Anonymous Rayfin Fabric Apps – Ep.555

Tommy hears the feature name and says it sounds dangerous, which is where this conversation should start. Mike then walks through the admin switch, the contact-form wall he hit a month ago, and the gold-silver-bronze order he and Tommy both accept.

News & Announcements

• Chicago Fabric & Power BI user group — Fabric Task Flow Studio — Two days after this episode, Thursday, August 20, the group meets at 3 p.m. at the Microsoft office in downtown Chicago, and Tommy needs RSVPs by the end of Wednesday so every name is in the building system. The session walks through Fabric Task Flow Studio, the community tool built partly by Microsoft and partly by Alex Powers, which creates items from a single prompt and can manage a project you already have. September is already on the calendar, with October and November aimed at the Desktop Bridge and MCPs.

• Fabric Runtime 2.0 is generally available — Released about five days before the episode, Runtime 2.0 is the Spark-and-notebooks path Tommy points to for transforms when mirroring and shortcuts are not in play. The stack moves from Spark 3.5 to 4.1, with Delta Lake 4.2, Python 3.13, Java 21, and Scala 2.13, and Mike highlights liquid clustering plus the native execution engine (Gluten, with Velox), which you opt into per workspace or environment for more work on less compute, with no code changes and no extra cost. That engine becomes the default in late September, so he wants it tested first: a package that ran on Spark 3 has already failed a job for him on Spark 4, and a few minutes spent enabling the engine can still save heavy Spark users hundreds or thousands of dollars.

• SQL database skills for end-to-end Fabric solutions (preview) — About five days old, these skills live in Microsoft’s Fabric skills collection and load only when the task needs them, while an MCP server or plugin stays present the whole session. General coding agents still miss Fabric’s authentication and item-discovery APIs, so the skills teach an agent to create a SQL database — tables, keys, relationships, indexes, triggers, and sample data, plus schemas with vector columns — then explore, query, and check schema changes, from the GitHub Copilot CLI, Claude Code, Cursor, or Windsurf, using Entra and a Fabric API token. Mike still wants a GraphQL skill on the list, because the database can already expose a GraphQL API and an agent would rather call that than install a SQL driver.

• Anonymous data access in Fabric apps — The line Tommy reads from the article, its third sentence, is that anonymous access is an authorization option, not a replacement for authentication. A tenant admin has to enable it under Fabric apps (preview) before an app can give the anonymous role its own privileges on a table. Mike asked Microsoft for the feature about a month ago, and he still ranks it behind data that stays in Fabric and an agent that can build the app.

Main Discussion

Topic: Anonymous data access in Rayfin, and why Mike ranks it third

A few episodes after the Rayfin sessions, and after Mike’s Chicago user-group talk where he built more than one app, the missing piece showed up. About a month ago he tried to add a contact form, the agent told him it could not be done, and he asked the Microsoft team when it was coming. It is here, it is off by default, and both hosts spend the hour deciding what it is actually for.

• Rayfin is the app. The data source decides what a stranger can do. Standard Power BI visuals still miss interactions people ask for. Mike’s example is a line chart with a brush that filters the rest of the page, which an agent can now build as a custom visual in a day or a weekend instead of weeks. A Fabric app goes further: the agent can use Vega, Plotly, or D3, against a SQL database that is read-write or a semantic model that is read-only and fast. Anonymous access matters on the SQL side, where someone with no account needs to insert a row.

• The tenant admin has to turn it on. In the admin portal, under tenant settings, Fabric apps (preview) now includes Anonymous data access. Mike likes that it ships disabled, because the people who run the tenant keep the decision. Tommy’s first reaction is that the name sounds bad, and Mike treats that caution as reasonable.

• It authorizes an operation for someone who is not signed in. A normal Fabric app passes Entra identity through to the SQL database, so the app knows Tommy’s email and row-level security can follow. Anonymous access is the case with no person attached: you allow the site to write a table anyway. Tommy stops the explanation to read the docs: anonymous access is an authorization option, not a replacement for authentication. The anonymous role and a signed-in role can carry different privileges, which is how a public form and a logged-in admin page can share one app. Mike’s own picture is an admin page where you configure the questions, then a public page you embed in an iframe.

• The wall it removes is the public write. His examples are ordinary and specific. A company site that is almost entirely one-way, until the contact form. A Replit game, built while he learns the tool Microsoft is partnering with on Fabric apps, that should save a name and a score with no organizational login. A government page that today uses Publish to web, and now also needs parameters or a write. A utility outage map, which he wished he had during the hurricane that hit Chicago the week before: type an address, land the row, show the map, and leave the customer outside Entra. Microsoft Forms, in his view, embeds poorly and only lands in Excel. Google Forms embeds cleanly and lands in Google Sheets. Either path still means hauling a file into Fabric. A Rayfin form writes the SQL database, and a notebook can use the row immediately. Tommy adds the cost: that app consumes capacity units, which a Microsoft Form does not.

• The toggle is a gap he hit while building, and a weak adoption pitch. Tommy asks what you tell a technology manager, especially when the same database may hold private data, and when someone could hang a basic email-and-password page on anonymous access without Entra. Mike sets the pitch aside. Fabric apps are still short of what he can do with Azure Static Web Apps. He found this feature by hitting an obstacle, arguing with an agent for fifteen to thirty minutes, and recognizing a missing piece from the static-web world. Adopting Fabric apps because anonymous access exists is the sentence he refuses to say.

• Gold is the data. Silver is the agent. Bronze is the anonymous user. Both hosts land on the same podium. Gold: create, read, and write stay in Fabric, so he can stop moving rows between services. He is blunt that “Fabric as a backend” is the unimpressive part of the story; one system of record is the win. Silver: an agent can build the app from a screenshot and a description. Hand him Fabric apps with no agents and he would still use them, and he thinks most people will stop at the hundreds of lines of code, so he calls this the step after Power Apps, where you describe the app instead of clicking it together. Bronze: the anonymous user, necessary for the public case and last on the podium. Tommy expected the top two to swap, then agrees with Mike’s order, and still thinks “prototype before lunch” sits closer to gold than Mike is allowing. The feature Mike wants next is a function layer, the piece he sees everywhere in Azure Static Web Apps, so an app can call external APIs and handle security there. He does not know if it is on the backlog. If it arrives, he says the medals get reordered.

• Who maintains it is the objection he keeps hearing. Customers ask what happens after an agent builds the app in half a day and something breaks. His answer is the same agent, as long as the project is in git: paste the complaint, pull the repo, ship the change. You still have to be able to tell the agent what you were building. Start on something small, live through a few debugging passes, and judge the maintenance story after that.

External visitors stay outside the tenant

Mike’s government and utility cases are people who will never have an Entra account in your tenant. Anyone who can open the app can do what the anonymous role allows, so that role belongs on the public table alone. When a page does need to know who the visitor is, he describes a second layer, a user table the app checks, rather than inviting every customer into Fabric. Tommy’s question sits on top of that pattern: a technology manager should stay cautious whenever private data shares the database.

Looking Forward

Pick one public write that has no business asking for an Entra account. Have a tenant admin enable Anonymous data access, put the anonymous role on that single table, and leave everything else on a signed-in role. Mike said he hopes to publish a build of an anonymous Fabric form in the next couple of days. Until that video is up, the order both hosts agreed on is the one to rehearse: data in Fabric first, an agent to build and fix the app, and anonymous access reserved for the visitor with no Entra account.

Episode Transcript

0:01 Tommy and Mike light up the sky. Dance today to laugh at this mix. Fabric and artificial intelligence give you what you need. Explicit standards. Start the rhythm now. The pumpkin feels the crowds. Explicit standards. Explicit standards. Welcome back to ” back to ” Frank Measures” with Tommy and Mike. Tommy, good morning. How are you? you? Good morning, Mike. I am perfectly fine, but a little sad. The children are at school.

