Building Scalable AI Apps and Agents with VS Code, GitHub Copilot, and Agent Framework
GitHub presents a demo-driven walkthrough from Amanda Silver, Rong Lu, and Shayne Boyer, exploring how developers can build scalable intelligent apps and multi-agent workflows using VS Code, GitHub Copilot, Agent Framework, and Azure AI Foundry.
Building Scalable AI Apps and Agents with VS Code, GitHub Copilot, and Agent Framework
Presented by: Amanda Silver, Rong Lu, Shayne Boyer (Microsoft)
Event: GitHub Universe 2025
Overview
This session demonstrates a modern AI-powered developer workflow using VS Code, GitHub Copilot, AI Toolkit, and the open-source Agent Framework. The presenters walk through:
- Turning GitHub Copilot and the AI Toolkit into a flexible AI development workbench
- Using Spec Kit for spec-driven development to move from requirements to working code
- Building and orchestrating intelligent AI agents that scale
- Deploying multi-agent workflows to Azure AI Foundry with built-in observability, tracing, and safety
- Managing live site operations such as incident response with Azure SRE agents
Key Demos and Highlights
1. Spec-Driven App Development
- Begin with high-level specifications, leveraging Spec Kit to automate code generation and ensure the implementation matches business requirements.
- Demonstrates the iterative process from initial spec to production-level code using VS Code and Copilot.
2. Adding Intelligence: Building and Automating AI Agents
- Use the AI Toolkit to architect, implement, and automate AI-powered agents that adapt to developer workflows.
- Walkthrough of integrating GitHub Copilot to assist with agent behavior creation and task automation.
3. Multi-Agent Orchestration
- Demo of orchestrating complex workflows involving multiple cooperating AI agents using the Agent Framework.
- Showcases best practices for communication, coordination, and scaling within multi-agent systems.
4. Operations and Observability
- How to deploy these intelligent apps and agent workflows to Azure AI Foundry, incorporating tools for observability and tracing.
- Segment on operationalizing AI: managing incidents using the Azure SRE agent, automating incident response for live systems.
5. Security and Scalability
- Patterns and guidance for building secure, scalable intelligent applications (including built-in safety and tracing mechanisms).
- Advice for shipping robust solutions from first commit to production.
Takeaways
- Real-world patterns for integrating AI into developer workflows
- Step-by-step demos illustrating end-to-end development, deployment, and operations
- Guidance on using Microsoft and GitHub developer tools for AI-powered apps and agent orchestration
- Emphasis on security, scalability, and reliability across the software lifecycle
Resources
- GitHub Copilot
- Agent Framework (Open-source)
- Spec Kit
- Azure AI Foundry
- GitHub Universe 2025 Playlist
About the Speakers
- Amanda Silver – CVP, Apps & Agents + 1ES GM, Microsoft
- Rong Lu – Principal Manager, Microsoft
- Shayne Boyer – Principal Program Manager, Microsoft
For more sessions and resources, visit GitHub Universe.