Last night Bill, Udai, Veronika and I gave our top picks from Microsoft Build at the Virtual Boston Azure AI meetup.

In order to not overlap topics, we each picked areas to cover - mine was Copilot and Copilot Studio.

My picks:

Copilot Wave 2 Spring release features – this was announced in April but still hasn’t landed for us. You’ll know when this lands due to the look of Copilot changing. The navigation will be on the left side of the screen with the new release (instead of the right where it is now).

This weekend I upgraded the semantic-kernel-getting-started repository to the latest code.

It was mostly to move to .NET 9 (I’d forgotten it was still all .NET 8) and the latest version of the Semantic Kernel libraries. However there were a few things that needed changed:

Planners Were Deprecated

The Planners were deprecated last year (more information here: Semantic Kernel: Package previews, Graduations & Deprecations).

This means the blog posts Semantic Kernel Hello World Planners Part 1 and Semantic Kernel Hello World Planners Part 2 are not longer relevant. I’ve updated those blog posts indicating the deprecation and have removed the code projects from the solution in Github (the code is still there but not loaded with the solution).

If you want a video that provides a good overview, I’d recommend watching Pablo Castro’s talk Knowledge Retrieval: RAG for enterprise agents with Azure AI Search, he did a great job highlighting most of the new features.

Yesterday was Boston Azure AI’s Edition of the Global AI Bootcamp at the NERD in Cambridge.

We had a lot of interest in the event - 85 people on the RSVP list and almost that on the waitlist. Since the space we had couldn’t handle any more people, unfortunately we couldn’t let more people in. However since there was such good interest in a Saturday AI oriented bootcamp, we are working on planning another one soon - so if you happen to be one of those people that couldn’t attend - stay tuned!

In January, I volunteered at that New York City AI Tour Event and got to attend a few sessions in the afternoon.

I spent the morning at the OpenAI booth answering general questions about Azure OpenAI. It was fun to hear what sort of projects people have and how they are using AI in general.

I’d forgotten how sore you legs can get from just standing on a cement floor all day … and was reminded to wear sneakers to my next conference - NOT my normal leather boots.

Yesterday Bill Wilder, Juan Pablo Garcia Gonzalez and I had the second Hands-on AI Workshop - this time in Cambridge MA. Like the last time, the goal was to give the attendees a hands on experience using Semantic Kernel. The target audience was .NET developers using Visual Studio Code looking to learn GenAI.

My presentation and labs focused on RAG.

Talk: Retrieval Augmented Generation (RAG)

I made a few small changes since the December workshop, mostly in the part where I discuss chunking.

Late last year I was looking for an AZD template that uses Chainlit for a UI application and deploys to Azure Container Apps, but couldn’t find one … so now I’m working on building one.

Chainlit OpenAI AssistantsAPI Azure Container Apps Starter

This is a starter template that works with Azure Developer CLI (AZD) to create a Chainlit application and deploy it to Azure in a matter of minutes. The application code currently uses OpenAI and OpenAI Assistants API - I’m working on adding more functionality to it soon.

This is the seventh part of the eShopSupport Series which covers the details of the eShopSupport GitHub repository.

NOTE: This entry will make the most sense to you if you’ve first read eShopSupport: Evaluator Project

EvaluationTests Project

The EvaluationTests project is a recently added xUnit Test Project that uses the new Microsoft.Extensions.AI.Evaluation packages highlighted in this blog: Evaluate the quality of your AI applications with ease. I think the idea is to replace the Evaluator project but, as is, it doesn’t run all 500 evaluation questions in the dev folder (which is good since it would not be cheap to do unless you are running locally).