.NET Rocks! Visual Studio Code AI with James Montemagno
18 snips
Sep 25, 2025 James Montemagno, Microsoft developer community lead, shares insights on integrating AI with Visual Studio Code. He explains the differences between VS Code and Visual Studio for AI workflows and discusses which LLM models work best for coding tasks. James highlights his experience building a production app using AI agents and emphasizes the importance of human oversight in coding processes. He also covers the use of GitHub agents to automate workflows and offers practical advice for developers looking to enhance productivity with AI.
AI Snips
Chapters
Transcript
Episode notes
Two Complementary Developer Experiences
- VS Code and Visual Studio serve different developer needs and surprisingly converge on AI features.
- Choose VS Code for fast AI-first workflows and Visual Studio for deep IDE-specific debugging and profiling.
A Year Of Shipping With AI
- James said 2025 became the year he shipped more code than ever using AI.
- He built production apps where nearly none of the code was handwritten.
Pick Models By Task And Lock The Prompt
- Match model choice to the task: use GPT variants for fast, ticket-like tasks and Claude for deep exploratory code work.
- Explicitly set system prompts and project instructions to steer model behavior and consistency.