Trusting In Microsoft Fabric – Ep. 502

Mike and Tommy dive deep into whether Microsoft Fabric has earned our trust after two years. Plus, the SaaS apocalypse is here as Claude Code disrupts enterprise software, and we celebrate Semantic Link going GA.

News & Announcements

The SaaS Apocalypse Is Here

Peter’s viral article on how “11 free plugins exposed tech’s biggest lie” sent shockwaves through enterprise software. Anthropic’s Claude Code released plugins that connect to Excel, Word, and PowerPoint—doing what many expected Microsoft to build themselves. The market reaction was brutal:

Thomson Reuters — Down 18% in a single day, 33% year-to-date
LegalZoom — Down 20%
Salesforce — Down 26% year-to-date
HubSpot — Down 39% year-to-date
Atlassian — Down 35% year-to-date
DocuSign — Down 52% over 12 months, 85% from all-time high

Mike’s take: “These $100-200/month AI tools have changed how people work. If you’re not already using AI, you’re in a competitive race—and falling behind.”

AI Doesn’t Reduce Work—It Intensifies It

Harvard Business Review: AI Doesn’t Reduce Work—It Intensifies It — This HBR article explores the counterintuitive reality that AI doesn’t lighten workloads—it amplifies them. While AI handles drafting documents, summarizing information, and debugging code, workers find themselves multitasking more intensely and managing AI agents alongside their regular responsibilities. Mike confirms: “I am physically working more intensely than ever before.”

Semantic Link Now Generally Available

Semantic Link has graduated from beta to GA in Microsoft Fabric. Built largely by Michael Kowalski and the community, this tool enables data scientists to use semantic models directly in notebooks. It’s become a Swiss Army knife for:

Automating Power BI tasks against semantic models
Translating and migrating workloads
SQL and Spark integrations

Mike’s advice: “Microsoft should focus on API-first development. Build the infrastructure, let developers create on top of it—that’s where you stay relevant.”

Excel/CSV Import Deprecation

The old Excel and CSV import experience in Power BI Service is being retired:

May 2026 — Reports from the old experience stop refreshing
July 2026 — No data syncing

This is a good cleanup—we have lakehouses, OneLake, and Dataflows Gen 2 now. Time to modernize those legacy workflows.

Main Discussion: Trusting Microsoft Fabric

After 500+ episodes talking about Power BI and two years with Fabric, do we trust it?

The Verdict: Yes, But It Took Time

Tommy and Mike agree: Fabric has matured significantly in the last 3-6 months. The rough edges from the GA release have been smoothed out. The day-to-day experience—Git integration, notebook workflows, lakehouse operations—just works now.

Mike’s trust list:

✅ Spark and Python notebooks — “By far my go-to. The notebook experience is top-notch.”
✅ Lakehouse — Solid foundation for data engineering
✅ SQL Database — “Works like a normal SQL database. I’m actually building web apps on top of it.”
✅ Pay-as-you-go Spark (Autoscale) — Flexible pricing, no limits on cluster size

Tommy’s trust list:

✅ Lakehouse — “The number one area where I’m finding source of truth.”
✅ Git integration — “The things that were buggy before just work now.”
✅ Direct Lake — Building semantic models this way is the preferred workflow

The Friction Points

Copilot: Both hosts find the in-Fabric Copilot experience lacking. “Every time I see Copilot show up, it doesn’t quite do what I want.” The pricing/licensing model is confusing, and they often use free browser-based AI tools instead.

Service Principals: Item ownership in Fabric is still built around individual users. “I want the service principal of the workspace to create and own the items—not individual users.”

Connections: Setting up connections properly requires more expertise than the Power BI days.

Real-time/Webhooks: “There’s nowhere in Fabric where you can turn on a webhook with basic authentication.” For event-driven applications, you still need an Azure Function as middleware.

Who Thrives with Fabric?

Data-curious teams that blend data engineering and BI analysis
Organizations with leadership that views data as a strategic asset
Teams already comfortable with federated, collaborative approaches

Who Struggles?

Small businesses (under 25 people) — The ramp-up time and complexity may not be worth it
Excel-centric organizations — There’s a significant trust jump from local Excel macros to cloud-based notebooks
Highly centralized IT teams — If everything must go through a single team, Fabric’s self-service model creates friction

The Bottom Line

Episode Transcript

Full verbatim transcript — click any timestamp to jump to that moment:

0:00 Lighting up the sky. Dance to the day to laugh in the [music] mix. Fabric and A. I get your fix. Explicit measures. Drop [music] the beat now. H feel the crowd. [music] Explicit measures.

