Recent News in Analytics and AI: December 2025 Edition
2nd January 2026 . By Michael A
The last wave of news from 2025 reflects a year full of rapid progress in data and AI platforms. Highlights include Microsoft's GPT-5.2 rollout, Fabric's Lakehouse maturity, and pioneering work in audio and voice models. Looking ahead, 2026 is likely to bring deeper integration of AI into everyday workflows, stronger emphasis on responsible adoption, and accelerated use of multimodal technologies across industries.
Read on and get up to speed.
Power BI
-
Power BI's 2025 recap marks a decade of innovation, showcasing how the platform has evolved from basic dashboards to a comprehensive analytics solution. Highlights include Fabric integration, Direct Lake for near-instant queries, and advanced AI capabilities that automate insights. Microsoft credits its vibrant community for shaping features and driving adoption globally. The article also hints at future priorities: tighter integration with Microsoft 365 and expanded AI-driven experiences. This retrospective underscores Power BI's role in democratising data and enabling organisations to scale analytics securely and efficiently. Learn more.
-
Excel now supports drillthrough for Direct Lake and DirectQuery models, enabling users to access detailed data directly from aggregated views. This improvement ensures real-time connectivity to Power BI datasets while preserving Excel's flexibility for analysis. Users can drill into transaction-level details without exporting static files, streamlining workflows for teams that rely on Excel for reporting. The update strengthens interoperability between Power BI and Excel, making it easier for organisations to maintain a single source of truth while empowering analysts with deeper insights. It's a significant enhancement for enterprise analytics. Learn more.
-
Power BI Q&A, which allowed users to query data using natural language, is being deprecated as Microsoft moves towards advanced AI-powered experiences. Support will end in 2026, giving organisations time to adopt alternatives such as Copilot and other AI-driven tools. The change aligns with Microsoft's strategy to provide more intuitive, context-aware analytics that go beyond simple keyword matching. While Q&A was innovative, newer solutions promise deeper insights and better integration across Power BI and Microsoft 365. Businesses should begin planning migration to maintain continuity. Learn more.
Microsoft Fabric
-
Microsoft shares how it built a petabyte-scale telemetry platform using Fabric to manage SQL workloads efficiently. The solution leverages Lakehouse architecture, Delta tables, and advanced caching to handle billions of telemetry events daily. Key design principles include scalability, cost optimisation, and real-time analytics for engineering teams. By integrating Fabric's unified compute and storage, Microsoft achieved faster query performance and simplified data governance. The article highlights lessons learned, such as schema evolution and partitioning strategies, making it a valuable read for architects designing large-scale telemetry systems. Learn more.
-
In Gartner's 2025 Magic Quadrant for Data Integration Tools, Microsoft secured a Leader position thanks to its robust capabilities in Fabric and Azure. Gartner cited strengths in unified data experiences, AI-powered transformation, and support for hybrid deployments. Microsoft's focus on scalability, governance, and integration across its ecosystem contributed to this recognition. The article emphasises how Fabric simplifies complex data landscapes, enabling organisations to accelerate analytics and decision-making. This achievement reinforces Microsoft's role as a trusted provider in enterprise data integration. Learn more.
-
Lakehouse schemas in Microsoft Fabric are now generally available, providing structured organisation for tables within Lakehouses. This feature simplifies data modelling by grouping related tables under schemas, improving clarity and governance. It supports enterprise-scale analytics by enabling better collaboration and security management. The update also aligns Lakehouse with traditional database practices, making it easier for teams to adopt. Organisations can now design more maintainable and scalable architectures within Fabric, enhancing productivity and compliance. Learn more.
-
Microsoft Fabric now allows notebooks to run in pipelines using service principals or workspace identities, providing a secure alternative to user-based authentication. This capability improves automation for data engineering and machine learning workflows while ensuring compliance with enterprise identity standards. Organisations can schedule jobs and manage pipelines without exposing personal credentials, reducing operational risk. The update strengthens Fabric's integration and supports DevOps best practices, making it easier to build scalable, secure analytics solutions. Learn more.
Microsoft 365 Copilot and Copilot Studio
-
Microsoft has rolled out significant security and governance upgrades for Microsoft 365 Copilot and its agents, designed to meet enterprise compliance needs. New capabilities include retention policies, sensitivity labels, and detailed audit trails for AI-generated outputs. Integration with Microsoft Purview ensures consistent governance across Microsoft 365, while identity management improvements strengthen access control. These features allow organisations to adopt Microsoft 365 Copilot responsibly, safeguarding sensitive data and meeting regulatory obligations. The update reflects Microsoft's commitment to secure, transparent AI deployment in business environments. Learn more.
-
Sales Agent is Microsoft's new Copilot agent tailored for sales teams, offering AI-driven assistance across the entire sales cycle. It connects with CRM platforms to deliver actionable insights, automate routine tasks, and personalise customer interactions. Features include predictive analytics for opportunity scoring, proposal generation, and conversational queries for quick data access. By reducing administrative overhead and enhancing decision-making, Sales Agent empowers sales professionals to focus on building relationships and closing deals. This release marks a major step in embedding AI into frontline business operations. Learn more.
