How do you measure the quality of a large language model? Carl and Richard talk to Dr. Jodie Burchell about her work measuring large language models for accuracy, reliability, and consistency. Jodie talks about the variety of benchmarks that exist for LLMs and the problems they have. A broader conversation about quality digs into the idea that LLMs should be targeted to the particular topic area they are being used for - often, smaller is better! Building a good test suite for your LLM is challenging but can increase your confidence that the tool will work as expected.

What are JavaScript promises, and why do you want to make them? Carl and Richard talk to Martine Dowden about all the various async options available in Javascript today, including Callbacks, Promises, Async/Await, and even ReactiveJS! Martine digs into some of the more remarkable features available, including grouping sync calls together so code is only called when they all complete, or the race option where only one needs to complete, and everything else is thrown away. Lots of power is available in Javascript today. Have you taken advantage of it?

What's the latest with .NET Aspire? Carl and Richard talk to Rob Richardson about his experiences with .NET Aspire to help build great .NET cloud apps. Rob talks about all the goodness that comes out of the box with Aspire, including OpenTelemetry, containerization, good security practices, and the excellent dashboard. The discussion turns to the challenges of evolving .NET to be better in the cloud, retrofitting existing applications with Aspire, and all the container choices you have in front of you with these tools. There's more than one way to fall into the pit of success!

React version 19 has been released! Carl and Richard talk to Aurora Scharff about the long-awaited version of React that incorporates React Server Components and many other features. Aurora talks about the rethink involved in switching to a server-first implementation of a React website, which is best suited for greenfield implementations. For existing React apps, you'll want to look at React Router, which has V7, incorporates Remix features, and provides a bridge between React 18 and 19. Lots of progress from the library that runs Facebook!

How can vertical architecture help you? Carl and Richard talk to Jeremy Miller about using vertical architecture to help build applications quickly and reliably. Jeremy talks about resisting the over-thinking of architecture leaving room for developers to build the app and get to results rapidly - by taking a vertical slice of the problem space, end-to-end, and getting something running as soon as possible. The conversation digs into many concerns about taking shortcuts, collaborating with other teams, duplicated work, and more!

Ready to speed up your cross-platform development? Carl and Richard chat with Francois Tanguay and Sasha Krsmanovic about Uno Hot Design. First shown at .NET Conf in 2024, Hot Design brings the Hot Reload experience to UX onto your various client devices. Francois talks about the evolution of the Uno Platform into a place where you can use a variety of client libraries to build your app and deploy your client of choice. But when it comes to iterating the implementation, Hot Design speeds your dev cycle so you can get more done in less time!

How do customers take control of their data from merchants? Carl and Richard chat with Richard Reukema about his book The Empowered Customer. Richard discusses building a data cooperative between customers and merchants using ethical data handling techniques and technology to create mutual benefit. The conversation dives into how to get merchants to migrate from their loyalty programs into this more constructive and broader model.

Can tooling make implementing AI features in your applications easier? Steve Sanderson says yes! Carl and Richard talk to Steve about the Microsoft.Extensions.AI preview toolset for OpenAI and oLlama. Steve discusses ideas around useful places for AI technologies to appear in your application, not just chat. The conversation digs into more ambient ideas, like providing suggested cut-and-paste items when entering forms and even dynamic changes to UI based on how a user interacts with the application. Want to get started? Get the extensions on NuGet!

Ready for a migration story? Carl and Richard talk to Nathan Westfall about his experiences moving an application for school buses from Xamarin Forms to Blazor. Nathan describes the interplay between a tablet on the bus for the driver, cloud services in AWS, and parent smartphones. The discussion dives into the advantages of Blazor on the client from a server resources perspective when dealing with hundreds of thousands of parents, plus being compliant with all of the rules and expectations of a public service sector product. Great insights on how to make apps people use every day!

What's a Microsoft DevBox, and why do you want one? Carl and Richard talk to Isaac Levin about the power of DevBox to help you get up and running fast with a development project. Issac describes a virtual workstation designed for software development with much more processing, memory, and storage options. With the management tools, you can quickly build templates to create new instances,and only pay for what you use. You can have instances for different projects, even different versions!

A visit from one of Scott Guthrie's Ninja Army! Carl and Richard chat with Rob Conery about his latest work with Microsoft technologies, including a VS Code extension for Copilot to understand Postgres databases! Rob talks about spending time in other programming platforms besides .NET to expand his horizons, which led him to create a tool called Viper.NET, similar to the tool from the Go platform, to help manage configuration. The conversation also visits and revisits the impact of GitHub Copilot, now with a free tier, and how it is helping software developers - and generating controversy!

How is your architectural intelligence? Carl and Richard talk to Thomas Betts about his thoughts on implementing AI-related technologies into applications. Thomas talks about stripping the magic out of AI and focusing on the realities - in the end, it's just another API you can call. The conversation digs into what useful implementations of large language models look like, as UX alternatives, summarizers, and tools for reviewing existing work.