0:23 Good morning and welcome back to the explicit measures podcast with Tommy and Mike. Hello everyone and I hope you’re liking our new intro because I’m loving it. It sounds so good.

0:34 You’re not sick of it yet?

0:35 I’m not sick of it, Tommy. It’s on repeat all day long on my Spotify account. other things that we’ve done,, all of the songs you’re going to hear on this show moving forward will now be also on Spotify. we’re going to have a number of songs coming out.

0:49 So, make sure you go follow us at PowerBI Tips on Spotify. You can literally look up PowerBI tips or EMP forever is our first initial release. We’re going to be DJs, Tommy. I’ve always wanted to be a DJ. Here we are.

1:02 Dude, I admit though I I don’t like that type of music, but I really enjoyed that one.

1:10 Well, it’s cuz it’s about you.

1:12 There’s a lot. And you also packed a lot of words in that, too. Like

1:16 there you got to listen multiple times to get all the funny lyrics and quips in there. Tommy bringing the fire. Tommy Ribbon Mike.

1:24 Tom Ribs Mike. Everyone explodes. Yeah.

1:27 so good. [laughter]

1:28 Yeah, dude. I How many songs you got at this point?

1:33 I think so there’s like a little bit of mix of like personal and like fun work businessy things. I think I’m up about 45 songs now.

1:43 Yeah, it’s been I’m having so much fun. And there’s a little bit of so my you’ll see the particularly on this PowerBI tips Spotify channel. It’s a little mix between like super nerdy data lyrics all electronic dance m music pretty much. It’s all like high energy bubbly my kids call it bouncy. It’s like bouncy music. It’s so bouncy. So that’s like what they do.

2:08 It’s bouncy is a good way to put it, dude.

2:10 It’s bouncy, right? It’s happy that that energy high energy stuff. And then I’ve been exploring what would what it would sound like. So one thing is I like guitar solos, Tommy. I like rock and roll. I like heavy stuff. But I don’t like any of the lyrics on any of this stuff. I really I literally could just just give me like huge drums, massive guitars, and just solo upon solo upon solo. So I’ve also built another album called Silicon Metal. Yeah. Made by AI.

2:38 Silicon Metal. And it is all AI making what I would call metal music without any lyrics. So I can listen to it all day long as much as I want and just get that energy through the roof., I’m having so much fun right now. AI is incredible. I’m having a blast. All right, let’s get into our well news here. But before we do that, let’s go into what our main topic will be today is just talking about do we trust fabric? We’ve been at episode 500. We’ve been talking about fabric for well over two years now. how does it look? What

3:11 what does it what do we feel about it? are there any friction points? Is it getting better? What’s our trust level in fabric? How if you’re a new user and you’re like I’m not quite sure about fabric. Where do we go from here? So we are going to definitely dive in a bit more in this topic and talk more about what do we find that’s working for us in fabric. Let’s talk about how we trust it or areas we’re not trusting as much. So let’s we’ll dive into that in a little bit. Let’s jump into some news articles. Tommy, I got some fun ones today. I believe

3:37 I I do that.

3:38 Okay, so I’m going to throw down a couple links here. One of them is al is on Twitter and I don’t even know how to say the name of this article. but Peter, I can say his name. So Peter is his name and Peter is talking about the SAS apocalypse. Sassalypse. SAS apocalypse. I don’t know how he says it in the article. It sounds like, I I won’t get it right. How 11 free plugins exposed tech’s biggest lie is the article here. And I found this article to be very interesting. Yeah. Okay. So really the

4:12 premise here is Anthropic one our our lovely AI agent person out there. Anthropic is building Claude Code. That’s the the parent company of what we use all the time. Tommy, you love Cloud Code. You’ve seen a lot of that. I play with it a bit. Their models. Opus is one that I use heavily., but since the release, they released a plugin basically that connects to all of Microsoft’s products. It’s in Excel, it’s in Word, it’s in PowerPoint. And I think there’s like an elephant in

4:44 the room here around this. People are finding this is incredibly valuable and wondering to themselves, why didn’t Microsoft just release this? What happened? So there’s this whole idea of like Microsoft has this like very,, longstanding security and everything that’s got to move quick with through security all these different levels. Yeah, I agree. I understand. But cloud code shows up, builds a plugin, drops into your existing tooling, and oh my word, it’s saving people tons of times all over the place and can build things at very rapid

5:18 rates. Let me I’ll just say something here, Tommy. This is a company that understands the importance of creation, the importance of working with an expert and helping them build something. That’s where AI is really good at. It’s good at building things and doing repetitive tasks all the time. I don’t want AI to go autogenerate me a bunch of insights or things that I should be thinking about. I want the AI to produce most of content, right? Go build these files. go build the mundane things. I want to tell

5:53 it what I want, but it should just go build it.

