Microsoft Ignite 2023 Day 2 General Session Announcements

The annual Microsoft Ignite 2023 Conference started this week both in-person in Seattle and virtually. While the conference is normally all week, it is virtually available only two days this year. Today, November 16, 2023, was day two of the virtual part of the conference. Today’s general session was centered on Microsoft Cloud and how it supports AI technologies. Scott Guthrie, from Microsoft, spoke on how they are taking a two-prong approach to supporting customers in this realm by implementing both Copilot and AI into all Microsoft Cloud solutions.

GitHub Copilot

In addition to these features, GitHub is now combined with Copilot to create “GitHub Copilot”. This is made up of several other services like Copilot Chat, Copilot in GitHub Mobile, and Copilot Enterprise. All these features create a comprehensive tool for developers by enabling the ability to turn natural language into code and providing predictive suggestions for what comes next in the code. Not only does it help create code, with commands like “/doc”, it adds code comments explaining what the function is of that section of code. ?GitHub Copilot improves productivity and code for all developers.

Copilot Studio

Copilot is now combined with Power Platform to better enable coding from with other applications such as Teams. During a demo, it was shown that Copilot did not know the answer to a question. However, from within Teams, there was an option to select a plug-in. By selecting Copilot Studio it opened directly into “Conversational Plugins”, then moves into Power Platform to create metrics to enable Copilot to obtain the information. There can be up to four SharePoint sites and four websites, or even Azure OpenAI added that will collect and pool information for a response to that or any other question. Once created, the code is then published for everyone to use. This was just one example of the many ways questions and sources can be coded within Copilot Studio to create custom code quickly.

Azure Application Services

Developer tools like Visual Studio, GitHub, and Power Platform are made more powerful due to their integration in Azure Application Services. This is a combination of powerful services that includes features like Azure Kubernetes, Azure Container Apps, and others to build enterprise class applications and services. Azure Application Services enables organizations to run environments that are more flexible and scalable. Within this stack of services, there have been new features added to Azure Kubernetes Services to better manage and report on costs. As well as consumption and resource usage.

Azure Data Services

The Microsoft Intelligent Data Platform combines the power of databases, analytics, AI, and governance to create a comprehensive service that integrates directly into Microsoft Cloud. Surprisingly, most people probably do not realize that the popular ChatGPT runs on the Azure Cosmos DB and Kubernetes, with the platform developed using GitHub. This enables it to grow and scale at a tremendous speed. Improvements to the product were announces such as Search for Cosmo DB and AI integrations\SSD support for PostgreSQL. All of which improves the speed and responsiveness of data within the Microsoft Cloud.

Microsoft Fabric and One Lake

Microsoft Fabric, which was announced earlier this year, now has a central repository for all data called One Lake. The means that any customer can connect any data source to One Lake to create a central data source for all analytics workloads. Basically, it is a way to virtualize storage no matter where it exists. Even in another cloud service such as AWS and Google. There were several announcements regarding Microsoft Fabric. First, while Microsoft Fabric has only been available to a limited number of customers, it was announced as now being generally available. Second, Copilot in Fabric is now in preview mode. Third, there is now a feature called “Mirroring” which allows work with databases as if they are already imported into Microsoft Fabric. Fourth, Purview is now in preview mode to better enable security and governance to enable auto tagging and protection of data not matter where it moves.

Azure Databricks

For those not familiar with Databricks, it is a unified platform for building, sharing, and deploying data analytics within enterprises. This product is now available on Azure which means everything needed for analytics resides in one place with governance now included via Purview. During the demo, it was shown how the incorporated Azure OpenAI could be used to generate the needed SQL query just by asking the assistant, in natural language, to show them how to do it. A second demonstration was done showing how Databricks could be used to create a chatbot by uploading a collection of PDF documents that are then indexed and analyzed, resulting in the chatbot being able to answer questions regarding the uploaded documents. Databricks then logs all the transactions with the chatbot and can download this straight to PowerBI enabling concise reporting of the chatbots usage and performance.

Azure OpenAI

Microsoft has built their entire platform on top of Azure OpenAI. There were several new models announced with this service which are all currently in preview mode. These are GPT-4 Turbo, DALL-E-3 and GPT-4 Turbo with Vision. The new functionalities, among other things, better enable the analysis of videos that are uploaded to the platform. Model as a Service was also announced as a new built-in feature. This eliminates the need and complexity of managing the needed GPU hardware and works with other models from other companies. The models can even be fined tuned against an organization’s data with out fear to the data being leaked publicly. They believe that your data is “your data” and it will not be used to train any other foundation models within Azure. In fact, they have added a feature called the “Customer Copyright Commitment” to ensure customers that they do not have to fear about results being copyrighted material from another person or organization.

Conclusion

The session lasted for nearly an hour and a half and covered a wide range of topics related to the Microsoft Cloud and the many services that make up the Azure AI Studio of products. Combine these annoucements with the previous announcements regarding their new in-house designed Cobalt processor and Hollow Fiber technologies, it is obvious that Microsoft it is working hard to become the fasted and most versatile platform for AI and LLM services. Many of these are covered more in-depth during other breakout sessions and it will be interesting to hear more about where Microsoft is headed with all of these technologies.