
Sign up to save your podcasts
Or


Can Claude build interactive dashboards directly on top of Power BI semantic models? Absolutely! But does that mean these dashboards are ready for real business environments?
In this episode of BI Insight, we go beyond the AI hype and critically examine Claude Dashboards, their integration with Microsoft Fabric and Power BI, and the important gotchas that are easy to overlook when watching impressive AI demonstrations.
Based on Soheil Bakhshi's hands-on investigation, we discuss how Claude uses the Power BI MCP server and DAX queries to generate interactive dashboards without exporting or copying the underlying datasets.
But building a dashboard is only part of the story. What happens when we introduce Row-Level Security (RLS), user permissions, sharing, and multiple Fabric tenants?
We explore one of the most important findings: RLS works, but not necessarily the way you would expect when sharing Claude Dashboards with other users. We also discuss Read vs Build permissions, authentication, governance, and why something that works perfectly in a demonstration might not be practical in production.
And of course, we ask the bigger questions: Can Claude Dashboards replace Power BI reports? Are they a serious competitor to Microsoft Fabric Apps? Or do they serve a completely different purpose?
This is not another podcast celebrating how quickly AI can generate dashboards. It is a discussion about what actually works, what doesn't, and why understanding the limitations is just as important as exploring the possibilities.
🎬 Watch the original video, including the complete step-by-step setup and demonstrations on YouTube or here on Spotify.
🌐 Blog: https://biinsight.com
📚 Book: Expert Data Modeling with Power BI, 2nd Edition
💼 LinkedIn: https://www.linkedin.com/in/bakhshi/
🦋 Bluesky: https://bsky.app/profile/biinsight.com
𝕏: https://x.com/_SoheilBakhshi
Topics: Claude AI, Claude Dashboards, Power BI, Microsoft Fabric, Semantic Models, Model Context Protocol (MCP), DAX, Row-Level Security (RLS), Microsoft Entra ID, Data Security, Fabric Apps, AI Dashboards, Business Intelligence, Enterprise AI.
Can Claude build live, interactive dashboards directly on top of Power BI semantic models? Yes, it can! But building a dashboard is one thing, making it work in a real business environment is another.
In this episode of BI Insight, we take a deep dive into Claude Dashboards and their integration with Microsoft Fabric and Power BI. We explore how Claude uses the Power BI MCP server to query semantic models using DAX, without exporting or copying the underlying data.
We walk through the setup step by step, from registering an application in Microsoft Entra ID and configuring Fabric tenant settings to creating a custom connector in Claude and generating an interactive dashboard using natural language.
But the interesting part starts when we look at Row-Level Security (RLS), user permissions, and dashboard sharing. While RLS works, there is an important catch with the identity used by the connector. We also investigate Read vs Build permissions, connecting to multiple Fabric tenants, and the limitations you should understand before considering this approach for production environments.
And of course, the big questions: Can Claude Dashboards replace Power BI reports? Can they compete with Microsoft Fabric Apps? Are they actually practical for enterprise use?
This episode goes beyond generating a dashboard. It is about understanding what works, what doesn't, and the gotchas that come with it.
🎬 Watch the full video demonstration:
https://www.youtube.com/@BI.Insight
🌐 Blog: https://biinsight.com
📚 Book: Expert Data Modeling with Power BI, 2nd Edition
💼 LinkedIn: https://www.linkedin.com/in/bakhshi/
🦋 BlueSky: https://bsky.app/profile/biinsight.com
𝕏: https://x.com/_SoheilBakhshi
Topics: Claude AI, Claude Dashboards, Power BI, Microsoft Fabric, Semantic Models, MCP, DAX, Row-Level Security, RLS, Microsoft Entra ID, Fabric Apps, AI Dashboards, Business Intelligence.
I first shared this Power BI workaround in 2020. In 2026, a customer needed the same solution. Some things age very well.
Welcome to Old Dogs, Smart Tricks, my new series revisiting older techniques that still solve real reporting problems. Short, practical and straight to the point.
In this episode, I explain how to sort Matrix column headers in descending year order, without changing the sorting behaviour of other visuals.
Watch the full YouTube walkthrough
Read the original blog post
Have you ever wanted to sort the column headers of a Power BI Matrix in descending order, so the latest year appears first?
I first shared a workaround for this in March 2020. In October 2026, a customer needed the same thing. Same requirement. Same workaround. Apparently, some things age very well.
In this first episode of Old Dogs, Smart Tricks, I revisit the solution in a full video walkthrough.
What you will learn:
About the series
Old Dogs, Smart Tricks is a new BI Insight blog and video series revisiting older techniques that still solve real reporting problems. Not every useful tip needs to come from this month's release notes.