0:31 The children are at school. Oh, yes. It’s that time of year. Everyone is sending their children now. Tommy, all my children go to school, all of them, for several years now. This is one of your first years where every child will be out of the house, all seventeen of them. They will all go all go like an Italian family. So, they So, they will all go to school and your house will be quiet during the day. Know. That would be strange. It will be sad, but it will be good. I am happy for them. Your wife is going to take a long nap. It will be

1:02 long nap. It will be extremely tiring. She will say, “You extremely tiring. She will say, ” what? I’ll just have coffee know what? I’ll just have coffee and stay in my pajamas until noon and have a quiet, relaxing day without any children in the house for a while.” Yes. She will simply tell me: ” Wake me up at lunchtime.” So, So, yes. exactly. correct. Yes. exactly. Good., before we get into our news and articles here, our main topic today looks interesting, but we will break it down here. A recently updated article on Microsoft changed how

1:32 how Fabric applications are used with an SQL database. So I want to start unpacking this a little. I really want to, I think this is a very pivotal moment for Fabric apps as we are starting to see some investments in Fabric apps that are really changing how they are used and what they can do for your business. How do you build things? What does this look like? And I think this will be a turning point, I hope, for many for many more features to come that will

2:02 more features to come that will likely enhance Fabric apps even further and make them central to what I do every day, right? ? That’s something I’m really excited about. Good. The topic is anonymous access to data in data in Fabric applications. That’s what we’ll talk about. It’s a rather technical name, but we’ll delve into the details here. This will be a very technical discussion, and I want to explain why I feel it is so important to me as a developer and creator within Fabric. Before we get into that, let’s review some news. Tony, what do you have for us? My friend, we have some

2:32 friend, we have some great news articles. The first one is this week, starting on August 18. Two days later, the next meeting of the Power PI and Microsoft Fabric user group will be held in Chicago. We will meet at the Microsoft office in downtown Chicago at 3 PM. If 3 PM. If you have not done so yet, please confirm your attendance by the end of Wednesday as we need to register all names in the system in order for you to be able to enter the building. So, this is very important. We will be reviewing Fabric Taskflow

3:02 Fabric Taskflow Studio, a Studio, a great community tool, partly built by Microsoft and partly by this humble servant, Alex Powers, that really helps you build all all your elements, manage those elements with just one prompt, cover the entire project, and even, so far, far, manage manage existing projects you may have been working on. So we are really excited. We will present a complete solution here. We are very excited about that. And if you can’t make it, guess what? We have already announced the

3:33 already announced the September meeting. We are about to announce our October and November meetups, which will be about ” Bridge” and “MCBS”. So, we’ll be doing all the demonstrations, Mike. That’s what we’re going to do. We will… yes. A question from the chat, Tommy. Someone is asking: “Is there a parking lot in the Ion building in downtown Chicago?” I think there’s a parking lot underneath it. I think you can park your car there. there. There… it costs money. It’s Chicago. Everything costs money. Just look at the building. It costs you money. But I think there is a parking lot under the building. You can drive directly to the Microsoft office and there is also a parking garage under the building.

4:03 under the building. Therefore, there are some parking spaces nearby. There is no outdoor parking area. It costs some money to get into parking lots and garages, but you can park there if you want. I advise you to park somewhere else. Use the train to get there. Use the “Parking Hero” app. Or a bicycle. Great app. Oh yes, “Spot Hero”. Spot Hero. It was great for Chicago, and I found it too. So, if you want to take a short walk, it’s located right on Randolph Street near the water. Okay, yes. It’s a bit difficult difficult to find a parking spot, but once we get there, we’ll have a great time.

4:33 we’ll have a great time. amazing. Hmm, that’s all we need there. Good. So, let’s move on to the next topic. Okay, Mike. So, this is a general availability announcement. This was released five days ago, and it relates to the Fabric 2. 0 operating environment. This represents a major achievement for data engineering and data science across the entire Microsoft Fabric. It is the foundation of a modern, high-performance implementation built on Apache Spark and deeply integrated into the

5:03 Apache Spark and deeply integrated into the ecosystem. What are some of the key updates and improvements? They have made many system upgrades. Delta Lake 4. 2, Python 3. 13, Java 21, Scala 2. 13, and Apache Spark 4. 1. Other Other performance-related issues include immediate access to smoother transitions for updated components such as Python 3. 13. Continued investment in the native execution engine provides accelerated support and improved

5:34 accelerated support and improved query performance, requires no code changes, and incurs no additional cost. You can choose to use it today at the workspace or environment level, and it will immediately become the default in late September. So this is really great. I think this is a big achievement because, Mike, we are increasingly realizing that the actual way to merge, transform, and load data in Fabric is through Spark and through notebooks if you are not using

6:04 if you are not using mirroring or shortcuts. Therefore, this is something that anyone doing this should be fully aware of and informed about. . So I will really point out that there are a few things here. I have been using Spark for a number of years. There are a lot of things going on here. The things I’m most excited about, and the , and the more I’ve worked, or at least more deeply, within Databricks, there are a few areas here that seem to make a lot of sense. One of them is to go ahead with the Spark engine. Many improvements have made things better. There are many things

6:34 many things inside the Spark engine that do a better job. So, Apache Spark 4. 1, a really big win. Anytime I see such big steps. Now, I would say that we are moving, I believe, from version 3. 5 previously and now we are in 4. 1. millimeter. Understood, but this is a major change in the version of the Spark engine. So, just be aware of that. The other thing that I think is really useful here is Delta Lake. Delta Lake has been getting a lot of features.

7:04 of features. Data bricks are working on building the concept of what is called “Delta Lake”. They use a lot of Delta Lake’s features. They do, in a way, own that open- open- source part of the project. DataBrex often creates things in Delta Lake that no one else can use because they are exclusive for 6 to 12 6 to 12 months, and then they start rolling out those features for everyone to get. One thing I would like to point out here is that I think we have now achieved liquid clustering with this new release of Spark and the Fabric runtime environment. Liquid assemblies are

7:35 Liquid assemblies are extremely complex, but they are truly fascinating. DataBrex has had this feature for years or months. I honestly don’t know how long it’s actually been around. But now Fabric offers Liquid Aggregation, which allows you to instantly repartition data. It is intended to make reading and writing data more efficient and improved. So this is another big win. The last thing I really want to point out here is the Native Execution Engine, or NE. He is NE. He is constantly working to make Spark and other things

8:06 Spark and other things more efficient. Spark is inherently efficient. It’s fast. It has It has parallel processing capabilities, but certain parts of it are slow to operate. So Microsoft said, ” Look, we’re going to take a look at the Spark engine and replace parts of it with the original execution engine or parts of a project called ‘Gluten’.” There is something else called ” Velox” that is somewhat integrated into this original engine, and it makes things faster with better performance. Therefore, you can process more

8:37 process more data with data with fewer computing resources. This is what we are looking for in this field. Therefore, I am very excited about further improvements around the native execution engine. The fact that they are highlighting this tells me that they are doing a good job here, that things are improving, and that there are additional performance gains to be made. All this to say that it is now generally available available. It is not enabled by default. Therefore, you need to test test your workloads and see if they are working or not. I would like to warn you against migrating from Spark 3

9:07 against migrating from Spark 3 to Spark 4. Definitely check and test your workloads if you have dedicated connections or packages that you are currently using. These are the things you really need to test with these new engines because they are the things you have found in the past. You’ve caused me problems. There was a package that worked on version 3 but did not work on version 4. And so one of the tasks began to fail. So you don’t want to make these updates blindly. You should definitely deal with this matter with some… Just with a little caution. But be aware that spending five

9:40 But be aware that spending five minutes clicking a button could save you hundreds, if not thousands, of dollars in computing costs if you use Spark heavily. So, this is something to be aware of, as it may save you money. No, this is a very big update. Let’s continue with more updates, Mike. It’s all about skills, my friend. We have a new article here as well, which is the availability of skills now to support the SQL database in Fabric. This is part of the

10:11 in Fabric. This is part of the Fabric skills platform or repository that Microsoft already owns. That’s amazing, Mike. I use it use it almost every day, and we can review the various use cases, not only in construction, but even in scaling and surveying. In fact, I just published a blog post about it, specifically about conducting the survey. But one of the big parts of MCP or Fabric remotely when working with AI is AI is that you have to use skills. skills. At this stage, it has become

