This course teaches developers how to design, build, and deploy agentic AI systems—AI solutions that can plan, reason, and act autonomously using large language models (LLMs), tools, memory, and orchestration frameworks.

Course Outline GH-600T00: Developing in Agentic AI Systems with GitHub CoPilot

Module 1: Introduction to Agentic AI

• What is Agentic AI (vs. traditional AI & copilots)
• Characteristics of AI agents:
  • Autonomy
  • Planning & reasoning
  • Tool usage
• Common architectures (ReAct, Plan-and-Execute, multi-agent systems)
• Use cases:
  • Task automation
  • Decision support
  • Complex workflows

Module 2: Foundations of Large Language Models (LLMs)

• LLM capabilities and limitations
• Prompt engineering fundamentals
• System vs. user prompts
• Token usage and context windows
• Controlling behavior with:
  • Temperature, top-p
  • Stop sequences
• Responsible AI considerations

 Module 3: Designing AI Agents

• Core components of an agent:
  • Planning
  • Memory
  • Tools
  • Execution loop
• Agent design patterns:
  • Single-agent systems
  • Multi-agent collaboration
• Task decomposition strategies
• State management

Module 4: Tool Integration and Function Calling

• Connecting agents to external tools/APIs
• Function/tool calling with LLMs
• Designing tool schemas
• Handling tool responses
• Error handling and retries
• Security considerations when invoking tools

Module 5: Memory and Context Management

• Types of memory:
  • Short-term (conversation context)
  • Long-term (persistent storage)
• Retrieval-Augmented Generation (RAG)
• Embeddings and vector databases
• Context pruning and summarization
• Personalization using memory

Module 6: Planning and Orchestration

• Planning strategies:
  • Chain-of-thought (internal)
  • Explicit planning models
• Execution loops and control flow
• Orchestration frameworks:
  • Semantic Kernel
  • LangChain (conceptual comparison)
• Handling multi-step reasoning tasks

Module 7: Multi-Agent Systems

• When to use multiple agents
• Communication protocols between agents
• Role-based agents:
  • Planner
  • Executor
  • Critic
• Collaboration patterns:
  • Sequential
  • Parallel
  • Hierarchical
• Conflict resolution and coordination

Module 8: Building with Azure AI Services

• Azure OpenAI Service for agentic systems
• Integrating:
  • Azure AI Search
  • Azure Functions / APIs
• Identity and access management
• Deployment architectures on Azure

Module 9: Evaluation and Debugging

• Evaluating agent behavior:
  • Accuracy
  • Reliability
  • Latency
• Testing strategies:
  • Prompt testing
  • Scenario-based evaluation
• Observability:
  • Logging
  • Tracing agent decisions
• Debugging reasoning failures

 Module 10: Responsible and Secure Agent Design

• Risks of autonomous agents:
  • Hallucination
  • Tool misuse
  • Data leakage
• Guardrails:
  • Content filtering
  • Policy enforcement
• Human-in-the-loop design
• Compliance and governance

 Module 11: Deployment and Scaling

• Packaging agent applications
• API-based deployment
• Scaling considerations:
  • Cost optimization
  • Token usage
• Monitoring and maintenance
• CI/CD for AI solutions

Module 12: Capstone Project

• Build an end-to-end agentic system:
  • Example projects:
    • Autonomous research assistant
    • Workflow automation agent
    • Multi-agent problem solver
• Design documentation
• Evaluation and optimization

 Skills Gained

• Build autonomous AI agents using LLMs
• Integrate tools and APIs into agent workflows
• Implement memory and retrieval systems
• Design multi-agent architectures
• Deploy scalable agentic applications on Azure

Prerequisites

• Intermediate Python or C# development
• Basic understanding of REST APIs
• Familiarity with AI/ML concepts (helpful but not required)