5:57 You are bringing up I have a great news article for next week., I just found it. It was all over in terms of basically what’s happening now too. And honestly, not with the coding and I so much more than just coding now with the building of things. before it was like yeah A yeah A yeah A yeah A yeah A yeah A yeah A yeah A yeah A yeah A yeah it could help but you’re we’re finding for most developers they would have to admit today that AI does it better than them at a faster rate and I think yes so the article’s name is something big is happening and

6:32 basically the idea here is if you do remember when co the corona virus hit and like the two months before it got real like

6:42 yeah this seems like something but I’m I don’t know, you’re like, “It’s not around me.” And you just thinking some people were over,, like,, it was overblown, yada yada yada. And that phase of,

6:57 yeah, I don’t think it’s going to be a big deal. And obviously, it was a big

7:01 this Yeah, go ahead. I was saying this is going to lead this is going to lead in very well to my next article. Keep going, Tommy.

7:06 Okay. Yeah. And really if you’ve been working in AI at all, but I think especially thank goodness we’re not just only developers, but at all if you’ve been utilizing AI, you’ve seen this rapid incredible growth not just in this ability to talk to you but what it can do and that is where things are really occurring now where it’s happening so fast

7:36 I’ll pause there. Yeah. in this article, it talks about the market reckoning. the fact that this $100 or $200 a month tool has been released, Claude Code, Claude Co-work, this tooling is now existing in the marketplace. It took a lot of sassified programs or companies and they took a massive hit. Let me give you some of the names. Thompson Vers went down 18% in a single day and has been down 33% year year to date. Legal Zoom down 20%., some of the other big ones here, Salesforce down 26%, HubSpot down 39%

8:09 year-to- date. Alassian down 35% year-to date. Docuine 52% down over 12 months. 85% down from alltime high. ASA 59 60% down over 12 months. These are crazy numbers, Tommy. And what I think is happening here is these tools were around buy our seat, we’ll give you a software product that would be very difficult for you to make. That’s that’s the principle here. And so these this cloud code and all these new AI agents are just now entering the market. We’re

8:40 just hitting again we’re still in the early waves of this. It’s coming in and it’s changing how people work on things. I saw a Super Bowl ad talking about I don’t remember what company it was, but they’re talking about building web apps or building apps right inside their office.

8:54 Oh yeah. Yeah. We talked about that.

8:56 This is no this is no new concept. This has been around like power apps, power pages have been around forever for business users to build things with low or no code solutions. It’s just that we just went to the the 11th degree. We’re we’re on we’re we’ve upshifted a gear here and now we can do all this stuff with AI which is incredible. So, I thought this article was interesting. it basically was how do you defend against some of these companies that are being hit hard by AI. I would argue if you are not already using AI, so I think this way it’s a

9:28 competitive race, right? Your company needs people to do something. There’s you can’t get rid of everyone’s job at a company because then there’d be no one left to do anything. And I think that’s not a good reasonable explanation of what will happen with AI. there’s some experiments being made now with a company called Macrohard where they can build the entire company from AI agents. Ah, I don’t know. We’ll see how it turns out. Proof will be in the pudding, I think. However, I think if you’re looking at a race here against you and everyone else at your company,

9:57 the people that are going to hang around the longest are the ones that are understanding how to use AI and the couple use cases or a couple people that I’ve seen really adopting and bringing AI to the forefront of what they’re doing.

10:09 They’re building many applications. They’re building AI agentic like stuff with their code. They’re not afraid of code to step in, right? They’re bu they’re stepping into problems in the company and solving problems by building solutions around them. They’re problem solvers. So I think for me what I’m looking at here is like look there’s going to be this wave of AI. You better embrace it. It’s coming whether you like it or not. And your goal should be learn how it works. Figure out how to build many products, applications, web apps, little things

10:41 that tools that help you do your job better.