Watch the full video on YouTube
Read the original blog post: How to Sort Matrix by Column in Descending Order
Explore more from BI Insight
Do you use a different approach to sort Matrix column headers? Share your thoughts, I am interested to hear your examples.
In this episode of BI Insight, we explore how to transform Dynamics 365 Business Central sales data into meaningful insights using Power BI, from connecting to the source data and building the semantic model to creating DAX measures and generating report pages with Copilot.
Building sales reports on top of Business Central might sound straightforward. We need sales, gross profit, gross margin, customer performance, product performance, and some trends for executives. But before creating any visuals, we need to understand the underlying data.
Which tables do we need? How are they related? What filters should we apply? And what do the values actually mean?
We start by connecting Power BI Desktop to Business Central using Standard APIs v2.0 and selecting four tables: customers, items, itemLedgerEntries, and itemCategories.
From there, we look at preparing the data in Power Query, filtering sales transactions, cleaning unnecessary columns, fixing data types, creating relationships, and building a proper Date table.
We also explore how DAX Query View can help us create 50 measures in one go, and what happens when some of those measures fail because we removed columns that turned out to be necessary.
Finally, we look at using Copilot in Power BI to generate report pages, including executive dashboards, sales performance, customer insights, and product analysis. We also discuss why running the same Copilot prompt twice does not necessarily produce the same results, and why human review is still essential.
The key takeaway from this episode is simple:
The data model comes first. Copilot can help, nice visuals can help, and having 50 measures can help. But none of them can compensate for a model that does not represent the source data correctly.
📖 Explore more Power BI tutorials on BI Insight
🌏 Read the blog on BI Insight website
▶️ Watch the full Business Central to Power BI walkthrough on YouTube
💻 Get the Power BI DAX measures from GitHub
Follow BI Insight on Spotify for more practical content around Power BI, Microsoft Fabric, data modelling, DAX, Power Query, and AI-assisted analytics.
🎙️ Follow BI Insight
🌐 BI Insight
▶️ YouTube
💼 LinkedIn
🦋 BlueSky
𝕏 X
🟠 Reddit
Are you looking for a Power BI Business Central tutorial that shows you exactly how to build Sales Reports and Dashboards step by step?
In this video, I show you how to connect Dynamics 365 Business Central (D365 BC) to Power BI, model your data, and use Copilot to analyse your sales performance faster and smarter.
What You’ll Learn:
✅ How to connect D365 Business Central to Power BI the right way
✅ How to create professional Sales Reports and Dashboards
✅ How to use Copilot in Power BI for faster insights
✅ Data transformation, data modelling and performance tips
Whether you are new to Power BI or already using D365 BC, this Power BI Business Central tutorial will help you build insightful reports and take your sales analytics to the next level.
🔊 A big shout out to my fellow Business Central experts whose support was instrumental in building the Power BI report. This video wouldn’t have been possible without their collaboration:🗣️ Tharanga Chandrasekara🗣️ ManMohan Bhatt
🌐 Related blog post coming soon.
📙 Purchase my book, "Expert Data Modeling with Power BI, 2nd Edition" here.
🎙️ Follow BI Insight
🌐 BI Insight
▶️ YouTube
💼 LinkedIn
🦋 BlueSky
𝕏 X
🌐 Useful Links:
🔗 Data types in Power Query
🔗 Power BI Sales Measures for D365 Business Central
Music by: CreatorMix.com
Song: Moth
In this first episode of the Agentic AI in Power BI and Fabric series, we start with the foundations: what Agentic AI actually is, the terminology around it, and how to think about it in the context of Power BI and Microsoft Fabric.
There is a lot of noise around AI agents, agentic workflows, tools, reasoning, autonomy, and Human in the Loop. Before jumping into implementations, MCP servers, GitHub Copilot, and AI-assisted development, it is important to understand what these concepts actually mean and how they fit together.
We look at the difference between traditional AI interactions and more agentic approaches, where an AI system can reason about a goal, create a plan, use available tools, take actions, evaluate the results, and continue working towards the objective.
But agentic AI does not necessarily mean giving AI unlimited autonomy.
Human oversight, clearly defined tools, context, instructions, permissions, and checkpoints are all important parts of building reliable agentic workflows. These ideas become especially important when AI starts interacting with real Power BI and Microsoft Fabric environments.
The key takeaway from this episode is simple:
Agentic AI is not just about asking AI better questions. It is about giving AI the context, tools, and controlled autonomy to work towards a goal.
📖 Read the full Part 1 blog
➡️ Continue with Part 2: Getting Started with VS Code, GitHub Copilot, and Safe MCP Setup
Part 2 moves from the concepts into the practical setup. We start preparing the environment with VS Code and GitHub Copilot, introduce MCP servers, and look at how to connect AI agents to external tools safely before allowing them to interact with real Power BI and Microsoft Fabric environments.
