As a data professional, you know the world of data science is always evolving - but every now and then, something comes along that makes us all stop and take notice. That’s exactly what’s happening with Microsoft Fabric's new and planned features for Fabric Data Science.
What you will learn: In this edition, we'll walk through what’s new and upcoming for Fabric Data Science and understand why these updates are worth your attention. Whether you’re a senior data scientist or someone just getting started, there’s something here for you. I think this is a big deal, and I can't wait to share with you exactly what's coming and how these updates will make your data science work more efficient, powerful, and downright enjoyable.
Read Time: 5 minutes
There’s a lot to be excited about when it comes to Fabric Data Science in Microsoft Fabric. One of the standout new features that I find particularly thrilling is the introduction of AI Functions for LLM-Powered Text Enrichment and Transformation, planned for public preview in Q4 2024. Imagine being able to perform common NLP tasks like text summarization, translation, and sentiment analysis - directly in your notebook, without needing to write complex code. Yes, you read that right! I believe these AI functions will make it incredibly easy to enrich data, enhancing the quality of the data we feed into our models with just a few lines of code.
As we look forward to Q1 2025, I’m especially enthusiastic about the AI skill integration with Azure AI Foundry. This integration is set to make Fabric's AI skills a powerful knowledge source for Microsoft’s Azure AI agents. Imagine interacting with an AI agent that can query your data in Fabric to provide you with detailed, accurate answers - all without leaving the conversation. I foresee that this will effectively bridge the gap between data storage and AI interaction, making workflows more efficient and significantly smoother.
Moving further into Q1 2025, we’re also set to see the introduction of Semantic Models as a new data source for AI Skill. This is another feature that has me genuinely excited because it will allow us to query Power BI Semantic Models using natural language. Imagine asking a question like, “What were the total sales over the last 12 months?” and not only receiving the result but also the DAX query used to calculate it. For me, this is about more than just convenience, it’s about transparency and learning. I love the idea of seeing exactly how results are derived, which helps deepen our understanding of DAX and data modeling in general.
Another feature coming in Q1 2025 that I’m eager to explore is the ability to query KQL databases as a new data source in AI Skill. This means we’ll be able to query our Kusto databases in Fabric using natural language, receiving both a concise answer and the corresponding KQL (Kusto Query Language) query. How cool would it be to ask, “What was the total number of logins last week?” and receiving not only the result but also the underlying KQL query. The feature will also support providing few-shot examples and refining queries, which I believe would make it incredibly powerful. This is all about accessibility and control, and I can see how this will make data analysis much more approachable for everyone while empowering advanced users to get exactly what they need.
Also arriving in Q1 2025 is the transformation of AI Skill into a conversational AI agent. This is one of the features I’m keen to see come to life as it will enable us to engage in natural, back-and-forth conversations to understand our data more deeply. I think the ability to ask follow-up questions, refine queries, and receive dynamic insights, all in a conversational way just makes data exploration more intuitive and engaging. It feels like we’re moving towards having an AI collaborator that genuinely understands our needs and helps us get maximum value from our data.
Another feature that stands out in Q1 2025 is the introduction of Low-Code AI-Powered Operations in Data Wrangler. This suite of AI-powered operations will let us describe code transformations in natural language and generate the corresponding Python code; translate custom Python code into PySpark; and apply SynapseML transformations like text translation and sentiment analysis in just a few clicks. When I had started my career in data science, I can’t count the number of times I’ve struggled to translate an idea into code. While I believe this tool will help newcomers take that struggle away, not just for beginners but also a huge time-saver for more experienced users (something I'll definitely be exploring!)
Looking ahead to Q2 2025, I’m really excited about the Real-Time Endpoints for Machine Learning Models. This is a huge deal for anyone who needs real-time predictions. Previously, deploying a machine learning model to serve real-time predictions required significant configuration. Now, with real-time endpoints that are automatically configured, we can easily call predictions from external applications - whether integrating them into another Microsoft Fabric component or into an entirely separate system. This feature is going to make real-time applications far more accessible and straightforward.
What’s exciting to me is how these features will make everyday data science tasks more intuitive, removing a lot of the friction that often comes with disjointed workflows. I think this is a huge leap towards democratizing machine learning and making advanced analytics more approachable for everyone. If you’re a data professional, now is the perfect time to get hands-on with these new capabilities. Don't wait for the updates to drop - start exploring Microsoft Fabric today and get familiar with the platform. The more you know now, the better prepared you'll be to make the most of these exciting features when they become available. Go ahead, explore what Microsoft Fabric has to offer, and see how these tools can simplify your workflows. And if you’re just getting started in the data world, experiment, get comfortable, and set yourself up to leverage these features to embark on your data journey.