10:41 At this stage, it has become like needing the cloud or downloading the tool. Well, you need skills too. I feel that they complement each other. What is available now are now are SQL database skills. The reason is that general AI programming agents do not do not really understand Fabric’s Fabric’s proprietary APIs for authentication or item discovery. You may have have SQL database skills, but this is specifically designed for the SQL database in Fabric. These are

11:11 workspace workload-specific guidelines, enabling agents to think more precisely about the Fabric SQL database in broader solution patterns. Skills are loaded only when only when they are relevant, unlike an MCP or plugin which is which is always present. What you can do is do is really three main things. You can compile, create an SQL database, and generate tables, keys, relationships, indexes, triggers, and sample data. You can build AI-ready diagrams with vector columns for semantic research,

11:35 vector columns for semantic research, which is amazing. You can consume, explore schemas, generate and run SQL queries, use SQL results in applications or analysis workflows, perform operations, discover items, identify endpoints, manage resources, or check schema and state changes. Therefore, it produces reviewable implementation steps for for execution. The team helps understand how to fit an SQL rule into a larger Fabric solution. It truly supports the transition from intention to execution, starting with the demand to ensure you are guided in the

12:05 you are guided in the right way. This works with the GitHub Copilot command-line interface. It also works with Cloud Code, Cursor and Windsurf to load Fabric directives from workspace configuration files. Therefore, it uses Microsoft Entra login and Fabric API access tokens. That’s fantastic, Mike. This is amazing. So I’ll put this here too. This is the Git repository for Fabric skills. So, I added the article here in that place as well. So, So, SQL database skills in “Fabrik”., if you look at

12:35 , if you look at the article linked here, “Fabrik” skills, you will find many of them. And I’m looking here, Tommy, in the skills that are available, there’s a command-line interface for an SQL database. There is a command-line interface ( CLI) directly located there. In fact, there are three main interfaces. I think they are modifying this a little. There is a command-line interface for Spark, an interface for an SQL database, and an interface for an SQL data warehouse, all of which are now listed among the skills available there as well. It’s really great to see see additional skills being released.

13:05 additional skills being released. These skills were released 5 days ago. They are constantly adding new things there as well. Okay. And honestly, this is the hardest part because they do it so often that I have to go back to it. . If you’re using it in the cloud desktop, Mike, or even if you just downloaded it, the problem is that when you change, the skill name sometimes changes. For example, the ” PowerBI” skills used to be called “PBI-”, and now they are called “PowerBI-”. And again, if there are other updates, it’s always difficult to keep up with them. However, it is

13:36 keep up with them. However, it is worth the effort. So, make sure you use your “fabrik” skills if you haven’t already. One thing I would like to note here as well. This only involves talking to the Fabric SQL database using OLTP, and there are some command-line tools that I believe they integrate here as well. Another area I would really like Microsoft to add. I don’t see this in the skills section, and I may have missed something here, Tommy, but there is a “GraphQL” feature

14:06 “GraphQL” feature in SQL databases. I do n’t know if Microsoft is listening here too, but if you think about programmatic access to Fabric as a background, we were talking about Fabric as a background. Today, we will talk more about “Rayfen” as well. If you think about what’s going on there, I want everything to talk to PowerPI and Fabric via an via an API. API. Almost, or at least as much as possible, possible, right? What this does is make it effective effective for me. It makes it effective for my robotics. It makes it more efficient for me to talk about different things. Having a Fabric command-line interface

14:37 Having a Fabric command-line interface that relies on an SQL database, where I need to use OOLTP or install other drivers or languages, is not efficient for me when I use agents. Proxies love API calls, like: ” Give me access, let me make the connection.” He can write API calls and do what he needs to do, but once I need to install SQL definitions, this is an old-fashioned method and I honestly don’t like it. Therefore, I would like to see another skill. Within Fabric

15:07 another skill. Within Fabric you can run an SQL database, and on top of that you can add a GraphQL API interface, allowing you to talk to the database as if it were an API point of contact. I love GraphQL. Many of the applications we use now, I believe, use GraphQL in the background anyway. It is a . It is a very popular technology. Therefore, I would like Microsoft to adopt this in skills. Where is the GraphQL skill? Someone needs to add it here so we can get

15:37 add it here so we can get something specific about what GraphQL does. I agree with you, but Mike, you always make me laugh about these things because you sound like my Italian grandmother when she says, ” me.” I don’t like to complain. There’s a draft here. Don’t you think so? I know the pizza is good, but it would have been better if better if they had served extra sauce on the side. Mike, that’s great. I understand what you mean. I can hear the Microsoft team saying, ” Man, we have SQL skills now.” Be happy. But no, we have to… wait, do? We always reach a

16:07 always reach a higher level. The high standard we set in the Explicit Measures podcast. A very high level. An extremely high level. Very high level. Good. Good. I just find… well, that’s… … Mike, he’s still working great. great. But it’s also a reflection of where I see other tools heading, isn’t it? If I use other tools, for example AirTable is a table that can be manipulated via API. It’s just an API that talks, so… , so… so… so… when you’re talking about more

16:37 when you’re talking about more agents and systems communicating with each other, people don’t want to install definitions everywhere. It only requires working with a URL and a verification code. This is the way the way the world is currently operating. And I think agents are accelerating this need more and more, frankly. So this is the main point, that things are moving in this direction more and more., honestly, I think what’s likely going to happen is , I think I saw something about this called an ” AI API,” where we’ll start writing APIs for agents like we do

17:07 agents like we do now. Do now. Do? And I ? And I think GraphQL is part of that journey. Yes. There will be summonses specifically for the agent. I know that a long time ago, before , before agents even appeared on the scene, Tommy, we used to talk a lot. I remember I made a comment - I have to find it in the podcast - but we were talking about how the world works with application programming interfaces (APIs). We live in the age of programming interfaces, where everything must become an API endpoint, a link containing

17:38 link containing input and output data. It is the most common way to exchange data, and I think we talked before ( this was a long time ago) that if your organization is building applications and does not have a mechanism for collecting, billing, or measuring API connections for data input and output, then you are missing an opportunity for an organization or person who needs the data to access it. The reason I’m complaining about this particular thing, Tommy, is that

18:08 , is that when I look at what XAI is doing, XAI, Twitter, or X. com are doing something great; they have a simple and inexpensive way to access the API. You can simply connect to it, turn it on, and get the publications easily. It works simply. You can give it to your AI: “Hello, this is what I want to use.” ” Go and use this programming interface.” It works simply. important. When I look at all the other social media platforms, YouTube is perhaps a slight exception, but it’s still difficult to set up. But when I look at

18:39 But when I look at any other social media platform, TikTok, Facebook, LinkedIn, Twitch, it’s like, it’s all complete chaos. None . None of them give you access to the things I things I actually need. So, I’ve reached a point where I’m very disappointed with all these companies and how badly they treat you in granting you access to the data they produce and create on their own platforms. There should be a minimum; if you post more, you should

19:09 you post more, you should earn points so that at least at least those posts can be read and you can see how much influence they have. a point. I don’t want to go to the portal page. I don’t want to go to the go to the analytics page. All your analytics pages are ridiculous. It’s just one page from one platform. I need a system that brings all platforms together on one page, don’t you? This is something that none of these tools seem to offer us very well. However, I am only complaining because I find the way way companies are handling this very annoying. I think it would have been a missed opportunity

19:40 I think it would have been a missed opportunity if they had just said, ” Look, put your credit card here.” Here’s $5 a month. Just pay us. Just cover the cost. “Whatever the API costs API costs, it doesn’t mean anything to you as companies. Calculate the cost plus a plus a 10% to 30% profit margin on those calls. calls. Just do it. Make it happen. Don’t do it, because agents will find ways to do these things, and we’ll be creative in how we build things. But, you things. But,, platforms like LinkedIn and

