Using AI tools to teach old apps new tricks | BRK220
Nish Anil, Hazem El-Hammamy, and Jeff Fritz present a Microsoft Build 2026 breakout on using agentic AI to modernize legacy applications with GitHub Copilot, focusing on analysis, dependency mapping, upgrade planning, safe refactoring, and governance at scale.
Overview
Modernizing applications often involves more than rewriting code: teams need to untangle dependencies, trace data flows, and make changes without breaking production. This session focuses on using agentic AI—specifically GitHub Copilot modernization capabilities—to take on the hardest parts of modernization across large portfolios.
What the session covers
AI-driven modernization challenges
- Analyzing large, complex codebases
- Mapping dependencies and understanding data flows
- Planning upgrades and sequencing changes
- Refactoring in a way that reduces risk to production systems
Principles for building modernization agents
The speakers outline principles for modernization agents, emphasizing:
- Scale (handling large portfolios and large codebases)
- Customization (adapting to organization-specific standards and constraints)
- Governance (controls to keep changes safe and compliant)
Azure and GitHub Copilot as a modernization platform
- Positioning Azure and GitHub Copilot together as a unified approach for modernization work
Demos and scenarios highlighted
GitHub Copilot modernization agent and custom skills
- Demonstration of a modernization agent workflow
- Use of custom skills to tailor the agent to specific modernization tasks
Mainframe modernization: COBOL to Java
- Demonstration scenario focused on translating/modernizing mainframe workloads
- Example path: COBOL to Java
Portfolio modernization at scale using CLI agents
- Using command-line driven agents to modernize multiple applications across a portfolio
Governance: command center and rule books
- A “command center” concept for overseeing modernization work
- “Rule books” used to guide and constrain agent behavior for governance
- Execution plans tied to rule books to control how changes are proposed and applied
.NET legacy modernization and deployment to Azure
- A scenario focused on modernizing legacy .NET applications
- Deployment of the modernized application to Azure
Resources
Session metadata
- Event: Microsoft Build 2026
- Session: BRK220 (Breakout, Advanced)
- Language: English (US)
- Track: Cloud platform & data
Chapters (from the video)
- 0:00 - Welcome and Introduction to AI-Driven Modernization
- 00:01:11 - Modernization Challenges and the Role of AI Agents
- 00:04:09 - Principles of Building Modernization Agents: Scale, Customization, Governance
- 00:06:01 - Azure and GitHub Copilot: Unified Modernization Platform
- 00:10:23 - Demo: GitHub Copilot Modernization Agent and Custom Skills
- 00:12:07 - Mainframe Modernization Demo: COBOL to Java
- 00:17:00 - Portfolio Modernization at Scale Using CLI Agents
- 00:26:00 - Command Center and Rule Books for Governance
- 00:31:04 - Custom Skills and Execution Plans with Rule Books
- 00:33:19 - .NET Legacy Modernization, Deployment to Azure, and Recap of New Capabilities