Add agent framework integration examples (#3898)

* Added agent framework examples

* Rename

* Update ZenML logo image path in README.md

* Add OPENAI_API_KEY export to all agent frameworks.- Set OPENAI_API_KEY for each agent framework.- Add export command to set the API key

* Add production vs. demos explanation in README.md

* Update user guide examples and table with new agent data

* Update docs/book/user-guide/toc.md

Co-authored-by: Alex Strick van Linschoten <[email protected]>

* Update examples/agent_framework_integrations/README.md

Co-authored-by: Alex Strick van Linschoten <[email protected]>

---------

Co-authored-by: Alex Strick van Linschoten <[email protected]>
(cherry picked from commit 653cfd1)

Author: htahir1

docs/book/user-guide/README.md

@@ -24,4 +24,4 @@ Complete end-to-end implementations that showcase ZenML in real-world scenarios.
 Focused code snippets and templates that address specific ML workflow challenges.\
 [See all examples in GitHub →](https://github.com/zenml-io/zenml-projects)
 
-<table data-view="cards"><thead><tr><th></th><th></th><th data-hidden data-card-cover data-type="files"></th><th data-hidden data-card-target data-type="content-ref"></th></tr></thead><tbody><tr><td><strong>Quickstart</strong></td><td>Bridging Local Development and Cloud Deployment</td><td><a href=".gitbook/assets/example-01.png">example-01.png</a></td><td><a href="https://github.com/zenml-io/zenml/blob/main/examples/quickstart">https://github.com/zenml-io/zenml/blob/main/examples/quickstart</a></td></tr><tr><td><strong>End-to-End Batch Inference</strong></td><td>Supervised ML project built with the ZenML framework and its integration.</td><td><a href=".gitbook/assets/example-02.png">example-02.png</a></td><td><a href="https://github.com/zenml-io/zenml/tree/main/examples/e2e">https://github.com/zenml-io/zenml/tree/main/examples/e2e</a></td></tr><tr><td><strong>Basic NLP with BERT</strong></td><td>Build NLP models with production-ready ML pipeline framework</td><td><a href=".gitbook/assets/example-03.png">example-03.png</a></td><td><a href="https://github.com/zenml-io/zenml/tree/main/examples/e2e_nlp">https://github.com/zenml-io/zenml/tree/main/examples/e2e_nlp</a></td></tr><tr><td><strong>Computer Vision with YoloV8</strong></td><td>End-to-end computer vision pipeline with modular design</td><td><a href=".gitbook/assets/example-04.png">example-04.png</a></td><td><a href="https://github.com/zenml-io/zenml-projects/tree/main/end-to-end-computer-vision">https://github.com/zenml-io/zenml-projects/tree/main/end-to-end-computer-vision</a></td></tr><tr><td><strong>LLM Finetuning</strong></td><td>LLM fine-tuning pipeline with PEFT approach</td><td><a href=".gitbook/assets/example-05.png">example-05.png</a></td><td><a href="https://github.com/zenml-io/zenml/tree/main/examples/llm_finetuning">https://github.com/zenml-io/zenml/tree/main/examples/llm_finetuning</a></td></tr><tr><td><strong>Agent Architecture Comparison</strong></td><td>Compare AI agents with LangGraph workflows, LiteLLM integration, and automatic visualizations.</td><td><a href=".gitbook/assets/example-06.png">example-06.png</a></td><td><a href="https://github.com/zenml-io/zenml/blob/main/examples/agent_comparison">https://github.com/zenml-io/zenml/blob/main/examples/agent_comparison</a></td></tr><tr><td><strong>Minimal Agent Production</strong></td><td>Document analysis service with pipelines, evaluation, and a simple web UI.</td><td><a href=".gitbook/assets/example-07.png">example-07.png</a></td><td><a href="https://github.com/zenml-io/zenml/blob/main/examples/minimal_agent_production">https://github.com/zenml-io/zenml/blob/main/examples/minimal_agent_production</a></td></tr></tbody></table>
+<table data-view="cards"><thead><tr><th></th><th></th><th data-hidden data-card-cover data-type="files"></th><th data-hidden data-card-target data-type="content-ref"></th></tr></thead><tbody><tr><td><strong>Quickstart</strong></td><td>Bridging Local Development and Cloud Deployment</td><td><a href=".gitbook/assets/example-01.png">example-01.png</a></td><td><a href="https://github.com/zenml-io/zenml/blob/main/examples/quickstart">https://github.com/zenml-io/zenml/blob/main/examples/quickstart</a></td></tr><tr><td><strong>End-to-End Batch Inference</strong></td><td>Supervised ML project built with the ZenML framework and its integration.</td><td><a href=".gitbook/assets/example-02.png">example-02.png</a></td><td><a href="https://github.com/zenml-io/zenml/tree/main/examples/e2e">https://github.com/zenml-io/zenml/tree/main/examples/e2e</a></td></tr><tr><td><strong>Agent Architecture Comparison</strong></td><td>Compare AI agents with LangGraph workflows, LiteLLM integration, and automatic visualizations.</td><td><a href=".gitbook/assets/example-06.png">example-06.png</a></td><td><a href="https://github.com/zenml-io/zenml/blob/main/examples/agent_comparison">https://github.com/zenml-io/zenml/blob/main/examples/agent_comparison</a></td></tr><tr><td><strong>Agent Framework Integrations</strong></td><td>Production-ready integrations for 11 popular agent frameworks including LangChain, CrewAI, AutoGen, and more.</td><td><a href=".gitbook/assets/example-06.png">example-06.png</a></td><td><a href="https://github.com/zenml-io/zenml/tree/main/examples/agent_framework_integrations">https://github.com/zenml-io/zenml/tree/main/examples/agent_framework_integrations</a></td></tr><tr><td><strong>Minimal Agent Production</strong></td><td>Document analysis service with pipelines, evaluation, and a simple web UI.</td><td><a href=".gitbook/assets/example-07.png">example-07.png</a></td><td><a href="https://github.com/zenml-io/zenml/blob/main/examples/minimal_agent_production">https://github.com/zenml-io/zenml/blob/main/examples/minimal_agent_production</a></td></tr><tr><td><strong>Basic NLP with BERT</strong></td><td>Build NLP models with production-ready ML pipeline framework</td><td><a href=".gitbook/assets/example-03.png">example-03.png</a></td><td><a href="https://github.com/zenml-io/zenml/tree/main/examples/e2e_nlp">https://github.com/zenml-io/zenml/tree/main/examples/e2e_nlp</a></td></tr><tr><td><strong>Computer Vision with YoloV8</strong></td><td>End-to-end computer vision pipeline with modular design</td><td><a href=".gitbook/assets/example-04.png">example-04.png</a></td><td><a href="https://github.com/zenml-io/zenml-projects/tree/main/end-to-end-computer-vision">https://github.com/zenml-io/zenml-projects/tree/main/end-to-end-computer-vision</a></td></tr><tr><td><strong>LLM Finetuning</strong></td><td>LLM fine-tuning pipeline with PEFT approach</td><td><a href=".gitbook/assets/example-05.png">example-05.png</a></td><td><a href="https://github.com/zenml-io/zenml/tree/main/examples/llm_finetuning">https://github.com/zenml-io/zenml/tree/main/examples/llm_finetuning</a></td></tr></tbody></table>