20:10 know, platforms like LinkedIn and Facebook won’t be happy about it because they simply don’t work well with these systems. Anyway, I think that’s another part. Yes. We’ll save that for another episode because I liked it. But I have a news article for you, Mike. That’s what I’d really like. Yes. Yes. Let’s see if you have any objections to this. And honestly, based on the previous 554 episodes, that’s all I wanted in one article. Tada. Tada. Tada. Web TML is

20:42 Tada. Web TML is now available in preview. preview. What again is that the Tabular Form Definition Language (TML), which lets you write code to edit the semantic form, is now available directly in the browser. It’s a browser-based experience that allows Developers can use, view, and modify semantic models as code. It extends existing existing web modeling tools with a rich code editor that provides transparency into all objects and semantic properties. You can explore metadata for the

21:14 published semantic model. You can really boost development efficiency. You can edit code with IntelliSense support and perform batch updates. You can work with AI tools like GitHub Copilot to modify TML. You can modify any property or object in the semantic model. So it’s not limited. You can see what ‘s available. It allows complete control without external tools, and you can again copy and paste code to code to reuse it. reuse it. It integrates with the Tindle Gallery for sharing components; I know there’s a discussion platform there and a release history. So,

21:44 release history. So, Mike, that’s all I wanted; you can use your browser, and there don’t seem to be many limitations yet. , this is the knockout blow to the desktop version. What does the desktop have now that the web doesn’t? ? For me, I love this feature., Tommy, internally I’m doing this… This feature here has

22:15 … This feature here has basically allowed me to redirect my organization away from PCs and toward toward fully Mac-based systems. I… we’re actually buying Macintosh computers for our next round of hardware for our company. You’re kidding. kidding. No. We’ll use Macs and we won’t need them; we’ll have Parallels. We’ll have, you Parallels. We’ll have,, Power Desktop if know, Power Desktop if we need it. But we don’t need it anymore. That’s the knockout blow. And when you combine that with the ability to see agents who can

22:46 see agents who can read the model. They can see the model. You see the model., we have know, we have MCP servers. Tommy, I think I’m done, man. Desktop. To me, there’s nothing the desktop adds that the web doesn’t already have, and I think it’s a smart move. And I think going forward, Microsoft should always build every Power BI or modeling feature; it should come first through the service, and if it gains enough traction, it should be put on the desktop.

23:12 put on the desktop. Okay, I’m going to take my podium here. Go ahead. . Because I’m going to argue with you about this. You may use Power BI Desktop less, but again, that’s only for modeling, not for designing reports. Second, Mike, where do you use your MCP server? Do you use it in the cloud, on the web? Do you use it on VS Code on the web? No, it still works on computers. Secondly, you have to design your report at some point. I’m not sure I’m following you on that point.

23:42 following you on that point. Okay, let me repeat: when you want to use an MCP server, you want to use Fabric skills. You want to use Fabric remote MCP or Semantic modeling MCP. That runs on the computer and not on the web. Not remote MCP. That’s just an HTTP call. You just set it up yesterday, right? But the ? But the semantic modeling MCP runs on your computer. . There’s a remote MCP server you can use as well. Ray also works remotely. You don’t need

24:13 remotely. You don’t need anything there, okay, you need a need a computer to build the code. Right. Right. But you need a Git repository, and most of the proxies I’m building with now have have pull-out capabilities, like cloud code. They have machines ready, they pull, you go to the cloud copy, they create a small virtual machine, you pull the repository, and you do what you need to do, no problem. problem., I understand your point, Tommy, but I think… think… we have the proxies and the systems. If you’re already paying For a standard version, you can’t really call this a $100-

24:43 $100- a-month subscription. You’d have to start paying for the big, professional tools in the proxy world here. But when you do, you get a lot more value. So, I understand, Tommy, that a remote MCP server has some limitations. I checked this out yesterday. A remote MCP server will let you talk to and query the form. So, it’s basically a one-way street for reading from the form only, right?

25:16 You could combine that with a Fabric Remote Core MCP server that has API access to talk to the elements. Now, regarding your point, Tommy, you can read things from the form, extract data from it, and it will run queries on them. That’s what we’re talking about with an MCP server. If you want to modify the form, what you’re describing here is where you would have to write or read certain files, and then push those files back to the server. Oh, Donald, in

25:46 Oh, Donald, in our comments, says, ” What about Excel?” ” Yes, I think we… I have some comments on that. I think there’s something coming up for that later in this conversation. Sorry. But let’s move on to the PowerBI Desktop Bridge, which you praised, and I commended Mike for it. This only works on Windows, and the thing is: there are more and more cases for using Windows. Going back to your point, could you use the use the remote version? Absolutely. You still need things to work with MPM, so you’ll still need to run CodeSpaces

26:16 need to run CodeSpaces or provide some computing resources. I could do it on a PC, and for me, Mike, I really like that feature. But I find my workflow is very MCPS-dependent at the moment, as I don’t type a lot of DAX formulas into the formula bar, and I do n’t use Tim for myself as much as I make sure to set a Tim DLE for a proxy. Because having those skills here, you skills here,, is

26:46 know, is great for me. This would be more… Let me put it this way: this would have had a much bigger impact if AI hadn’t come along. If we hadn’t had functional experiences with Fabric, this would have been a game-changer, really amazing and important, but because My current workflow is with functional solutions in Fabric and semantic modeling. Well, it’s cool, but it doesn’t change anything for me. me. Let me… yes. So, I’ll ask

27:17 Let me… yes. So, I’ll ask you this question. Are you actually going to start using Tindle on the web with your current workflow? I can… I know you have parts. Okay. Yes. Just anything, modifying anything I want to do. If I want to add another column, or add tables, I do that there. If I want to add new data sources, I go to Power Query online like this. Again, Power Query online within the service feels much better than Power Query on your local machine., the screen there. Yes. So,, all the extra features I feel are now now in the service. Now, I agree with you,

27:48 Now, I agree with you, Tommy. The desktop bridge where I can talk to an AI agent AI agent locally and build report objects, that’s a big win. I think it’s still very much like the deeper we get into business intelligence (BI), the more our build pattern shifts toward agents. I think, for me personally, it’s a much better experience to talk to an AI agent and say, ” What’s the I want it, but what do I want to build? “And I can really start to take a step back step back and say, ‘Okay, let’s talk about… I’m

28:18 let’s talk about… I’m now seeing posts on LinkedIn about people building, for lack of a better term,, people are now building the filter pop-up window again.’” People have now started rebuilding it. Why not use the built-in ones that come with… … I know, but people like having their own custom pop-up menus, and , and what? It’s a major nuisance. We have a filtering section within Power BI reports, but unfortunately, you cannot move it to the left side of the screen. You can’t place it at the top of the Power BI report screen like you do in

28:48 like you do in other tools. So people complain, saying: Well, the filter panel is on the right side. Man, just move on. It is located on the right side. This is where it belongs. But also, it shouldn’t be complicated for Microsoft to let you move the filter pane to any side of the screen. As far as I’m concerned, someone should have come up with this feature a long time ago. You should have had this feature a long time ago . Anyway, regardless, I think this is really powerful, and the reason “Tindle” is here, Tommy, is that

29:18 here, Tommy, is that when Tindle pops up, it gives you full access to everything in the model. Okay. . If they add a new feature, or use a user-defined function, you can build it into TIDLE. It’s specified there. Any new feature they offer can be accessed through that. This means that Microsoft can release any new feature within the semantic models and it will be immediately supported by the Tindle editor. So, people are always complaining about Microsoft, saying: Why did you create perspectives on the desktop that we can’t modify? modify? Microsoft can

29:48 Microsoft can now say: Okay, we now have the Tindle editor in the desktop and service. You can go ahead and create new perspectives or update and add to them as you wish. It’s very easy. So I think this is a big moment for the semantic model. I really like this. This is a great step. I love this tool. This is amazing. I am very happy about this. . Do what wasn’t in my predictions for 2026? Do what Do what Power BI needs? A feature from ” Quilk”. This is what we need. I never expected that. What are we saying? What are

30:18 we saying? What are we talking about here? People migrate from other tools. Okay. . But this is the notion that people are stubborn. People are stubborn and they see things the way they see them, aren’t they?, ?, when I switch between different tools, I say: Why does n’t it work the way that that other tool did? Well, it’s different. It’s a new tool. You need to adapt and change your style a little. little. But I will say with minimal effort. In the Filters section, if you allow this section to be moved

30:49 if you allow this section to be moved around the page, I think it’s obvious and should be available. And great, anyone coming from Click will find all the filters at the top of the page, choose what you want, then press play and the page works. Excellent, let’s do the same thing with PowerBI, and let this happen. I think this is a simple thing that is easy to achieve, but for some reason we do not have the ability to do it. I was working for a client on

31:19 for a client on embedded PowerBI, Tommy, and the funny thing is the client was so adamant that the filters section had to be on the left side of the screen, that in the in the embedded application we had to rebuild the side filter page to move it to the left. In any case, we would have adopted the inclusion. We were unable to use the filtering section on the right, in addition to their desire for other other mandatory filters to be applied before the report could be loaded.

