Microsoft surely knows how to make some noise and keep us on our toes! With the Microsoft Ignite 2023 officially behind us, let us take you through our Purple Peeps’ most important takeaways, including the latest in AI, Cloud and the announcement of Microsoft Fabric becoming generally available.


Ravi’s Takeaways

  • Getting your enterprise ready for Microsoft 365 Copilot
  • Learn to Live: Prepare for, implement, and secure Microsoft 365 Copilot
  • Extend Microsoft 365 Copilot with your line of business apps and data

This year’s Microsoft Ignite is all about AI, and Microsoft 365 Copilot is taking centre stage. Copilot is an AI assistant that enables users to focus on creating rather than recreating. It is built around each user, grounded on their data and permissions. For organisations already using the Microsoft ecosystem, deploying Microsoft 365 Copilot is quick since the Microsoft Graph is already built for their users.

Like any other AI product, Copilot’s large language model (LLM) has two main aspects – skills (what the AI can do) and knowledge (what the AI knows). But Microsoft has enabled a lot of customisability in both areas by allowing developers to build plugins to extend the AI’s skillset and allowing the extensibility of the Graph by connecting it to external data. Microsoft is also releasing developer toolkits for these extensions!

Copilot also complies with strict privacy and security measures, allowing admins to manage policies and access controls easily via a dedicated dashboard. The user data, including externally connected data, is never used to train Microsoft’s models and is never exposed to the OpenAI LLM. Admins can also set security policies in just one location and have it automatically flow through to all instances of Copilot implementation.

Microsoft’s intent to change how we work and make AI accessible is clear. Copilot’s implementation helps with tasks like content summarisation and generation and unlocks advanced functionality such as querying and editing business data using natural language directly from Teams. Strict governance ensures the AI acts in accordance with various company policies, just as expected of any individual.

The future of Copilot is bright, with expanding possibilities. And that future can be here soon if only Microsoft reduces the barrier to entry…


Reagan’s Takeaways

  • Make your data AI-ready with Microsoft Fabric and Azure Databricks
  • Build and manage enterprise-scale data warehouses in Microsoft Fabric

This session discussed building and managing enterprise-scale data warehouses in Microsoft Fabric. The Speakers were Bogdan Crivat, Vice President and head of Azure Data Analytics and Priya Sathy, head of Microsoft Fabric Data Warehouse, Azure Synapse Dedicated SQL Pools and Serverless SQL Pools.

The session explored some of the new features and improvements within Fabric and allowed data analytics customers and analytics specialists to dive into deep and tech heavy questions on what and how Fabric could work in their world.

I am blown away by what the Fabric team have brought to the table. Microsoft claims that Fabric will not compromise on scalability, technology or open-source nature using open data formats. The pricing and performance of Fabric is expected to match or beat traditional data analytics workspaces. However, Microsoft is up front that they are happy for customers to continue with their existing platforms and only try Fabric when they are ready and have made Fabric migration guidance and tools publicly available.

Fabric now features more SQL functions, including Dynamic Data Masking, Rest API Support, SQLPackage Support and some other QoL benefits data engineers will appreciate. For the release management people, Microsoft will be landing GIT integration across Fabric, Warehouse will have GIT available in the next few months. Every piece of code can integrate with GIT and some workloads have ADO Deployment Pipeline capability.

Governance… There is no substitute to Microsoft Purview, and Fabric Governance will integrate into the Purview Full data estate, or ‘MetaSystem’. It is clear they are focusing on a full data estate across Microsoft, with Purview being the backbone of the operation.

The biggest announcement is the Private Preview Release of Fabric Mirroring. Fabric Mirroring is the evolution of Synapse Link, which provides Deltas, Sub-select schemas, and table Change Data Capture ingestion into Onelake as Delta Parquets. Mirroring is a dedicated ETL without any configuration and gives you a lightweight, cheap and fast data warehousing solution. Ingesting data via Mirroring will be free, however, data stored and queried in OneLake will consume Fabric Capacity units and storage costs. This pricing model is attractive and advantageous for Microsoft and their customers.


Ruby’s Takeaways

  • ERP in the era of AI: A new face of business efficiency and agility
  • Transforming service organisations with generative AI
  • Accelerating your business with adaptive banking solutions

Microsoft’s Copilot revolutionises the way businesses operate by seamlessly integrating advanced AI capabilities into their workflows. Acting as an intelligent assistant, Copilot enhances productivity and efficiency by automating repetitive tasks, assisting in code development, and providing valuable insights. This innovative tool accelerates the software development process, enabling businesses to meet tight deadlines and deliver high-quality products. With its collaborative features, Copilot facilitates effective communication and knowledge sharing among team members, fostering a collaborative and dynamic work environment. Ultimately, Microsoft’s Copilot empowers businesses to stay ahead in the fast-paced world of technology, driving innovation and achieving success in today’s competitive landscape.

Microsoft’s Copilot transforms the development landscape and streamlines support services for businesses. Leveraging its advanced AI capabilities, Copilot assists support teams in efficiently handling customer inquiries and resolving issues. By understanding natural language and context, Copilot provides real-time suggestions and solutions, reducing response times and improving the overall customer experience. This tool empowers support agents by automating routine tasks, allowing them to focus on more complex and personalised interactions. Through its seamless integration, Copilot enhances the effectiveness of support services, ensuring that businesses can deliver prompt and effective solutions to their customers, thereby building stronger relationships and bolstering customer satisfaction.


