.NET Rocks! · Carl Franklin

Visual Studio Code AI with James Montemagno

September 25, 2025·1 hr 6 min·6 clips

James Montemagno compares GPT models as ticket-driven workers to Claude Sonnet's ambitious, exploratory approach in AI-assisted coding.

The setup starts in house-show mode. A Shopify read rolls into Carl and Richard bringing in Michelle LaRue Bustamante for Cybersecurity Intersection, a new 100% security-focused event tied to the Dev Intersection and NextGen AI organizers. Then the AI tooling talk gets more grounded. The guest talks about model choice less like picking a favorite brand and more like choosing the kind of help wanted that day. GPT gets the ticket metaphor. In his view, GPT models are useful when they get files, specs, and a pointed task, especially bug fixing or quick information gathering. Sonnet sits in another lane. The guest calls those models curious and ambitious, more willing to roam the codebase before touching anything. That is not magic. It is work style: speed, context appetite, and how much steering the developer wants to do. Richard and Carl keep it from turning into a product brochure. Their questions make the coding agent comparison feel like shop talk about real habits. GitHub's coding agent comes up too, with the group guessing at what it has used and how that mix may be changing. The episode knows the ground is moving. Near the end, the guest jokes that if he comes back in two months, everything will be different. That is the point. AI coding tools can help, but developers still need to know the model's personality, limits, and preferred mode. The close is classic .NET Rocks: codedwithai.com, production credits, and go write some code.

As heard by us

A practical comparison of coding models for developers using AI tools.

The episode weighs how different coding models behave in ordinary work. It draws a useful contrast between GPT models, which are quicker and fit a specific ticket, files, and specs, and Sonnet models, which tend to explore a codebase more deeply.

Read the full review in PlayNext →

Why you'd press play

Want a clearer way to choose between fast agents and exploratory ones?

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from .NET Rocks!