10:44 And that’s one. I think it’s fun to create stuff like this. I just really enjoy it. I think it’s super neat. But in addition to that, you got to be the one at the company that is showing people, building stuff, creating new things, solving problems with these kinds of tools. And then I think then you’re indestructible because why would they let you go if you’re building things and solving problems,

11:04 right? Well,

11:05 with the AI, the thing that’s coming like you have companies need AI integrators. Like I’m already seeing it in my business. I’m having a lot more conversations around demos and things and how to integrate AI into applications and appdev development with PowerBI.

11:21 Mike,, we’re really just getting to the point and again, I think we’re just going to be monitoring this more and more. I I really I’m going to save the article for next week, but our dreams what’s possible is closer to us than never than ever before. But at the same time, you almost feel relevant, too, because there’s,, I don’t want to use the word disruption. I never lived in Silicon Valley, but my gosh, Mike, I’m doing stuff throughout my day where I would have to hire someone. I would

11:53 have to hire multiple people. I could replace myself for it

11:57 like and there are things that I’m doing now that

12:00 I used to find doing like writing documentation or researching I’m tending to done

12:08 and it’s done better which is the crazy thing. I just you got me inspired. I just created a skill for how I write my contracts and because I do it in markdown.

12:20 I’m like I wonder I know. Listen listen listen. Good.

12:24 I’m doing the same thing exactly as you speak right now. Keep going. I I love this. I’m doing the exact same thing in my business as well cuz I told you Tommy, write everything you can as much as you can. Don’t use any word docs. Just write everything in Mark.

12:35 So, since I already have all these markdown files and they have a format like I wonder and I told Claude, I’m like, “Hey, build me a skill. Here’s my repo, of all my contracts.” And it’s pretty awesome, dude.

12:53 and and also what with AI and stuff like this, you can actually build yourself like a a client estimator tool, right? Really fast. Hey, here is my transcript from the meeting. Here’s the summary from cat chat JP Copilot. Here’s all of my historical statements of work. The the bot can have access to all these things. It knows and it can read through everything and get context for whatever it needs all right there. Boom. Read it all. This is incredible. And so I think if with adding so inside your markdown file area, Tommy, what you should

13:26 probably do is you should probably have a a bot go through and write down an agent.mmd file and say summarize all my old workers of statement. What types of activities or key exercises were done? Build out a company profile based on my past work. What do we do? And then when you ask it to build a new statement of work, it can go there, read that first, and then understand, okay, these are the projects that we’ve done with different skills. I can go look at those files and get information directly from them and draw from them and put them into a new statement of work. This is gamecher for

13:59 Yeah. I could keep talking about this.

14:02 All right, let’s move to the next one.

14:04 Yeah. let’s there are more things to talk about surprisingly.

14:08 So, let’s go to this other article I’ve got here. I got another one from Harvard Business Reviewed. AI doesn’t reduce work, it intensifies this. I’m gonna go back to this article, Tommy. We’ve already talked about this one once before. I I’ve read it twice now. I am very convinced that this is a very true statement., and this this article is really good. I just want to bring it up again. I feel like I’m working more intensely on things than ever before. I’m multitasking a bit more. I’m giving agents and bots to do things on the side. I I can focus my attention on something, have it the agent or bot solve it and then move on to the next

14:40 thing. I can’t go I can’t argue enough for this article. It’s really good. I like it. I think this is how I am physically working on stuff. I’ll just I’ll leave that one there. Tommy, you got some articles, too.

14:50 Yeah., if you guys have not seen it, it will be in the podcast description. This is this may be a series, this article, actually, Mike.

15:00 supercharge. Read the article title. Let’s let’s hear it.

15:03 no, I’m saying the article that you’re talking about again, the AI doesn’t supercharge. We may have to make a series out of that.

15:09 I think you’re right about that.