Ben’s Takeaways

  • The AI-era of Industrial Transformation

At the MS Ignite session BRK271H, the focus was on the AI-driven industrial transformation era. One major concern highlighted was the impending retirement of skilled experts and the challenge of attracting and training new talent on cutting-edge technologies before losing the expertise of retiring professionals. Intelligent factories aim to tackle this by harnessing operational and information data and integrating it across various business facets like engineering, systems, manufacturing, service, and supply chain. This comprehensive data usage intends to enhance product quality, optimise processes, and mitigate the talent gap.

The integration of AI in this sphere is already yielding notable results, particularly in improving safety and operational efficiency. The demo of Azure IoT operations was a highlight, showcasing the process of digitising physical manufacturing assets and enabling operators to access near real-time information about these assets. The platform also allows for the creation of data pipelines within the application itself, enabling the transformation of data into business-ready information. This information is seamlessly integrated into the Microsoft fabric environment, where data specialists can further analyse it to glean insights, enhance production lines, and train models for digital assets.

  • Becoming an AI-Powered Organisation with Microsoft Copilot
  • GitHub Copilot and AI for Developers: Potential and Pitfalls with Scott Hanselman

The BRK231H session delved into the potential pitfalls of GitHub Copilot and AI for developers. The discussion emphasised the importance of context in AI’s capabilities, noting that while AI is not capable of mind-reading….yet, prompt engineering is emerging as a crucial skill. Prompt engineering involves blending business, technical, and linguistic skills to define problems for AI solutions effectively. Understanding problems thoroughly before engaging AI becomes pivotal as “Bad Input = Bad Output” – indicating that the quality of input data significantly impacts AI-generated outcomes.

Furthermore, the session highlighted parallels between human interaction and AI responses, emphasising that the nature of input affects AI output just as it influences human interactions. Context is key in both scenarios.

A notable demonstration involved using GitHub Copilot integrated with Visual Studio Code to resolve an unresolved feature from an old repository bug list. While it required some adjustments, eventually, the developers achieved the desired outcome. This showcased the potential efficiency boost in development work with Copilot and the promising prospect of AI scanning GitHub backlogs to address issues. However, it was stressed that AI isn’t flawless, and thorough testing of code, fixes, and features remains imperative before acceptance. In essence, AI can offer the recipe for tacos, yet the actual taste test is essential to guarantee their flavour. The session provided a glimpse into the future of code management and deployment but underscored the need for human validation and testing in the development process.


Juan’s Takeaways


Cloud services have become indispensable for businesses to access cutting-edge innovations and maintain competitiveness. And Microsoft Ignite 2023 has made it clear: Azure AI is transitioning from just an add-on to increasingly becoming the core of its cloud business model.

The event today, 16th Nov (15th in the US), showcased how AI advancements such as LLM, retrieval augmented generation, and sophisticated search technologies are becoming more integral to Azure’s data architecture, merging seamlessly into platforms like Cosmos DB, PostgreSQL, Fabric, and PowerBI. This shift is reflective of the dramatic transformation in the perception and usage of AI, moving from initial curiosity to proactive integration in business strategies and products, indicative of a broader shift in client expectations from mere awareness to active implementation.

Furthermore, the introduction of Model-as-a-Service adds a new dimension to the ‘Everything-as-a-Service’ paradigm: the service offers ready-to-use APIs, ranging from GPT-4 Turbo to Meta’s Llama 2, and even Jais, a model trained on Arabic-language data. This expansion signifies an important leap towards a more interconnected and intelligent cloud ecosystem.


 As Microsoft Fabric becomes generally available, a common question arises: is this tool right for me? Fabric is notably tailored for productivity and seamless integration, boasting built-in capabilities like auto-tuning and auto-optimisation. Its primary aim is to simplify complex infrastructure tasks, eliminating the need to set up VMs or clusters for data workloads—a requirement in platforms like Databricks, Snowflake, and Azure Synapse. Essentially, Fabric transforms the workspace with a level of simplicity reminiscent of Office 365’s user-friendly design, distancing itself from some traditional PaaS experiences.

However, this simplicity doesn’t equate to a lack of robustness, especially for expert teams. Another key update for Fabric is the planned support for third-party and open-source features, including upserts using merge statements, version control, time travel, cloning, and vacuuming. While not all features are available immediately, the roadmap includes incremental additions over the next 12 months, progressively aligning it with the capabilities seen in Delta and Spark. This strategy highlights Fabric’s ambition to encompass comprehensive analytics and BI systems, offering a more complete solution than its competitors.

Furthermore, Fabric’s cost-effectiveness was emphasised. The team assured that running workloads on its SaaS solution would be more economical compared to other technologies. This cost efficiency goes beyond just storage and compute costs; it encompasses the overall project cost due to the significant reduction in data movement and duplication.

This is particularly advantageous for Budget Owners, thanks to the integrated Data-Lake-as-a-Service feature, OneLake, which offers functionalities like Shortcuts, akin to Snowflake’s Zero-Copy Cloning and Replication, allowing users to create virtual table references to data stored in various locations, including other clouds. This provides a comprehensive view across different data sources while obviating the need for physical data movement, thereby streamlining operations and cutting costs. The era of laborious storage account management and excessive cost overheads is effectively over.

In conclusion, whether you’re a seasoned data engineer, a database administrator a novice analyst, or a business leader seeking deeper insights, Fabric’s suite of tools and features is designed to meet a broad spectrum of needs.


Previous Post Next Post