Skip to content

GitHub-Insight-ANZ-Lab/agentic-aiapp-in-a-day

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 

Repository files navigation

Agentic AI App in a Day (Draft)

Agentic AI App in a Day Cover

Overview

  • Duration: 9am to 4pm
  • Demo: Azure Portal & VS Code
  • Lab: GitHub Codespaces

Goal

  • Build on the Agentic AI aspects
  • Inspire and Motivate
  • Real world ideas for Agentic AI
  • Focused on End-to-end building AI-powered applications
  • Frictionless learning environment, learn via lab, demo and teaching moments
  • Azure AI Protections
  • GitHub Copilot Agent Model to accelerate development
  • For some people without laptops or limited technical skills, allowing them to watch walkthroughs, whilst others do them

Audience

  • Technical teams
  • Senior management like C-levels

Labs

  • Keep lab simple and able to be done end to end
  • For experienced developers, share the accelerator repos and let them explore

Note: some of the labs will be adjusted based on the resources and licences available for the participants to use.

Presentation #1: Morning, Kick-start – 30 min

Topics: Basic Concepts and Overview

  • Azure AI + Agentic AI overview
  • Agentic AI design patterns (Planning, Tool Use, Reflection, Multi-agent)
  • Agent architectures: ReAct, Chain-of-Thought
  • Memory systems for agents (short-term, long-term, episodic)
  • Tool integration patterns and API orchestration
  • Intelligent data layers (Azure AI Search, RAG)
  • Low/No Code option in Copilot Studio (vs pro-code)

Demo #1: AI App Demo - 15min

Showcasing a fun AI App

  • use a Accelerator Repo
  • Daniel's LEGO coding chatbot app demo

Lab #1: AI Foundry Integration & Extension - 50min

Explore AI Foundry services:

  • Use Playground and Model Catalog
  • Various Azure AI Services
  • Customize agent orchestration using no-code / low-code blocks (Functions)
  • Connect additional tools and knowledge bases (e.g., a calculator API, web search API)
  • Connect Agents

Demo #2: Complex AI App Demo - 15min

*Demonstrating Advanced AI Solutions

  • use a Accelerator Repo
  • refer to the repo list in the end

Lab #2: Add Intelligent Data Layer - 50min

Integrate Azure AI Search or Cosmos DB:

  • Ingest doc into Azure AI Search
  • Use Document Intelligence Service
  • Connect agent to knowledge base
  • Demonstrate contextual reasoning
  • Build a memory/retrieval plugin with Semantic Kernel
  • Advance query, score profile, index, indexer in AI Search

Presentation #2: After Lunch – 30min

Topic: Building Production-Ready AI Apps with Agentic Concepts

  • Technical Deep Dive: Advanced reasoning architectures and hybrid - approaches for complex problem-solving
  • Platform-First Strategies: Scaling, API management, DevOps, and platform selection best practices
  • Practical Considerations: Memory, cost, performance optimization - and error handling strategies
  • Enterprise Readiness: Security, evaluation, testing, and compliance frameworks for production deployment
  • Real-World Implementation Patterns: Customer service, document processing, developer tools, and multi-modal applications

Demo #3: GitHub Copilot Coding Assistant for AI App - 15min

Developer experience showcase:

  • Prompt-driven coding assistant (e.g., auto-generate unit tests, fix bugs)
  • Refactoring, adding comments/images, reformatting examples
  • Show how devs stay in control but move faster

Lab #3: AI Agent 101 & Semantic Kernel + MCP - 70min

Build and extend your first agentic AI app:

  • Implement the Prompt → Plan → Act → Reflect loop
  • Use Semantic Kernel for memory, planning, and tool integration
  • Chain outputs between tools and functions
  • Build and integrate MCP for advanced orchestration and extensibility
  • Experiment with adding new skills,

Demo #4: Multi-Agent Solution Walkthrough - 15min

Showcase advanced agent collaboration and orchestration:

  • Demonstrate multi-agent workflows using Azure AI Foundry Services
  • Highlight agent collaboration: perception → planning → action loop
  • Explore coordination strategies and communication between agents
  • Present a real-world scenario (Daniel's agentic LEGO demo) to illustrate practical applications

Lab #4: AI Integration, Deployment & Productionise - 70min

Deploy and operationalize your agentic AI app:

  • Containerize the agent app for portability and scalability
  • Deploy to Azure using ACA, AKS, or App Service with CI/CD
  • Expose APIs securely via Azure API Management (auth, rate limits, monitoring)
  • Implement logging, monitoring, and alerting for production readiness
  • Apply security best practices and cost controls

Presentation #3: Afternoon – 30min

Topic: Wrap-Up: Responsible AI & Future Considerations

  • Responsible AI and Guiderails: Content filtering, prompt injection defense, evaluation, and monitoring
  • Observability & Operations: Agent analytics, performance metrics, cost tracking, and debugging
  • Advanced Topics & Future Trends: Fine-tuning, multi-modal agents, edge deployment, and emerging frameworks
  • Lessons Learned & Best Practices: Common pitfalls, testing strategies, and migration patterns
  • Next Steps & Resources: Certification paths, community support, and building your first production agent

Azure Services to be considered

  • Azure AI Foundry (Model Catalog, Playground) – To explore and test models
  • Azure AI Foundry – Agent Service – For orchestrating multi-agent systems
  • Azure AI Services – For integrating vision, speech, etc.
  • Azure OpenAI – LLMs for chat, planning, coding, summarization
  • Azure AI Search – Intelligent data layer for RAG
  • APIM – Secure, scalable API layer
  • Azure Functions – Serverless compute for tool integration
  • Azure Logic Apps – Workflow automation for tool orchestration
  • App Service / AKS / ACA – App hosting
  • Azure DevOps / GitHub Actions – CI/CD for agent apps
  • Azure Monitor / Application Insights – Monitoring and logging
  • Azure Key Vault – Secure secrets management
  • Azure Storage / Cosmos DB – Data storage for agent state and knowledge

Dev toolchain to be considered

  • Semantic Kernel – For agent memory, planning, and tool integration
  • Visual Studio Code – IDE for development
  • Jupyter Notebooks – For interactive data exploration and model testing
  • GitHub Codespaces – Cloud-based dev environment for labs
  • GitHub or Azure DevOps – Source control and CI/CD pipelines and project management
  • Azure CLI / PowerShell – Command-line tools for Azure management
  • Bicep – Infrastructure as Code for Azure resources

Lab Exercise to be considered

Accelerator Repo to be considered

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published