the panel. The system saves settings to property.json when you click Save changes. For further details, see Configure modules.

   3. Click the settings button to configure extensions with required API keys and parameters. For further details, see Configure properties.

   4. To configure STT and TTS languages, see Configure languages.

You can edit the configuration files directly if you prefer manual control. Open agents/property.json, locate the graph you want to modify, and update any values. After saving your changes, refresh the playground page and reconnect to the agent. Your changes take effect immediately.

This section contains content that completes the information on this page, or points you to documentation that explains other aspects to this product.