Follow BI Insight on Spotify to continue the series and explore more content around Power BI, Microsoft Fabric, data analytics, and Agentic AI.
🎙️ Follow BI Insight
🌐 BI Insight
▶️ YouTube
💼 LinkedIn
🦋 BlueSky
𝕏 X
🟠 Reddit
Agentic AI in Power BI and Fabric, Part 2
In Part 1 of this series, I introduced the concepts, terminology, and mental models behind Agentic AI. In Part 2, we move from understanding those concepts towards how they can actually be applied in the Power BI and Microsoft Fabric world.
This episode continues exploring what Agentic AI means for data and analytics professionals, and how AI agents, tools, context, instructions, and human oversight can work together as part of a structured workflow.
The goal of this series is not about replacing Power BI or Fabric professionals with AI. It is about understanding where agentic workflows can genuinely help us, what role the human still plays, and how these technologies can become part of the way we build and manage analytics solutions.
📖 Read the full Part 2 article on BI Insight
🎧 Continue the Agentic AI in Power BI and Fabric series
⬅️ Part 1: Concepts, Terminology, and How to Think About It
➡️ Part 3: Agentic AI in Action with GitHub Copilot and Power BI Modeling MCP
Part 3 takes these ideas into a practical workflow using Visual Studio Code, GitHub Copilot, Power BI Modeling MCP, structured prompting, and human oversight.
🎙️ Follow BI Insight
🌐 BI Insight
▶️ YouTube
💼 LinkedIn
🦋 BlueSky
𝕏 X
In this episode of the Agentic AI in Power BI and Fabric series, we move from setting things up into actually putting agentic AI to work.
I walk through a practical Power BI modelling scenario using GitHub Copilot, VS Code, and the Power BI Modeling MCP Server. The semantic model has 54 measures spread across multiple tables, and the goal is to let the AI agent inspect the model, create a dedicated measure table, organise the measures into meaningful display folders, and apply the changes directly to the model.
But the interesting part is not simply whether AI can do the job.
We look at the complete workflow using Ask, Plan, and Agent modes, with Human in the Loop checkpoints along the way. The agent plans the work, creates working documents and guardrails, uses the MCP server to interact with the semantic model, and pauses for approval before making important changes.
And, as you will hear, everything does not necessarily come out perfect the first time. That is part of working with agentic AI too.
The key takeaway from this episode is simple:
Agentic AI works best when the task is specific, the tools are clear, and the checkpoints are real.
📖 Read the full Part 3 blog
⬅️ Part 2: Getting Started with VS Code, GitHub Copilot, and Safe MCP Setup
➡️ Continue with Part 4: From Skills to an Agentic Report Theme Audit Workflow
Part 4 takes the idea further. Instead of working with one semantic model and one MCP server, we move into a much more advanced agentic workflow using multiple Agent Skills to audit and standardise Power BI report themes across an entire Microsoft Fabric workspace.
Follow BI Insight on Spotify to continue the series and explore more content around Power BI, Microsoft Fabric, data analytics, and Agentic AI.
🎙️ Follow BI Insight
🌐 BI Insight
▶️ YouTube
💼 LinkedIn
🦋 BlueSky
𝕏 X
🟠 Reddit
Microsoft Fabric Apps: Business vs Data Apps – What’s the Difference?
What is really different between a Microsoft Fabric Business App and a Data App? And how do those differences affect architecture, sharing, permissions, licensing and Fabric capacity?
In this AI-generated BI Insight podcast, the hosts explore my original content on Microsoft Fabric Apps and discuss the two quite different application patterns.
The conversation starts with the Business App pattern, where Fabric Apps can provide the application backend and a Fabric-managed SQL database. It then moves into Data Apps, where an existing Power BI semantic model becomes the analytical foundation for a custom application experience.
Topics discussed include:
🎧 About this episode
This podcast was AI-generated using Gemini Notebook from my own BI Insight content and research. It provides a conversational audio version of the topic, making it easier to catch up while driving, walking or simply away from the screen.
🎬 Watch the original video
Microsoft Fabric Apps: Business vs Data Apps – What’s the Difference?
📘 My Book
Expert Data Modelling with Power BI
🌐 Follow BI Insight
Website: BIInsight.com
YouTube: Subscribe to BI Insight
Spotify: Follow BI Insight on Spotify
LinkedIn: Follow me on LinkedIn
BlueSky: Follow BI Insight on BlueSky
X: Follow BI Insight on X
Reddit: Follow me on Reddit
#MicrosoftFabric #FabricApps #DataApps #BusinessApps #PowerBI #SemanticModels #DataAnalytics #BusinessIntelligence #AI #GenAI #BIInsight
From the publisher's feed