31:50 before the report could be loaded. Simple things, aren’t they? Some light and useful things. None of that is included in the built-in PowerBI. Therefore, we built it into the application. Do what I’m saying about that? that? As long as you’ve been paid, that’s one of those things… I’ve been paid. . This is like saying: Look, we can do that. Yes. Yes. Do you really want to spend your money on this? And… it was the right decision for them. for them. We were able to improve it in a way that made it very unique to them. Honestly, yes. That was wonderful. It no longer looked

32:21 wonderful. It no longer looked like PowerBI at that point. It seemed like a dedicated, standalone reporting solution, and it was a good solution, but now they have the technology to do it. Now they understand how to create their own filtering component and how to apply it before loading the reports. Really good things., this is a very powerful design for reports. But again, I feel these are features that should be automatically available when creating and including reports. But again, I don’t have the budget. The feature I’m talking about probably wo n’t bring in a million

32:52 n’t bring in a million monthly active users. Therefore, it is likely that Microsoft will not build it. But this leaves,, ,, some opportunities open to me. As a developer. Because I know how to do it. This means I can step in and profit from these unfinished projects or or unpolished aspects of Microsoft. This is why I built Power Designer and all the tools that come out of it, because there are some incomplete aspects to how to how artificial intelligence and agents work within the Microsoft system. Well, now I can find my own way. Well, let’s hope that

33:23 Well, let’s hope that Spanish in Power BI remains in the preview phase. So, I’m sure there are people talking about missed opportunities. Yes, exactly at the moment. Okay, Mike, let’s talk about something that’s already available. And Mike, a few episodes ago, we talked about Rafen. I think that was in July. Or June, and in Rafen, Mike, you actually came to Chicago and gave a great session to the user group about building with Rafen. You didn’t just build build one app, you built

33:54 one app, you built multiple apps. Once again, Mike, Rafen is a truly amazing feature, you could talk about it more, so before I delve into what we’ll be discussing today, what makes it different, Mike? Why should I care about Rafen? Man, Rafen is the most awesome thing we could get could get. So, . So, earlier in this episode, we were talking about how amazing artificial intelligence technologies are, and how things can be built with them. Tommy, you were saying, man, it would be

34:25 man, it would be really great if I didn’t have to own one. You weren’t saying that, but I was saying wouldn’t it be great if we didn’t have to use a desktop, or why do we need a desktop? Here are some reasons. Well, Rafen is a developer experience, isn’t it? There are many things in Power BI. In fact, I’ve been following a lot of content on LinkedIn lately, and there are a lot of people pointing out that the standard visual elements in Power BI do a

34:47 in Power BI do a good job of many things, but there’s a lot of lot of visual design that people want to create that isn’t within the standard elements, and you can’t do it. It doesn’t allow you to customize it the way you want. So instead, people are now turning to AI agents to build custom visual elements and place them within reports. I love that. I think it’s a great idea. It would be a great skill to add. Regarding Microsoft Skills for Fabric, I would like to see someone build a custom visualization software skill or skills

35:18 build a custom visualization software skill or skills that are integrated within the Fabric skills. That would make perfect sense to me; I to me; I want a want a bar chart that looks like this, or a , or a key performance indicator card, or something presented in a certain way, or maybe an idea for a data slicer. It was a linear layout, and they wanted to be able to drag a strip left and right in this divider to narrow the window. They said they want to display the entire partition with all data points, and

35:49 with all data points, and drag the strip across this line graph, right? right? And actually get a narrow window that filters out the rest of the report page. genius. That makes perfect sense. I loved this. Great idea. I really liked what they were offering here. The data and how to build this interaction between the line graph and the divider into one element would have made perfect sense, but it’s not there, is it? This is not something you can get from the standard visuals in Power BI. So, you build a custom visualization to do

36:19 custom visualization to do that. The idea here is that it used to take weeks to design, code, and build, and now we’ve shortened that to a weekend or even a single day. Custom visualizations within Power BI have become very simple at this stage. You can only speak to an AI agent. He understands how to build it. He understands all the documents related to it. This is truly wonderful. Fabric applications are an extension of this. So, anything you want to build now using Power BI, you are limited to the

36:50 Power BI, you are limited to the visual framework that has been provided, and that is all you can use with Fabric applications. If you want to use Vega, Flint visualizations (a new standard from Microsoft), Plotly, D3. js, or if you want to create any visualization style design, the agent knows all these languages, and you can then integrate and format a report or interactive report. What by that is that the data sources for

37:20 data sources for Fabric applications are currently SQL databases, which are readable and writable, meaning you can extract data from them and input data into them. The second part of this is semantic models. So, if you just want to display the data, you can have the semantic model display the data in the report. This is a one-way road. It is a . It is a read-only version, but it is fast and extremely efficient. Therefore, you will get very good response times when querying querying data through the semantic model. However, this changes everything

37:51 However, this changes everything. You are now building complete complete data-driven websites the way you want. And customize it as needed. This is the power of Rayfin and many people are excited about Rayfin. And I am definitely one of them. I absolutely love him. It’s a truly remarkable tool and an amazing innovation. You can again consider part of it as falling under the umbrella of Fabric applications, where you can build a report using Rayfin that is fully customized or just a simple application with inputs and control slides.

38:21 inputs and control slides. I believe the demonstration you gave gave was about registering your vacation, where you enter data, which is great within any organization, isn’t it? Because you simply go to Fabric or you can embed it anywhere and sign in using your Microsoft Entra. He knows who you are. And you enter that information. Naturally, it will be sent to Fabric. He prepares all of that for you. But the problem, Mike, is that when you have an application, often , often not everyone who needs access to that

38:51 access to that application is within my organization. So the apps are great, but the apps also work just as well as any general app; anyone can download the MLB app, you don’t have to be an MLB employee. And this brings us to this new type of document that talks about anonymous access to data here, Mike. Okay. This is . This is really interesting in what’s happening here. This is a feature that, first and foremost, is not automatically activated by default. It is not present by default. So, in order for this to work, you have to go for it. Yes,

39:21 this to work, you have to go for it. Yes, this is a good thing because it gives those in charge levels of control over it. This means that in order for this to work, you need to go to the administration portal, and then within within your tenant settings, there is a new item called Fabric App Preview. Then, in the Fabric app preview section, there is an option called anonymous data access. Let me be, this looks dangerous. Do you mean anonymous access to data? What does this mean? This looks very bad. Okay, so what this means is that whenever you log into the Fabric app by

39:51 Fabric app by default, as Microsoft does, someone has to authenticate to log into the app in order to be able to do anything, right? If you want to talk to an SQL database, this is a basic requirement. Someone needs to log in somewhere within that app. Good. Hmm, that’s pretty standard when going to websites, isn’t it? You go to your Gmail inbox. They will not give you direct access to Gmail. They will ask you to sign in to Gmail. A welcome screen will appear. You will find the phrase “Log in here”. great. You log in and access the application, right? Then that