docs/book/user-guide/toc.md

@@ -76,8 +76,11 @@
 ## Examples
 
 * [Quickstart](https://github.com/zenml-io/zenml/blob/main/examples/quickstart)
+* [Agent Quickstart](https://github.com/zenml-io/zenml/tree/main/examples/minimal_agent_production)
+* [Agent examples](https://github.com/zenml-io/zenml/tree/main/examples/agent_framework_integrations/)
 * [End-to-End Batch Inference](https://github.com/zenml-io/zenml/tree/main/examples/e2e)
-* [Basic NLP with BERT](https://github.com/zenml-io/zenml/tree/main/examples/e2e_nlp)
+* [Forecasting time-series prediction](https://github.com/zenml-io/zenml-projects/tree/main/floracasts)
+* [ML classification](https://github.com/zenml-io/zenml-projects/tree/main/oncoclear)
 * [Computer Vision with YoloV8](https://github.com/zenml-io/zenml-projects/tree/main/end-to-end-computer-vision)
-* [LLM Finetuning](https://github.com/zenml-io/zenml/tree/main/examples/llm_finetuning)
+* [Deep research agentic workflow](https://github.com/zenml-io/zenml-projects/tree/main/deep_research)
 * [More Projects...](https://github.com/zenml-io/zenml-projects)

