Skip to main content
🐍 Python SDK for AI Agents

Build AI Agents
That Actually Work

Three lines of code. Any LLM provider. Production-ready multi-agent workflows.

pip install openstackai

From Zero to AI Agent in 3 Lines

Simple API

from openstackai import ask, summarize, research

# One-liner AI calls
answer = ask("What is machine learning?")
summary = summarize("Long document text...")
report = research("Latest AI trends 2026")
🤖

Agent API

from openstackai import Agent, Runner

# Create an AI agent
agent = Agent(
name="Assistant",
instructions="You are a helpful AI assistant."
)

# Run synchronously
result = Runner.run_sync(agent, "Hello!")
print(result.final_output)
👥

Multi-Agent

from openstackai import Agent, Runner
from openstackai.blueprint import ChainWorkflow

# Multi-agent workflow
researcher = Agent(name="Researcher", instructions="Research topics")
writer = Agent(name="Writer", instructions="Write articles")

workflow = ChainWorkflow([researcher, writer])
result = Runner.run_sync(workflow, "AI in healthcare")

Everything You Need to Build AI Agents

🎯

Simple by Design

One-liner APIs like ask(), summarize(), research(). No boilerplate, just results.

🔌

Multi-Provider

OpenAI, Azure OpenAI, Anthropic, Ollama, Gemini - switch with one line.

👥

Multi-Agent

Chain, supervisor, and custom orchestration patterns out of the box.

🏢

Enterprise Ready

Azure AD auth, sessions, guardrails, tracing, and evaluation built-in.

🧩

Extensible

Custom skills, OpenAPI tools, A2A protocol, and plugin architecture.

📊

RAG & Vector DBs

ChromaDB, Pinecone, Qdrant, Weaviate - connect your knowledge base.

Any Provider. Your Choice.

🤖OpenAI
☁️Azure
🧠Anthropic
🦙Ollama
🔮ChromaDB
🌲Pinecone
🎯Qdrant
🕸️Weaviate

Ready to Build AI Agents?

Join developers building the next generation of AI applications.