Rui Romano

Automatic Identity Management for Entra ID is now GA on Azure Databricks! No more manual provisioning or complex SCIM setups: users, groups (even nested ones), and service principals sync automatically. Learn about how to simplify identity management and accelerate collaboration on Azure Databricks: https://lnkd.in/gVy4j32D

512

13 comentários

#CopilotStudio “Topics” are not what they seem. I often build topics without any conversation elements at all 😳 Here’s an example: This topic searches for a User in Office365 and adds it to a list variable if found. Why did I do it this way❓ Generative orchestration mode is smart enough to parse one or more User’s names and extract them from a message. Then the Agent can decide how many times to run this action itself. I think letting the Agent decide what to do is the optimal coding pattern. What about costs, isn’t that more expensive ⁉️ Sigh, it depends. Normally, this would require a prompt action (extra Copilot credits). But with Generative Orchestration mode, the Agent is smart enough to make its own decisions, and no added costs (assuming M365 Copilot User license). Then you have to consider that writing logic in topics instead of Agent flows is free-ish instead of paid. I think cost incentives to use topics instead of Agent flows will lead more people to use topics like flows. … I guess there’s three points I wanted to make: 1. Topics don’t have to include question or message nodes. They can be purely logic 2. Cost engineering will always play a role in Agent design. But don’t sell maintainability down the road. 3. Build in flexibility to let the Agent decide what to do instead of using the same thought process as a flow. 👉 If you want to learn how to build a real #CopilotStudio Agent using Generative Orchestration mode look for my full tutorial link in the comments section.

244

15 comentários

Microsoft recently published a template for implementing the medallion architecture in #Microsoft #Fabric. This architecture provides a structured approach to organizing and refining data in #OneLake, ensuring clarity, consistency, and scalability. Each zone of the #medallion #architecture should be separated into its own #lakehouse or data warehouse in OneLake, with data moving between the zones as it’s transformed and enriched. 🔹 Bronze (Raw Zone) This first layer stores data in its original format—structured, semi-structured, or unstructured. Data in the bronze zone is typically append-only and immutable. By preserving the raw data, organizations maintain a single source of truth that supports reprocessing, traceability, and future audits. If the source data resides in external storage (Azure Data Lake, Amazon S3, etc.), Fabric can even create a OneLake shortcut to that data instead of copying it. This allows the bronze zone to virtualize external data without duplication, making it more efficient and cost-effective. 🔹 Silver (Enriched Zone) The silver layer sources data from the bronze zone. Here, data is cleansed, standardized, and transformed into structured tables (rows and columns). It may also be integrated with other sources to provide a consistent, enterprise-wide view of key entities such as customers, products, or transactions. 🔹 Gold (Curated Zone) The gold layer sources data from silver and refines it further to meet specific business and analytics needs. This layer typically follows a star schema design, enabling performant, user-friendly data models that power dashboards, advanced analytics, and decision-making. Implement medallion lakehouse architecture in Fabric - Microsoft Fabric | Microsoft Learn

82

1 comentário

New in Databricks (Beta): Data Classification in Unity Catalog 🚀 We’ve introduced automated data classification in Unity Catalog—helping you quickly identify and tag sensitive data like PII or confidential information. Why it matters: 👉 Automated & incremental scanning – saves time and cost 👉 Better governance – pair with ABAC and tag policies to protect sensitive data automatically 👉 Less manual work – focus on action, not tagging Availability and prerequisites: 🔹 Available now (in Beta) on AWS, Azure, and GCP 🔹 Requires serverless compute As always, we’re very open to feedback! 👂 Feedback/questions: [email protected] 📄 Learn more: https://lnkd.in/d-PdfQDx PS. What is this "new" Beta release thing at Databricks?? 🤨 📄 Learn more: https://lnkd.in/dt9pj42d

378

14 comentários

MCP Dev Days 🚀 The talks from the Microsoft Developer MCP Dev Days conference are now available online. The conference focuses on different use cases of the Model Context Protocol (MCP) and its ecosystem. This includes topics such as: ✅ MCP in production ✅ MCP with AI agents ✅ Security ✅ Supporting tools 📽️: https://lnkd.in/gPJ_MXEE To learn more about AI/ML Ops, subscribe to the AIOps Newsletter 🗞️👇🏼 https://lnkd.in/gghdpeEF #mcp #ai #mlops #aiops

86

3 comentários

𝗜’𝘃𝗲 𝗯𝗲𝗲𝗻 𝘁𝗵𝗶𝗻𝗸𝗶𝗻𝗴 𝗮 𝗹𝗼𝘁 𝗮𝗯𝗼𝘂𝘁 𝘁𝗵𝗶𝘀 𝗿𝗲𝗰𝗲𝗻𝘁𝗹𝘆. For years, BI has been developer-led. Data models, DAX, pipelines. All absolutely vital - without them, nothing works. But here’s the thing: adoption is still painfully low. Reports don’t always fail because the numbers are wrong. They often fail because the experience is poor. People can’t find what they need, layouts feel messy, clutter hides the insight. So it got me wondering… 👉 𝗪𝗶𝗹𝗹 𝗕𝗜 𝗗𝗲𝘀𝗶𝗴𝗻𝗲𝗿𝘀 𝗿𝗲𝗽𝗹𝗮𝗰𝗲 𝗕𝗜 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿𝘀? In my first ever blog post (under the Doccrs Analytics banner) I explore: • Why adoption is the real bottleneck now • The value developers and designers each bring • The friction between them • And why I think “BI Designer” will soon be a standalone role I don’t think one replaces the other. But the balance of power is shifting. 📖 Read the full piece here: https://lnkd.in/eKqCt3mW ✉️ And if you want more of these kinds of thoughts, subscribe to Doccrs Analytics on Substack. What do you think - is design treated as optional in your BI team, or as critical as development? #PowerBI #DataViz #UIUX

141

13 comentários

kipi.ai leads the Snowflake ❄️ #Innovation charge yet again with this "Complete MCP Guide on Snowflake" by Ilyas Mohammad. It covers the entire #ModelContextProtocol (MCP) architecture and explains how you can build, deploy and execute all the pieces inside Snowflake, with ease! ➡️ Learn how MCP architecture works ➡️ Get Step by Step directions on how to build and host an entire MCP system (Server, Client, Host) in Snowflake ➡️ Try a Quickstart yourself with the provided sample code https://lnkd.in/d3rma4TN Well done Ilyas Mohammad! 👏 #AgenticAI #MCP #SnowflakeAICloud #kipiai

153