15:11 They’re, we got two PowerBI and fabric articles. So the first one is semantic link baby is now generally available in Microsoft fabric. So this is something that has been available. It was in beta but now it is in GA and semantic link was simply created to enable data scientists use semantic models directly in notebooks which is something that I I’ve always touted for. Everyone always wants the the developer experience of the raw

15:44 data. But why? Because you never had a nice way to connect to structured data. We can automate PowerBI task in semantic link against semantic models, translate the models and even migrate workloads. There’s a data engineering operations can use semantic link for SQL and spark t tests and semantic links built by really the community and led by Michael Kovalski no Kowalsski

16:09 Kovalsski K

16:12 Michael Kovalski I’ll say it go ahead

16:14 yeah so they’re really leaning into this even though

16:18 it was one of their employees who really just focused on it that’s usually when Microsoft the most successful

16:25 there’s a difference here Tommy. So this is you’re talking about semantic link

16:30 which is a project. This is talking about semantic link. Semantic link is part of the product and semantic link labs is the feeder for what features are being used most that gets sent over to the actual semantic link object. So semantic link is like a a library you can just go use. This is great. I’m happy this is becoming fully supported. It does so many things and it’s doing things across not just the model side Tommy. is doing things across the model, the fabric API report side, like it’s doing so many things. It’s just a really

17:03 good Swiss Army knife. And to be to be honest, Tommy, this is what Microsoft should be working on. Microsoft should be working on tooling like this, building things that aid. a lot of times we have a lot of,, Microsoft spends a ton of time on UI, UX development, user experience. Okay, you get some of that stuff right. But the strength of Microsoft is having an atscale rockolid API layer that lets you do anything you want to build. Let the take the take the mentality of what you’ve been doing for years, Microsoft. You’ve

17:34 been letting people build software on top of your products for years. That’s exactly what you should continue doing. Build the APIs, build the infrastructure, build everything has to be an API first mentality. If you can do that and bring things like semantic link, which are tools, all that does is enable developers to build more creative tools to what they need. You will stay relevant longer by having all this stuff. And now, Tommy, I’m using bots. I have an agent here on my computer that is helping me create items and artifacts in fabric. Like I’m I’m actually doing

18:08 I’ve given it access. It’s allowed to talk to the APIs. It can make me a notebook. I can say, “Hey, agent, go build me a notebook and put this process in the notebook.” It writes the whole thing in Python.

18:22 Here’s how to connect. Here’s how to do the transformations. All the cells are written and it documents the whole thing for me. So then I can go through and make tweaks to it after the fact and see what it wrote and test it. But this is incredible, Tommy. Just alone having the ability to do that is is awesome.

18:38 Yeah. And they they have actually really taken the official semantic links and they’ve actually taken a lot from labs too. So it is nice to see that. It really is.

18:46 I love it. It’s really good.

18:47 All right, you got one more.

18:48 Finally, something sad.

18:51 There’s another deprecation that has come our way.

18:55 They took away our natural Q&A

18:57 and now we have more. Yeah. So deprecation of the old Excel and CSV import experience into PowerBI. So this has been around since the very beginning and now it’s gone. Excel and CSV files are really removing that ability to import into the PowerBI service. by May of at the end of May of this year the reports that building from the old experience will stop refreshing in July. there will be just no data that’s

19:31 syncing. So the if you want to create a report, it’s really that get data experience.

19:37 I’m okay with this honestly. We’ve got we’ve got fabric now. We’ve got one lake. We’ve got places to put files. I can still use data flows gen 2 to go read the file from SharePoint from Lakehouse. I don’t really care. I’m not too worried about this one. This this one to me just feels like it’s been slowly dying and people are just going to use it less and less. So I think this is a good deprecation. I’m not I’m not too worried about it. Do you have anything, Tommy, that you have the Are you using a process today that’s using this feature right now,

20:08 Mike? I have not touched these in so long. I used to really try those once like, “Oh, this is cool.” But yeah,

20:14 there’s my point.

20:16 Microsoft, they’re really going through and they’ve been cleaning house with a lot of legacy products.

20:23 That’s good. Let’s get rid of the stuff that you don’t want to maintain. Let’s get you off of those features. Let’s deprecate them. Get them out the door. Let’s focus on what we’re good at. Make more APIs. Let’s build better experiences. I like this. This is good. I’m okay with this one. This one This one feels good to me. I feel like this is a a bad deprecation in any way, but it is good to call it out. If you have something you’re using on it, it will no longer work.

