Multi-Agent Systems: Build teams of agents that work together towards a common goal, with industry-leading architecture for reasoning, collaboration, and coordination. Create deterministic, stateful workflows with pure Python for maximum flexibility and control
Interactive Playground: Test and interact with your AI agents through an intuitive interface featuring real-time streaming, session history, user memory, multimodal support, and comprehensive configuration options
Monitoring & Debugging: Monitor your agents, teams, and workflows in real-time with built-in debugging capabilities, session tracking, and detailed logs for system prompts, user messages, and tool calls
Create your Agno agents with TrueFoundry Gateway configuration:
Copy
Ask AI
from agno.agent import Agentfrom agno.models.openai import OpenAIChat# Configure agent with TrueFoundry Gatewayagent = Agent( model=OpenAIChat( id="openai-main/gpt-4o", # Use TrueFoundry model name. Similarly you can call any model from any model provider like anthropic, gemini etc api_key="your-truefoundry-api-key", base_url="your-truefoundry-gateway-url" ), description="AI assistant powered by TrueFoundry Gateway", instructions=[ "You are a helpful AI assistant", "Provide accurate and concise responses" ])
Create a simple agent using the configured TrueFoundry Gateway:
Copy
Ask AI
from agno.agent import Agentfrom agno.models.openai import OpenAIChat# Single agent with TrueFoundry Gatewayagent = Agent( model=OpenAIChat( id="openai-main/gpt-4o", api_key="your-truefoundry-api-key", base_url="your-truefoundry-gateway-url" ), name="Assistant", description="General purpose AI assistant")# Run the agentresponse = agent.run("What is the capital of Brazil?")print(response.content)
For persistent configuration across all Agno agents, set these environment variables:
Copy
Ask AI
# Add to your ~/.bashrc, ~/.zshrc, or equivalentexport TRUEFOUNDRY_API_KEY="your-truefoundry-api-key"export TRUEFOUNDRY_BASE_URL="your-truefoundry-gateway-url"# Optional: Set default modelexport TRUEFOUNDRY_MODEL="openai-main/gpt-4o"