examples/agent_framework_integrations/README.md

@@ -0,0 +1,197 @@
+<br />
+<div align="center">
+  <a href="https://zenml.io">
+    <img src="../../docs/book/.gitbook/assets/header.png" alt="ZenML Logo" width="600">
+  </a>
+
+<h3 align="center">Agent Frameworks Integration Examples</h3>
+
+  <p align="center">
+    Production-ready agent orchestration with ZenML
+    <br />
+    <a href="https://zenml.io/features">Features</a>
+    ·
+    <a href="https://zenml.io/roadmap">Roadmap</a>
+    ·
+    <a href="https://github.com/zenml-io/zenml/issues">Report Bug</a>
+    ·
+    <a href="https://zenml.io/discussion">Vote New Features</a>
+    ·
+    <a href="https://blog.zenml.io/">Read Blog</a>
+    <br />
+    <br /> 
+    <a href="https://zenml.io/slack">
+    <img src="https://img.shields.io/badge/JOIN US ON SLACK-4A154B?style=for-the-badge&logo=slack&logoColor=white" alt="Slack">
+    </a>
+    <a href="https://www.linkedin.com/company/zenml/">
+    <img src="https://img.shields.io/badge/LinkedIn-0077B5?style=for-the-badge&logo=linkedin&logoColor=white" alt="LinkedIn">
+    </a>
+    <a href="https://twitter.com/zenml_io">
+    <img src="https://img.shields.io/badge/Twitter-1DA1F2?style=for-the-badge&logo=twitter&logoColor=white" alt="Twitter">
+    </a>
+  </p>
+</div>
+
+# 🤖 Agent Frameworks + ZenML
+
+This collection demonstrates how to integrate popular agent frameworks with ZenML for production-grade AI agent orchestration. Each example follows consistent patterns and best practices, making it easy to adapt any framework for your specific use case.
+
+## 🚀 Quick Start
+
+Choose any framework and get started in minutes:
+
+```bash
+export OPENAI_API_KEY="your-api-key-here"
+cd framework-name/
+uv venv --python 3.11
+source .venv/bin/activate  
+uv pip install -r requirements.txt
+python run.py
+```
+
+## 📊 Frameworks Overview
+
+| Framework | Type | Key Features | Technologies |
+|-----------|------|-------------|-------------|
+| [Autogen](autogen/) | 🤝 Multi-Agent | Multi-agent conversations, Role-based collaboration | autogen, openai |
+| [AWS Strands](aws-strands/) | ⚡ Simple | Direct agent calls, Built-in tools | aws-agents, bedrock |
+| [CrewAI](crewai/) | 👥 Crews | Agent crews, Task delegation | crewai, openai |
+| [Google ADK](google-adk/) | 🧠 Gemini | Google AI agents, Gemini models | google-adk, gemini |
+| [Haystack](haystack/) | 🔍 RAG | Retrieval pipelines, Document processing | haystack, openai |
+| [LangChain](langchain/) | 🔗 Chains | Runnable chains, Tool composition | langchain, openai |
+| [LangGraph](langgraph/) | 🕸️ Graphs | ReAct agents, Graph workflows | langgraph, openai |
+| [LlamaIndex](llama_index/) | 📚 Functions | Function agents, Async execution | llama-index, openai |
+| [OpenAI Agents SDK](openai_agents_sdk/) | 🏗️ Structured | Official OpenAI agents, Structured execution | openai-agents, openai |
+| [PydanticAI](pydanticai/) | ✅ Type-Safe | Type-safe agents, Validation | pydantic-ai, openai |
+| [Semantic Kernel](semantic-kernel/) | 🧩 Plugins | Plugin architecture, Microsoft ecosystem | semantic-kernel, openai |
+
+## 🎯 Core Patterns
+
+All examples follow these established patterns:
+
+### 🔧 Environment Setup
+- **uv for environments**: `uv venv --python 3.11`
+- **Fast installs**: `uv pip install -r requirements.txt`
+- **Consistent Python version**: 3.11 across all frameworks
+
+### 📦 Pipeline Architecture
+```python
+@pipeline
+def agent_pipeline() -> str:
+    # 1. External artifact input
+    query = ExternalArtifact(value="Your query")
+    
+    # 2. Agent execution
+    results = run_agent(query)
+    
+    # 3. Response formatting
+    summary = format_response(results)
+    
+    return summary
+```
+
+### 🛡️ Error Handling
+- Comprehensive try-catch blocks
+- Status tracking with success/error states
+- Graceful degradation for agent failures
+
+### 📊 Artifact Management
+- **Annotated outputs**: `Annotated[Type, "artifact_name"]`
+- **ZenML integration**: Full pipeline orchestration
