Stephen Toub: AI in .NET - Episode 344

Apr 7, 2025

Stephen Toub, a Partner Software Engineer at Microsoft with nearly 24 years of experience in .NET, delves into AI's transformative role in software development. He discusses the evolution of .NET libraries to support AI-enhanced tools improving developer productivity and user interfaces. Toub explains the concept of advanced agent systems powered by large language models, and highlights critical integrations in the .NET ecosystem that allow interoperability across various AI frameworks. His insights pave the way for a bold future in AI and .NET.

Ask episode

AI Snips

Chapters

Books

Transcript

Episode notes

ANECDOTE

Stephen's Varied Career Path

Stephen Toub found his passion in a blend of roles, from coding to management and writing, which took him half his career to discover.
This diverse experience led him to his current role in .NET libraries, evolving with industry shifts like parallel computing and open-source adoption.

INSIGHT

Dual Tracks of AI in Development

AI impacts developers in two main ways: AI-augmented developer tools and AI capabilities embedded in applications.
The latter involves AI as a computational resource, not just for chatbots but as functions integrated within software.

INSIGHT

Future of User Interfaces

Conversational and voice interfaces are becoming the dominant mode over traditional forms and text entry.
Humans will still review AI-generated content using visual confirmations, but data entry will evolve towards natural language interaction.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

Stephen Toub is a Partner Software Engineer at Microsoft. Stephen has been at Microsoft for almost 24 years and has done it all. If it has to do with .NET, he’s been in it. And today, he looks after all the .NET libraries making sure .NET continues to be the fastest platform on the planet. He ran the MSDN Magazine before it morphed into the docs and blogs of today. He was a leader in the concurrency and async and parallel programming developments, and now he’s turned his attention to pushing the AI envelope with .NET.

Topics of Discussion:

[4:37] Stephen morphed a few of his many interests into the role he has today.

[6:04] How moments like the push for parallel computing, the open-sourcing of .NET, cross-platform support, and performance goals shaped Steven’s current focus. [7:37] Steven highlights the two major AI tracks: (1) AI-augmented tools that improve developer productivity, and (2) building AI capabilities into applications themselves.

[10:37] The future of user interfaces.

[12:43] Why “agents” are basically advanced actor-based systems empowered by large language models and tool calling, enabling reasoning and orchestration beyond simple Q&A. [23:22] Model context protocol.

[24:08] A paradigm shift in system integration.

[30:24] How Stephen and the .NET libraries team factor out common AI abstractions for the ecosystem, so that different frameworks (Semantic Kernel, Onnx Runtime, Olama, etc.) can interoperate.

[32:15] Steven gives examples of how minimal C# code, combined with locally hosted LLMs or cloud endpoints, can solve real tasks.

Mentioned in this Episode:

Clear Measure Way

Architect Forum

Software Engineer Forum

Programming with Palermo — New Video Podcast! Email us at programming@palermo.net.

Clear Measure, Inc. (Sponsor)

.NET DevOps for Azure: A Developer’s Guide to DevOps Architecture the Right Way, by Jeffrey Palermo

Stephen on GitHub

Stephen on LinkedIn

Demystifying Retrieval Augmented Generation with .NET

Suno

csharp-sdk

Want to Learn More?

Visit AzureDevOps.Show for show notes and additional episodes.