40:21 that authentication is passed. Therefore, the Fabric app actually passes the authentication details to the thing you are connecting to. Good. So the Fabric app has an awareness of “Tommy” when I design an app and give it to you. He knows your email address. It knows who you are and how you access this application, which is great because if I have an have an SQL database, it says Tommy can see or not see some data in the database. I know who you are. I know what data you are allowed to see. I can apply row-level security

40:51 apply row-level security or or row-detail security, and the database will not return information to Tommy that he should not see. Again, good system design. This is how we expect it to work. However, there are certain things you might build that do not require a user context. For example, I will give you an example here, an example that I am I am currently trying to rebuild. Hmm, actually, it’s a funny story. I was trying to build this in Fabric applications about a month ago. I couldn’t do

41:21 . I couldn’t do that. I complained to the Microsoft team, asking when I would get this feature? Then they told me to watch out, she was coming. Okay, the feature is now available. Okay then. Tommy and I have Microsoft forms on our website. In our PowerBI tips podcast. podcast. If you go to powerbi tips. com you can read all the episodes. Oh, by the way, you’ve done a big update, Tommy. I don’t know if you’ve seen this yet. So if you go to PowerBI tips, Tommy, and go to see the episodes on the PowerBI tips website. He will go there now

41:51 tips website. He will go there now because he will not be able to resist looking at her now. now. I know that. When you say this, you change the subject. You have a person with a disability. Sorry. Sorry. I interrupted you while you were speaking. You need to go to the go to the website now. Now, when you go to the texts, clicking on them will not take you out of the site. Instead, you will click on parts of the text, and the video will advance with a small video player appearing in the right corner of YouTube. While it jumps to different texts, the video also jumps to the place we specified.

42:21 specified. Therefore, it is more comprehensive with regard to the texts within the site. Anyway, Anyway, well, look at that. Okay. Okay. Good. Okay. Okay. So, now you can, now you can, yes. Okay. This is truly amazing. Okay. Anyway. Good. Now, then, this also means that the PowerBI Tips website supports full-text search. Good. So, Good. So, having said all this, let’s say for example that I wanted to get a get a feedback form. Let’s say I wanted to build a website that says: ” Hello, this is my company’s website. Here’s everything

42:51 my company’s website. Here’s everything about us. This is all our information.” Isn’t that so? 90% of the site is one-way, displaying data only. This is what we are building. That’s what we do. great. Absolutely fantastic. I have many websites that do that. You might want to build an educational website. great. All of this can be applied as a Fabric application. No problem. The problem for me is, what happens when you have a “Contact Me” form? What happens when you want someone to visit your site and send you data to give you their opinion?

43:23 you data to give you their opinion? Another experiment I’m working on, Tommy. I’ve been trying a lot with Replet. Replet and Microsoft are collaborating to build Fabric applications. I said to myself, I’d better learn how to use Replet better. So, I subscribed to Replet. And I’m trying things out there. One of my experiences is building a game. I built a game using Replet, but there’s no way to save the save the records. I don’t want people logging in to play the game. I just want them to be able to set their records and know who they are. Enter a name. Yes, enter a name or something like that

43:53 enter a name or something like that. I have, . I have, as, data that needs to be sent from there; it might be information about that game. Okay, that’s fine. Where should I save that? How can I do that without the game requiring the user to log in through my organization to access it? Therefore, there is a whole layer of ” anonymity” that must occur between the application and the SQL database. Sometimes you just need to run the application, and you need it to connect anonymously from time to time to send

44:24 time to time to send data to an SQL database. You were never able to do that before. This is the feature that does that. Once the admin portal is running, you can specify a property within the table, and there are many details here. The role is “Anonymous”, and therefore that role gets different privileges. Therefore, anonymous anonymous or documented privileges can do different things. I think this is another subtle point about this tool, and it’s really useful. You can define a table, and then that table will tell you exactly

44:54 that table will tell you exactly how to define it. Let me pause here for a moment. I’ve said too much. Okay. I think there is an important distinction here because this is not a substitute for authentication. Let’s be very clear about this. This is a way to authorize the user, whether anonymous or not, to do what he can do, isn’t it? Because you don’t befriend anyone. Okay, wait a moment. I will change your wording slightly. You are not authenticating a user., an user., an anonymous user. You are

45:25 anonymous user. You are simply saying: “I will allow this site to send data to a table without the context of a person or user.” or user.” You are authorizing a user even if they are anonymous. But it is not a substitute for authentication. And this, my friend Luke, is the third sentence in this document here. This is where I get this information. . Anonymous access is an authorization option, not a replacement for authentication. correct. It’s just an option. So there are other things

45:55 So there are other things here that I think we may need to clarify as well. Okay. ., this concept related to the ” communication model,” right? right? certainly. I do n’t want to use Microsoft Forms because, firstly, when you add a form from it to a website, you ca n’t customize its appearance. Secondly, it contains a lot of extra junk at the bottom. It’s funny and not well included. Therefore, the embedding experience in Microsoft Forms is bad in my opinion. It’s just junk., the

46:23 It’s just junk., the only company that has a really good communication model is Google Forms. Google Forms does a great job of letting you format them. It is included very easily. It seems very straightforward. What if you wanted to build a website that included a contact form? The experience becomes completely disconnected between the parts. So, what I think now is that you can build a “fabrik” application that has this concept of models, which is much easier to work with, and then you can include and build the models. I have built a static web application on Azure as a

46:53 on Azure as a form tool. It contains a page for the administrator. You log in as an administrator. Therefore, there is likely an administrative aspect to the matter. On the administrator page, you can configure what the form does, how to use it, and the questions you want to ask; all of this can be configured via the administrator page. Okay, great. Once that’s done, you can publish. The output of that becomes a website that connects to a database to which it sends data. great. This makes it a really good use case for anonymous access. So

47:24 anonymous access. So half of your website is accessed accessed anonymously. It’s the model. It appears on your website. You can include it in an iframe. You go there, you activate and send your information. Boom. The data is displayed directly. And the best part of all this is that I love it because I can now collect data anonymously or ask people to send me information and it all goes to Fabric. It is already in the SQL database. I can immediately start working with it, looking at it, creating notebooks, and doing doing things with the data, which is really exciting for me. And I think, no, and I think the big thing here is,

47:54 big thing here is, let me ask you this. It’s clear that Microsoft is trying to solve the problem of mass-consumption applications, isn’t it? Information that is publicly available, or a publicly available method for people to enter information. Again, anyone can do that. I could put that in a government agency. Think of “web publishing” for many for many state counties; they have their reports on county or county or state data in a ” PowerPI” report published on the web. No authentication is required. All I have to

48:24 authentication is required. All I have to do is simply go to that page and it loads,, loads,, available for posting on the web. Now with Fabric, the Fabric app is much more than just a way to publish content on the web, because it has the ability to input data. But Mike, I want to ask you here: Why is this more valuable than dealing with the slowness or complexity of, say, Microsoft Forms or Google Forms? Because I’m thinking about what I said. My opinion on this matter is

48:55 My opinion on this matter is yes, I can’t format it and it goes to Excel, well, that’s okay. But it still gets the job done instead of having to build an application. And yes, Rayfin is incredibly easy to use. And me, it’s great that you only use one command. She says: ” Set up a contact form or mailbox with a name field and a question field.” amazing. Yes. Yes. Yes. Yes. As for me, I’m going to look at it from two perspectives, because my first thought here, Mike, is OK. Firstly, Microsoft Forms does that and you don’t have to set up a Fabric application or use

49:25 Fabric application or use CU modules, right? Or use any capacity. There is no need to worry about authentication. I can only use Microsoft Forms and deal with its slowness. For me, the selling point here is that Microsoft Forms is great for one-time things like “How are you?”. But the problem is that it goes to Excel. That’s the only destination Microsoft Forms goes to, and Google Forms goes to Google Sheets. Can you contact that and then get this get this convoluted way of transferring it to Fabric? certainly. certainly.