+- **Artifact storage**: S3-backed artifact management
+
+## 🏗️ Framework-Specific Features
+
+### Multi-Agent Systems
+- **Autogen**: Conversational multi-agent workflows
+- **CrewAI**: Role-based agent crews with task delegation
+
+### Simple Execution
+- **AWS Strands**: Direct callable interface with `agent(query)`
+- **PydanticAI**: Clean `agent.run_sync(query)` API
+- **Google ADK**: Gemini-powered agents with simple calls
+
+### Advanced Orchestration
+- **LangChain**: Composable chains with tool integration
+- **LangGraph**: ReAct pattern with graph-based workflows
+- **Semantic Kernel**: Plugin-based architecture
+
+### Specialized Use Cases
+- **Haystack**: RAG pipelines with retrieval components
+- **LlamaIndex**: Function agents with async capabilities
+- **OpenAI Agents SDK**: Structured execution with OpenAI
+
+## 🔄 Implementation Notes
+
+### Production vs. Demos
+**These examples demonstrate single-query execution for simplicity.** In production, ZenML's value comes from:
+- **Batch processing**: Process hundreds/thousands of queries overnight
+- **Agent evaluation**: Compare different frameworks on test datasets  
+- **Data pipelines**: Use agents to process document collections
+- **A/B testing**: Systematic comparison of agent configurations
+
+For real-time serving, use FastAPI/Flask directly. Use ZenML for the operational layer.
+
+### Async Frameworks
+Some frameworks require async handling within ZenML steps:
+- **LlamaIndex**: `asyncio.run(agent.run(query))`
+- **Semantic Kernel**: Event loop management for chat completion
+
+### Tool Integration
+Different frameworks have varying tool patterns:
+- **Decorators**: `@tool`, `@function_tool`, `@kernel_function`
+- **Functions**: Regular Python functions as tools
+- **Classes**: Tool classes with specific interfaces
+
+### Response Extraction
+Each framework returns different response types:
+- **String responses**: Direct text output
+- **Object responses**: `.output`, `.content`, `.final_output` attributes
+- **Complex responses**: Nested structures requiring extraction
+
+## 📋 Requirements
+
+- **Python**: 3.11+
+- **ZenML**: Latest version
+- **UV**: For fast package management
+- **OpenAI API Key**: Most examples use OpenAI (set `OPENAI_API_KEY`)
+
+## 🆘 Getting Help
+
+- 💬 [Join our Slack community](https://zenml.io/slack)
+- 📖 [Check our documentation](https://docs.zenml.io/)
+- 🐛 [Report issues](https://github.com/zenml-io/zenml/issues)
+- 💡 [Request features](https://zenml.io/discussion)
+
+## 🌟 About ZenML
+
+ZenML is an extensible, open-source MLOps framework for creating production-ready ML pipelines. These agent framework integrations showcase ZenML's flexibility in orchestrating AI workflows beyond traditional ML use cases.
+
+**Why ZenML for Agent Orchestration?**
+- 🔄 **Reproducible workflows**: Version and track agent executions
+- 📊 **Artifact management**: Store and version agent inputs/outputs
+- 🎯 **Production ready**: Built-in monitoring, logging, and error handling
+- 🔧 **Tool agnostic**: Works with any agent framework
+- ☁️ **Cloud native**: Deploy anywhere with consistent behavior
+
+## 📖 Learn More
+
+| Resource | Description |
+|----------|-------------|
+| 🧘 **[ZenML 101]** | New to ZenML? Start here! |
+| ⚛ **[Core Concepts]** | Understand ZenML fundamentals |
+| 🤖 **[LLMOps Guide]** | Complete guide to LLMOps with ZenML |
+| 📓 **[Documentation]** | Full ZenML documentation |
+| 📒 **[API Reference]** | Detailed API documentation |
+| ⚽ **[Examples]** | More ZenML examples |
+
+[ZenML 101]: https://docs.zenml.io/user-guides/starter-guide
+[Core Concepts]: https://docs.zenml.io/getting-started/core-concepts
+[LLMOps Guide]: https://docs.zenml.io/user-guides/llmops-guide
+[Documentation]: https://docs.zenml.io/
+[SDK Reference]: https://sdkdocs.zenml.io/
+[Examples]: https://github.com/zenml-io/zenml/tree/main/examples
+
+---
+
+*This collection demonstrates the power and flexibility of ZenML for orchestrating diverse agent frameworks in production environments.*

