3 Things: Analytics and AI Highlights from November 2024

This month's '3 Things' blog post includes topics ranging from the the announcement of price increases in Power BI, to the introduction of new agentic capabilities in Microsoft 365 Copilot, to guidance from Microsoft on how to modernise your business with AI.

Read on as we highlight three things from five technology areas that you should be aware of from last month.

Power BI

• Microsoft announced price increases for Power BI Pro and Premium Per User licenses. These won't take effect until April 2025, but it's something to be aware of now and prepare for. This is the second price increase for Power BI (the last one was in 2023, when Microsoft increased its prices by 9% across all its Microsoft Cloud services). However, this is the first price increase specifically for Power BI. Considering how much improvement there's been in Power BI, it's been quite an achievement that the price hasn't increased much all this time. Regardless, you should start assessing whether the new prices will impact your current and future plans. Learn more.

• It's easy to argue that Power BI is the industry leader when it comes to business intelligence platforms. However, this doesn't mean it's perfect. One area where it can sometimes fall short compared to other offerings on the market is the range and flexibility of its data visualisation options. Fortunately, this gap has been noted by the Core Visuals Team, a division of the Power BI engineering team. They recently published an elegant Power BI report called the Core Visual Vision Board which can help you track the development and execution of the Power BI Core Visuals roadmap. Learn more.

• Org Apps became available as a public preview. What are Org Apps? They're a new kind of app in Fabric, similar to Power BI apps, but you can have more than one per workspace, and they can include Fabric items like notebooks and real-time dashboards. Unlike Power BI apps, they require a Fabric capacity and allow you to preview the changes to the apps before saving and publishing. Learn more.

Microsoft Fabric

• One of the most significant announcements for Microsoft Fabric at Ignite 2024 is the introduction of Fabric Databases. Previously, if you needed a transactional SQL database in your Fabric solutions (e.g. to store the metadata for a metadata-driven ETL framework), you had to create this outside of Fabric. Using an Azure SQL database for this role was a common practice, and although this worked, it added some undesirable complexity. Now, with SQL databases in Fabric, you get all the great features of an Azure SQL database with the simplicity that comes with Fabric experiences. This includes auto-indexing, auto-pausing, auto-scaling, and automatic mirroring to OneLake for analytic workloads. With the addition of this feature, Microsoft Fabric has transitioned from being a unified analytics platform to a unified data platform. Learn more.

• Ignite 2024 saw the announcement of public REST API, Git integration, and deployment pipeline support for Dataflow Gen2. These features will bring the power of continuous integration and continuous deployment (CI/CD) to Dataflow Gen2 items, allowing for more streamlined and efficient development processes. Without these, managing and deploying Dataflow Gen2 requires manual interventions and lacks seamless integration with Git repositories. With this update, you can more easily automate your Dataflow Gen2 deployments. There is a delay as to when these features are rolled out, and although they were meant to be available by now, you can expect the Git and deployment pipeline support to arrive in January 2025, and the REST API to be available in early 2025. Learn more.

• Data Activator reached its general availability milestone, and with this, the 'Reflex' has been renamed to 'Activator', and billing meters based on compute and storage consumption are now in effect. There are also some updates to existing features, including more 'options for filtering, summarising, and scoping your data' when setting up the business rules. This enables you to carefully calibrate how changes to your data are tracked and the actions that they trigger. Learn more.

Microsoft 365 Copilot and Copilot Studio

• There were many Microsoft 365 Copilot announcements at Ignite 2024. Copilot Pages got support for rich artefacts such as code and charts, Copilot in Word gained the ability to present you with information sources to support your document drafting, and Copilot in Excel can now analyse text such as survey responses including sentiment analysis. Most notably, Microsoft announced the preview of Agents in Microsoft 365. These agents are specialised AI assistants with the skills and knowledge needed to perform specific tasks. Learn more.

• Like Microsoft 365 Copilot, Ignite 2024 brought in a flurry of Copilot Studio announcements. One of the most significant ones is the direct integration with Azure AI Search as a knowledge source and the support of over 1,800 models from Azure AI through model inference. This enables you to blend SaaS and PaaS AI development and deployment workflows with minimal friction. Another major announcement is the public preview of Autonomous Agents, which you can use to perform tasks on your behalf, including responding to events and taking action. Learn more.

• There's currently a lot of buzz on AI agents, and you may have heard terms like 'agentic AI systems' many times. This direction is all about going beyond AI chatbots and moving towards orchestrating processes and tasks that incorporate autonomous AI to achieve set goals. Although the term 'autonomous' is used, these agentic AI systems can and should keep humans in the loop to ensure quality and control. A recent article from Microsoft provides a non-technical introduction to the concept and provides examples of how it's changing the way we work. Learn more.

Azure Analytics and AI

• Ignite 2024 introduced Azure AI Foundry. This is a unified platform for AI that replaces Azure AI Studio and introduces many new capabilities. One of the biggest is the Azure AI Foundry SDK, which introduces new integrations with GitHub, Visual Studio Code, and Copilot Studio. Azure AI Foundry primarily targets developers, AI engineers, and IT professionals and enables you to build, host, run, monitor, measure, and govern your AI solutions. Learn more.

• The Mosaic AI Agent Framework in Databricks enables you to create autonomous AI agents with four core components: central agent, memory, planning, and tools. It also helps you log and monitor your AI agents' performance and internal behaviours as you develop them, supporting integrations with MLflow. A recent Databricks blog demonstrates what this could look like using a conversational agent for online retail as an example. Learn more.

• Modernising your business with AI is no small feat. It requires a blend of strategy, processes, and technology and getting all of these just right poses many challenges. A recent article from Microsoft guides on this topic by first highlighting the barriers that complicate AI adoption, then presenting a checklist of things that you can do to create a 'holistic modernisation approach'. Learn more.

Open-Source Analytics and AI

• Unity Catalog makes it easier to work with Apache Spark and Delta Lake by abstracting away the physical location of files, enforcing access control, and acting as an entry point for your data and ML/AI assets. In a recent blog post, the Unity Catalog team delves into how Unity Catalog can help you manage data at scale and what some of the integration points are with Apache Spark and Delta Lake. Learn more.

• Polars continues to be one of the most rapidly adopted DataFrame libraries in the industry right now. This is because of the incredible performance it can achieve when processing data on a single machine. When handling data, it is essential to use appropriate data types If you want the Polars query engine to process your data as efficiently as possible. The Polars team have published an article that explores the data types supported in Polars, and what the equivalent Python types are. Learn more.

• It's pretty easy to argue that optimisers are one of the most important parts of an analytical database system. DuckDB has one of the fastest execution engines today in the open-source analytics space, so the team behind it is well-positioned to explain what optimisers do and why they are crucial. That is precisely what they've done in their recent blog post about optimisers. Learn more.

Did You Find This Useful?

Get notified when we post something new by following us on X and LinkedIn.