49:55 certainly. correct. certainly. This is where your consumption of CU units comes from, , meaning you will now consume CU units. Oh yes. Oh yes. From that perspective, which is okay, now I have all these flat files and I don’t like flat files., we’re in an era where, if Fabric is really trying to become a background for your applications, use it and make it work that way, and I think that’s the big thing. That’s what we want. The big point here, Mike, is that the Rayfin app you’re

50:25 that the Rayfin app you’re building, if you have inputs where people are submitting information, even if it’s just once, once, you can store that in your Fabric ecosystem. And we are seeing more and more now. And Mike, I don’t know about you, but more of the projects I’m working on and the questions I’m getting in the field that we’re talking about, right, , is how do we focus our data in Fabric? If you’re going to do part of it, it’s really better to do it all so that the tools can communicate with each other. For

50:55 each other. For me, “Ray Fenin” is the main selling point here. It’s not just about building an application to input some information, because frankly I think you’re underestimating it when you call it a contact form, because that’s just one example. It is a very easy example to understand. It’s a simple example, but I don’t want listeners or users to limit their thinking to just a sophisticated communication model. Again, as Again, as with most AI tools, it , it depends on what you want or need.

51:25 need. Okay, what if, let me give you another quick example. What if you were a government company or a government institution, right? So, you are a government institution. A great example. Okay. correct. One of the reasons we use the ” Web Publishing” feature is our desire to include the report within a website and display some data. What if some of that data is located on an SQL server? What if that data,, data,, requires people to choose some input parameters and there is some interaction that takes place here? What if you wanted that data to come directly from an

51:55 directly from an SQL database? amazing. It creates an SQL database. You put the data into it, and then you now have an application that can anonymously connect to that SQL database and display tables or visuals containing the data. Again, you can deal with this in whatever way suits you. So, as, there are two layers to this. I think I’m trying to talk to her here too, Tommy. correct. One layer of this is the established method for implementing the application within your

52:25 for implementing the application within your organization. That’s how I think about it, isn’t , isn’t it? Tommy’s company has users and we want those users to use the “Fabrik ” app, right? This is what needs to happen; they need to get in there. They have to click on login on login and the system actually verifies their identity. Do they have access to ” Fabrik”? Do we know who they are? Can they actually log in to the app? Okay, this is inside information. The other side of this is what happens when you have you have external users. Let’s go back to that

52:55 external users. Let’s go back to that government agency that builds applications and needs to produce and send data to customers that they need to see. Well, if you’re trying to send data to customers, they don’t have an ” Entra ID” account for your application. So, you need to build a second mechanism. There is the concept of authorization followed by authentication that you can take advantage of here in this pattern, where I will build an SQL database with anonymous access, meaning that anyone

53:25 meaning that anyone can access the application. But there is a second layer: if you want to see this visual presentation, or see certain things, we have a user table, and within it we check who you are, what your permissions are, and what data you can see. So you can integrate other experiences here so that it does n’t require adding all these all these external users to your Fabric app. And I think this is another point that seems very logical to me. This is the thing we need. need. What also makes this stand out, honestly, is a wonderful example that just came to mind of a

53:56 mind of a situation in which it would have been useful last week last week when the hurricane hit Chicago; how most most electricity companies have very poor maps for displaying outages. outages. Well, if you have a database that shows the locations of outages and receives reports, then you can get a map. You can design that. Man, Mike, you don’t have many many limitations regarding what you can build. Do you want to add a login? correct? correct? Enter your address here, where are you experiencing an outage? Boom. The data goes

54:26 Boom. The data goes into the database, and now you have this database. You can give it to all your customers. All users can access the site, not “log in,” but go to the site and report the outage. amazing. Now you have a map. map. So, isn’t that great? As for me, we are really wondering where this all lies? Where does anonymous anonymous data access fit in terms of how much I should use it? use it? correct? We talk all the time about the usefulness of something. Is it a great feature? And frankly, if you’re going to own

54:56 own Fabric apps, supporting anonymous anonymous data access seems very logical to me as well. What is the point of sale? Because, well, let’s put a put a simple scenario in here. I am the IT or IT or technology manager at a company, and we have we have software needs, and I just tell them that we can support them through “Fabrik”. I imagine there will be some hesitation. There will be some concern. There will be a feeling that we will not push ourselves fully into this matter. For me

55:26 this matter. For me, I’m thinking about what the main selling point is here. How can I market this? Isn’t that so? Because I don’t think anyone will look at this at first glance and say it’s wonderful. Let’s build everything using Fabric apps. We will be transferring everything from a general viewpoint or using a basic login that people might have, instead of using, say, your say, your email and email and password, right? You can support basic login here without even using Entra, by using by using anonymous access. Give everyone, give all my sellers a

55:56 random email address and password; They need to authenticate the login page with this. You can support this however you like, but any technical manager would be cautious about starting with this, especially with regard to the general consumption aspect, would n’t they? Especially if the data is linked to an SQL database that you SQL database that contains know contains more private information. I’m thinking about how I’m actually going to present this, and I wouldn’t want to say ” contact model,” would I? Because again, a lot of things can do that.

56:26 can do that. So why build a Fabric app to access data anonymously instead of using the other tools available to me? What is the real tipping point or dividing line that makes this more logical? How do you convince a technical manager of this? of this? I want to rephrase your question here. I think it’s not that. I understand the question you’re asking; you’re asking, “What is the decision-making point?” Why use this? I don’t think that’s how I see it

56:56 use this? I don’t think that’s how I see it . I look at this and I understand how to build a static web app. I understand the full capabilities of what static web applications can do. When I look at Fabric, I examine it and ask: How does this compare to the technology I already understand? certainly. Currently, Fabric applications are not at the same level as Azure’s static web applications. It’s not that simple. Hmm.

57:27 . Hmm. So for me, this is another step closer to a gap that I can fill myself. myself. Without the need, it bridges the gap between full-fledged static web applications and what a what a Fabric application can do. Instead of saying, ” Well, we’d better use Fabric apps because they provide anonymous access to data,” here’s why: that’s not the selling point. This is just one of the things I found when I started using Fabric, I started

57:57 using Fabric, I started building it. She started building them

57:59 building it. She started building them one by one. Then I encountered an obstacle. This is the obstacle. The obstacle was: ” Hey, I want someone to be able to send a contact form.” This is literally what I was asking to be built. After 15, 20, or 30 minutes of discussion with my AI agent, I found that they were saying, “No, you can’t.” There is no anonymous access. You cannot do this. You can’t do that. You ca n’t do that. I said, “This is ridiculous.” I want to do that. This seems logical, this should work, shouldn’t it? Therefore, my original knowledge of static web applications made me know what was possible. This was a

58:31 what was possible. This was a missing feature. For me, this further closes the gap to get closer to the point where Fabric applications become more capable. There is another feature that I believe should be a high priority or on the backlog of tasks. The next feature is user-defined functions in Fabric, because once you have them you can run APIs from APIs from Fabric, you can run external APIs, and you can implement security things. So, the entire function layer,

59:01 entire function layer, when I look at Azure Static Web Apps, I find that functions are very central to what we are building, they are everywhere there. there. Now, when you get a static web application in Fabric and put functions on top of it, the world opens up to you. We have the ability to build many many wonderful things. We are in the early stages of this, and I think that once this feature is introduced, we will see another big win here. I don’t know if it’s on the on the schedule, I don’t even know if it’s on the on the to-do list, I just don’t know. But for me, when I look at the world of

59:31 when I look at the world of static web applications and what’s happening in Fabric, this is such a significant advantage that you now you now go back to your Chief Technology Officer ( CTO) and say, ” Look, parity, the decision now is: Should I build static web applications or should I build Fabric applications?” That’s the decision. What’s happening now is that all of this is contributing to raising the water level, bringing that ship closer to where static web applications need to operate. So I think this is

60:01 So I think this is confirmation. We are moving in the right direction. Good? ? We are getting the right features we need to bridge the gap with our current tools. That’s why this is important to me. So, it’s not that I’m trying to convince you to use it for that reason. Does this make sense? No, that’s good. You’ve got me thinking about something here, and as we’re nearing the end, help me refine these three features because I want to categorize them. This might actually be something new. I need you to give me the bronze, silver, and gold for what will convince someone, or in terms of impact