examples/agent_framework_integrations/autogen/README.md

@@ -0,0 +1,17 @@
+# Autogen + ZenML
+
+Multi-agent conversation framework integrated with ZenML.
+
+## Run
+```bash
+export OPENAI_API_KEY="your-api-key-here"
+uv venv --python 3.11
+source .venv/bin/activate
+uv pip install -r requirements.txt
+python run.py
+```
+
+## Features
+- Multi-agent conversations with UserProxy and Assistant agents
+- Async runtime management with proper cleanup
+- Travel planning use case with collaborative agents

examples/agent_framework_integrations/autogen/requirements.txt

@@ -0,0 +1,3 @@
+autogen-core>=0.4.0
+autogen-ext[openai]>=0.4.0
+zenml

examples/agent_framework_integrations/autogen/run.py

@@ -0,0 +1,118 @@
+"""ZenML Pipeline for Autogen Multi-Agent Travel Assistant.
+
+This pipeline encapsulates the autogen multi-agent system in a ZenML pipeline,
+demonstrating how to integrate agent frameworks with ZenML for orchestration
+and artifact management.
+"""
+
+import asyncio
+from typing import Annotated, Any, Dict
+
+from autogen_agent import TravelQuery, setup_runtime
+from autogen_core import AgentId
+
+from zenml import ExternalArtifact, pipeline, step
+
+
+@step
+def run_autogen_agents(
+    query: str,
+) -> Annotated[Dict[str, Any], "agent_results"]:
+    """Execute the autogen multi-agent system and return results."""
+    # Parse destination and days from query string
+    # Format: "Plan a X-day trip to DESTINATION"
+    parts = query.split(" to ")
+    destination = parts[1] if len(parts) > 1 else "Unknown"
+    days_part = [w for w in query.split() if w.endswith("-day")][0]
+    days = int(days_part.replace("-day", ""))
+
+    # Create TravelQuery for the agents
+    travel_query = TravelQuery(destination=destination, days=days)
+
+    async def _run_agents():
+        # Setup the autogen runtime
+        runtime = await setup_runtime()
+
+        # Start the runtime
+        runtime.start()
+
+        try:
+            # Send the query to the coordinator agent
+            result = await runtime.send_message(
+                travel_query, AgentId("coordinator", "default")
+            )
+            return result
+
+        finally:
+            # Always stop the runtime
+            await runtime.stop()
+
+    # Run the async function
+    travel_plan = asyncio.run(_run_agents())
+
+    # Convert to dict for ZenML artifact storage
+    return {
+        "destination": travel_plan.destination,
+        "days": travel_plan.days,
+        "weather": {
+            "destination": travel_plan.weather.destination,
+            "forecast": travel_plan.weather.forecast,
+            "temperature": travel_plan.weather.temperature,
+        },
+        "attractions": {
+            "destination": travel_plan.attractions.destination,
+            "attractions": travel_plan.attractions.attractions,
+        },
+    }
+
+
+@step
+def format_travel_plan(
+    plan_data: Dict[str, Any],
+) -> Annotated[str, "formatted_plan"]:
+    """Format the travel plan into a readable summary."""
+    weather = plan_data["weather"]
+    attractions = plan_data["attractions"]
+
+    summary = f"""🏖️ TRAVEL PLAN FOR {plan_data["destination"].upper()}
+{"=" * 50}
+
+📅 Duration: {plan_data["days"]} days
+
+🌤️ Weather Forecast:
+   • Condition: {weather["forecast"]}
+   • Temperature: {weather["temperature"]}
+
+🏛️ Top Attractions:
+"""
+
+    for i, attraction in enumerate(attractions["attractions"], 1):
+        summary += f"   {i}. {attraction}\n"
+
+    return summary.strip()
+
+
+@pipeline
+def autogen_travel_pipeline() -> str:
+    """ZenML pipeline that orchestrates the autogen multi-agent travel system.
+
+    Returns:
+        Formatted travel plan summary
+    """
+    # External artifact for travel query
+    travel_query = ExternalArtifact(value="Plan a 4-day trip to Paris")
+
+    # Run the autogen agents
+    plan_data = run_autogen_agents(travel_query)
+
+    # Format the results
+    summary = format_travel_plan(plan_data)
+
+    return summary
+
+
+if __name__ == "__main__":
+    print("🚀 Running autogen travel pipeline...")
+    run_result = autogen_travel_pipeline()
+    print("Pipeline completed successfully!")
+    print("Check the ZenML dashboard for detailed results and artifacts.")

examples/agent_framework_integrations/aws-strands/README.md

@@ -0,0 +1,17 @@
+# AWS Strands + ZenML
+
+AWS Strands agent framework integrated with ZenML.
+
+## Run
+```bash
+export OPENAI_API_KEY="your-api-key-here"
+uv venv --python 3.11
+source .venv/bin/activate
+uv pip install -r requirements.txt
+python run.py
+```
+
+## Features
+- Simple agent execution with callable interface
+- Built-in tools using `@tool` decorator
+- Math calculation capabilities

examples/agent_framework_integrations/aws-strands/requirements.txt

@@ -0,0 +1,3 @@
+strands-agents[openai]>=1.0.0
+strands-agents-tools>=0.0.1
+zenml