-
GPT-5.2 is now available in Microsoft 365 Copilot, offering significant improvements in language understanding and contextual reasoning. Users can expect more accurate document summaries, enhanced content generation, and better support for complex queries. The model also delivers stronger multilingual capabilities, catering to diverse global teams. Performance upgrades ensure faster responses and enterprise-grade security, making Copilot even more reliable for business-critical tasks. This release marks a major advancement in Microsoft's AI strategy, enabling organisations to achieve higher productivity through intelligent automation. Learn more.
-
Copilot Studio's November 2025 update delivered key improvements for building and managing AI agents. New capabilities include enhanced connectors for external systems, richer data integration, and advanced debugging tools for streamlined development. Governance features have been strengthened, allowing organisations to enforce compliance and security standards. These updates make it easier for teams to customise Copilot agents for specific workflows, accelerating innovation and adoption. Microsoft's goal is to provide a flexible, secure platform for creating intelligent solutions that meet diverse business needs. Learn more.
Azure
-
Microsoft Foundry introduces updated GPT voice models designed to deliver more natural, expressive, and context-aware speech. These models improve tone modulation, pacing, and multilingual support, making voice interactions feel more human-like. The update also enhances integration with Azure AI services, enabling developers to build richer conversational experiences for applications such as virtual assistants and customer service bots. With improved latency and scalability, the new voice models are optimised for enterprise-grade deployments. This advancement reflects Microsoft's commitment to creating seamless multimodal AI experiences that combine text and voice for better user engagement. Learn more.
-
An article on the Microsoft Foundry blog explores why evaluating AI agents requires more than assessing their underlying large language models (LLMs). It highlights critical factors such as orchestration, tool integration, and real-world task performance. Effective agents combine reasoning with action, leveraging APIs and workflows to deliver outcomes beyond text generation. The post also stresses the importance of measuring reliability, latency, and compliance in enterprise environments. By focusing on holistic evaluation, organisations can ensure AI agents meet operational needs and deliver consistent value. The piece offers practical guidance for building robust, production-ready AI solutions. Learn more.
-
An article from Databricks addresses common misconceptions about Lakehouse data modelling, clarifying that traditional modelling principles still apply. The article debunks myths such as “schemas are unnecessary” and stresses the importance of structured design for scalability and governance. It introduces best practices like using dimensional models for analytics and leveraging Delta Lake for consistency. The post also explains how Lakehouse combines flexibility with reliability, enabling organisations to unify data without sacrificing performance. Practical guidance helps teams design models that support BI, AI, and advanced analytics effectively. Learn more.
-
Microsoft reflects on 2025 innovations across SQL Server, Azure SQL, and SQL Database in Fabric. Enhancements include AI-driven analytics, improved performance, and stronger security measures. Fabric integration enables unified experiences for managing SQL alongside Lakehouse and data pipelines, streamlining operations. The article also highlights elasticity and cost optimisation features that support hybrid architectures. These developments demonstrate Microsoft's focus on evolving SQL platforms to meet modern data challenges, empowering organisations with scalable, secure, and intelligent solutions. Learn more.
Open-Source
-
DuckDB brings Apache Iceberg support to the browser using WebAssembly, allowing users to run queries locally without backend systems. This capability makes data analysis faster, more private, and easier to deploy for educational or demo purposes. By combining Iceberg's structured table format with DuckDB's in-browser execution, the solution offers a lightweight alternative to traditional lakehouse setups. It reflects DuckDB's mission to simplify analytics and empower users with modern tools that work anywhere, including browser environments. Learn more.
-
Daft.ai argues that the real challenge for AI is not search but knowledge curation. While search retrieves information, curation ensures relevance, accuracy, and context, which are critical for AI-driven decision-making. The article highlights how poorly curated data leads to hallucinations and unreliable outputs in large language models. It advocates for structured pipelines that prioritise quality over quantity, enabling AI systems to learn from trusted sources. By focusing on curation, organisations can improve model performance and reduce risks associated with misinformation, making this a key strategy for enterprise AI success. Learn more.
-
Rabobank adopted Polars to accelerate data processing and improve efficiency in its analytics workflows. The case study reveals how Polars' Rust-based engine delivered significant performance gains compared to traditional Python libraries like pandas. Rabobank leveraged Polars for large-scale financial datasets, reducing computation time and enabling faster insights. The article also highlights Polars' memory efficiency and scalability, which were critical for handling complex banking data. This success story demonstrates how modern dataframes can transform enterprise analytics by combining speed, flexibility, and reliability. Learn more.
-
SAM Audio is Meta AI's latest extension of the Segment Anything concept, designed for audio segmentation. It can break down complex audio streams into distinct elements, enabling use cases such as speech separation and sound event detection. Built on foundation model principles, SAM Audio offers strong generalisation across varied audio types, reducing the need for domain-specific training. The technology paves the way for multimodal AI applications that integrate audio with text and vision. This development promises breakthroughs in accessibility, media production, and real-time analytics. Learn more.