60:31 , or in terms of impact as you indicated. What are things again? You can tell me if you want me to revise or update it. Good. The Good. The three things that Rafen and Fabric applications are building today. One of them says: We need a communication model. We need to send this to people. I will finish that before lunch. lunch. Fabric integration. All the data they need to enter comes from Fabric. Or the other, and the data in the application goes directly to Fabric anyway. It is available directly directly from and to Fabric. Yes. And lastly, you want to

61:01 lastly, you want to log people in and secure this for the organization, understood. You need access to other people from outside our company; we take care of that. Completely separate authentication for users. users. These are the three I thought of. Again, do you want to update? Those are difficult choices. Those are difficult podiums. Umm, yes. Umm, yes. I will choose the person who came in first place. Data should be the first priority. Is this the gold one? Do you mean it’s gold? The gold? The gold medal goes to the fact that the data is present and stored in Fabric. To be honest, Fabric as a software background for Rayen is not very impressive. But the fact that Fabric becomes

61:31 the fact that Fabric becomes the background for everything you need in Rayfin. That’s what I want, isn’t it? I need the ability to create, create, transfer, input, and output data from users to the system. And I want that data to be in the Fabric system only. I don’t want to create many different services and multiple APIs. It gets complicated. Place the data in the same place where all the tools can communicate with each other. This is a great gain. There are many difficulties I encounter when transferring data between

62:01 data between different systems. It’s simply difficult. Put them all in the damn Fabric and be done with it. So from this perspective, I really like how easy it is. For me, that’s the gold medal., silver… I think it will be the nations of speed,, the fact that the agents know how to build it, right? The fact is that agents know how to build any framework. I believe that applications will not be effective without this era of proxies coming into effect. . If you gave me the

62:31 If you gave me the Fabric apps and told me that there were no agents available to me, I would like them. But most people will not like it because it will be too complicated for them to understand. And you write hundreds of lines of code to run some of these things. The fact that I can only describe it to an AI agent: “This is my screenshot.” ” This is my official handkerchief.” “Build something like this.” Boom. Go ahead. This is somewhat similar to what I’d say, perhaps a little controversial. This is the evolution of “Power Apps” where

63:02 evolution of “Power Apps” where “Power Apps” was based on low-level programming. We are in a space where it has become unprogrammed. I don’t think it’s a bold opinion anymore. It . It may not be a bold opinion, but I think there are some people who will get angry and say, “Wow, why are you still using PowerAbs?” It’s like, I understand you guys. But this is how I enjoy working in app development. Accept it as you wish. You may like your old tools, well, it doesn’t matter. I don’t care. I enjoy building apps this way. I build toys. I build

63:33 toys. I build things. I spend my weekends doing that. I build while I walk. I talk to I talk to artificial intelligence and make it build things for me. This is the is the new way of working in the world that I love. I like building this way. If you want your company to act like this, contact me and Tommy. We will teach your company how to do that. We will help you with this because we are changing our companies to be completely AI-centric in what we build, not just throwing throwing AI onto AI onto things. We use it in strategic locations to extract

64:05 value from it. So this is a really big win. And this is another part here as well. Then the third element here, Tommy, is the anonymous user. The unknown again, it’s a necessary feature, but it’s not the one that makes a real difference. This is my bronze medal at the moment. Honestly, I have the same arrangement as you. I thought you would change the silver and gold ones, but I agree with you. Fabric’s integration, and the fact that you can handle your data from end to end, is now

64:37 Fabric’s main competitive advantage overall. No, that will remain my first choice even after the introduction of user-defined functions. But when user- defined functions appear, we will have to rearrange the crowning platform because I believe it will change the final order. I want to share my perspective on Bild today, because frankly I think Fabric’s current integration is similar to Michael Jordan., you can’t outperform him. He is the best in this field. But today, Bild, I don’t think you’re giving it enough credit. He is not too far behind. The fact that I can tell my boss that we really need this

65:08 really need this, and we can secure everything by Wednesday, or in fact, let me give you a prototype before lunch today. Okay. The fact that we can do this integrated work, I think, is something that cannot be ignored. But Mike, what a great conversation today. Yes, I really liked this dialogue, Tommy. I also think that, when we talk about agents and building things, we greatly underestimate the extent of that impact. Tommy, there is one thing I want to point out, I have encountered objections from some

65:38 encountered objections from some customers. I don’t know if you’ve heard this too, Tommy. Customers say: “We will build this entire application and it will take half a day to do so. What happens when we need changes? What happens when we encounter problems? What happens when it does not work the way the way we expect?” I would argue that if you built the whole thing using an AI agent, it’s as simple as saying: I have a programming error, fix this thing. It will be over in a few minutes. So I find that when I build applications entirely using proxies,

66:09 using proxies, you really need to learn how to write commands (prompts). You really need to be able to tell the agent what you are trying to build in an appropriate way. But the agents have become so good lately that they build the application and have so much knowledge that they can update and fix it for you very quickly. So, maintenance and support, Tommy, as long as you connect this thing to “Git” and have it there. If you have GitHub

66:39 If you have GitHub, almost any agent can accommodate and build it. Someone will contact you over the weekend. Hello, the app does something specific and this is not what I expected. You can literally take this order, put it in your agent’s hand and say, “Hey, go fix this.” Pull this warehouse. Make this change. And make another change. And you can do it that quickly. So, weeks of error correction and review time are over. Agents are really good at correcting mistakes and figuring out what’s wrong with things because they’ve had to

67:09 things because they’ve had to do it from the start. Therefore, they are very good at correcting mistakes in things. Anyway, I’d say there will be a discussion in your organization that says, “Hey, we want to use this, but okay, who’s going to maintain it?” “What’s going to happen?” I understand that. Try it on something small and then get past a few debugging issues, because I guarantee that once you get past a few of them, you’ll say, “Yes, this is much better than having to debug myself. We’ll just use this going forward.” Therefore, not only is it easy to build applications, but it is also easy to correct their errors with the help of

67:39 their errors with the help of agents. Anyway, with all that said, Tommy, this was a great episode. I hope you have enjoyed this conversation here. Very detailed information about this new feature coming to coming to Fabric Apps. We are very excited about it. Stay tuned for more. I hope to have a video in the next couple of days in which I am building an anonymous version of these models we are talking about. I’ve been trying to get some some things ready on my end. Therefore, we will try to show you how to build this model using a Fabric proxy and build an anonymous communication model through Fabric applications. Okay, Tommy, where else can I

68:09 Okay, Tommy, where else can I find the podcast? ? You can find us on Apple, Spotify, and wherever you listen to your podcast. Make sure to subscribe and leave a review. It helps us a lot. Would you like to leave a question, topic, idea, or suggestion? Okay, go to powerbi. tipsodcast. Leave your name and question. And please, yes, leave your name. We would like to know who you are. The new form will probably actually have a name field, so they have to do it. Yes, that’s a good idea. You can add and finally join us live every Tuesday and Thursday, at AM AM Central Time on

68:39 Central Time on all of all of Power Tips’ social media channels. And if that happens to Ray Finn, Tommy, it should be,, ,, very easy to do. I should only tell my agent what I want. It should simply be a required field. Very fast. Okay. exactly. exactly. If not, we’ll put in a small AI button that,, ,, creates a name. So, it’s a funny thing that we can use as well. Only if you really don’t want to leave a name, that’s okay. okay. Thank you Thank you all very much. We appreciate you listening to today’s podcast. We hope you have found found our discussion today and all the new news coming from PowerBI and Fabric useful. Okay, see you next time. The metrics

69:12 next time. The metrics raise the level of enthusiasm. Let it be high. Tommy and Mike light up the sky. Dance for the day and laugh along with the mix. Fabric and artificial intelligence meet your needs. Explicit standards. Start the rhythm. now. I feel the crowds. Explicit standards. Explicit standards. Turn up the volume.

Thank You

Want to catch us live? Join every Tuesday and Thursday at 7:30 AM Central on YouTube and LinkedIn.

Got a question? Head to powerbi.tips/empodcast and submit your topic ideas.

Listen on Spotify, Apple Podcasts, or wherever you get your podcasts.