Supercharged profiling: Finding performance bugs with agents | DEM300

Microsoft Developer · Advanced ·🏗️ Systems Design & Architecture ·1mo ago

Key Takeaways

Using Visual Studio and AI-powered agents for profiling and debugging performance issues in distributed systems

Original Description

Modern performance problems are harder than ever to diagnose: distributed systems, async code, and massive data make traditional profiling insufficient. In this demo-driven session, watch how Visual Studio combines advanced diagnostics with AI-powered agents to surface bottlenecks, explain root causes, and guide fixes faster than ever. From CPU and memory issues to “it only fails in production” bugs, you’ll see how profiling evolves in an agentic world. Seating for this session is first-come, first-served. Add it to your schedule to plan your day and arrive early to secure a spot. To learn more, please check out these resources: * https://aka.ms/build26/DEM300 𝗦𝗽𝗲𝗮𝗸𝗲𝗿𝘀: * Nik Karpinsky 𝗦𝗲𝘀𝘀𝗶𝗼𝗻 𝗜𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻: This is one of many sessions from the Microsoft Build 2026 event. View even more sessions on-demand and learn about Microsoft Build at https://build.microsoft.com DEM300 | English (US) | Developer tools & frameworks Demo | (400) Expert #MSBuild Chapters: 0:00 - Using Visual Studio Profiler to identify performance issues 00:02:30 - Creating benchmarks for accurate performance testing 00:04:55 - Cloning the repository and opening the solution in Visual Studio 00:06:19 - Explanation of in-memory CSV setup to isolate enumeration performance 00:09:01 - Review and refinement of Copilot-generated benchmark code 00:12:05 - Profiling trace captured and loaded into Visual Studio for analysis 00:15:01 - Benchmark re-run shows marginal improvement; discusses measuring impact and continued optimization 00:17:52 - Copilot as a Learning Tool for Software Engineers 00:18:33 - Transition to Presentation – Encouraging Reproduction of Performance Results
Watch on YouTube ↗ (saves to browser)
Sign in to unlock AI tutor explanation · ⚡30

Related Reads

Chapters (9)

Using Visual Studio Profiler to identify performance issues
2:30 Creating benchmarks for accurate performance testing
4:55 Cloning the repository and opening the solution in Visual Studio
6:19 Explanation of in-memory CSV setup to isolate enumeration performance
9:01 Review and refinement of Copilot-generated benchmark code
12:05 Profiling trace captured and loaded into Visual Studio for analysis
15:01 Benchmark re-run shows marginal improvement; discusses measuring impact and co
17:52 Copilot as a Learning Tool for Software Engineers
18:33 Transition to Presentation – Encouraging Reproduction of Performance Results
Up next
Implement Microsoft Entra ID Auth with Delegated Graph API Calls in ASP.NET Core Web App Razor Pages
Dewiride Technologies
Watch →