20:44 For the first time, we’re seeing a lot more things being removed than I think added. And it’s good. I think we’re just really move a different direction. So

20:53 I I maybe would argue Tommy like the the richness of the features the or the spans of the features that fabric can provide. I think it’s pretty complete. You’ve got real time, you’ve got pipelines, you’ve got all the databases that Microsoft supports and uses. They’re all available. They’re they’re in the ecosystem now inside fabric. I don’t think we need more things. The Swiss Army knife is full. I have I have all the tools that I want. What we’re doing now, I think, is sharpening them, right? We’re making we’re honing the edges. We’re making it more effective. I think that’s I’m liking this. I like this next stage of

21:26 like, okay, we we’re not trying to,, Gartner our way into another feature. [laughter] So, there’s like a term,, this is a Gartner feature, right? This has to be a feature we have for the Gartner review every year. I think we’ve got all the features that we need. Now, it’s just refining those features and making them more seamlessly integrated together. I think that’s where we’re starting to see things move, which is exciting. All right, that being said, Tommy, let’s get into our main topic. So our main topic today is trusting in Microsoft fabric.

21:52 Where do you want to take this conversation today? How do you want to how do you want to pitch this?

21:56 Yeah, there’s a few things as I was going through and just thinking about what does it mean to trust something? And I think I took it really on the landscape of how we looked at PowerBI in the beginning. has our trust changed from the beginning where is something reliable? Is it something that can be used in multiple scenarios?, small business, enterprise, individual.

22:20 is it dependable?, in terms of all the things that I needed to do and honestly, the question that I really want to ask more than ever is, do we have the same, optimistic outlook to Microsoft Fabric as we did with PowerBI when it came out?

22:37 Good question. let me let’s talk about your first points here, right? You said,, where do you use it? where do we where do we see it fitting inside the spectrum of things? Is it going to fit the small, medium, or large businesses? Let’s start there. I’m going to say yes. I feel that I’m a small business. I pay for fabric, but I use the smaller SKUs. It’s very affordable. I get a lot of value out of it. It helps me with my standard reporting, and I’m not running into any limits. Now, I do have to be careful. I can’t go out and build crazy big jobs, but I also am a small business. I don’t

23:11 have crazy tons of volume of data. It’s just not required. I can get away with a low a lower-end skew, which I think is affordable. That’s the point of this. The point is you need to be able to have a tool, a solution that you use a little, you pay a little. You use a lot, you pay a lot. That seems fair. I think here is what I I would argue that’s really I I trust it in that regard. One comment you also made, Tommy, I want to maybe throw an idea back to you. here’s my idea, Tommy. When Fabric came out, when it just got released, I believe it was around going GA in September, I feel like it was the end of the year,

23:45 Microsoft either had announced it or went GA with it or something. I can’t remember exactly what happened., but Microsoft went GA with Fabric and I said, “Okay, it’s okay right now. There are some bumpy bumpy or rough edges on it right now.” I said, “Give it a year. Give it a year and I think you’ll really enjoy it.” What do you think, Tommy? We’re we’re probably two years out now. One year out definitely from when we had that comment originally. What do you think? Is it is the is it still rough around the edges to you?

24:15 No. And I think I’ve honestly this has really been changing probably I’ve seen in the last three to six months where you’re seeing just something more ref like not just more refined but

24:28 I agree. I agree with that statement.

24:30 It’s the little things too, Mike. It’s also like the bugs that would occur like on data flow when it came out the in the browser you’re like I can’t

24:38 I don’t know what it’s doing. Yeah, agreed.

24:41 So I think what we’re seeing now is from a user interf from just the usability point of view it’s just more and more comfortable. But also too I’m finding more and more that I am preferring actually this is a perfect use case. my wife, she’s gone back to the workforce and they’re using,, like this, they’re looking to do time sheet stuff.

25:04 And she’s responsible for it.

25:07 And I looked at what they’ve been, this company’s been doing, and it was all manual input into Excel, even though they’re doing this time entry and all this whole to-do. And I just looked, I’m like, “What are you guys using to do the time entry?” He’s like, “Oh, we’re using Clockify.” I’m like, “Okay.”, so went down a bit of a rabbit went down a bit of a rabbit hole like I’ll make I’ll impress your boss for you and I instead of building it in a semantic model, I was just like I’m just going to connect this to pieces and API to a like house and it’s just my preferred way of

25:41 going to really starting with data.

25:45 I never would have dreamed back in the day of the way I’m the mindset of going to Python before Power Query. Tommy, did I I feel like I called this one on you. Like you were a bit hesitant on me for a while and I was kind like you got to really try it. Like I think it’s I think you’re gonna and and here we

26:02 I’m glad you’re saying this because I think more companies need to focus their workloads on Python Spark notebooks. I think that that part of the product super solid, super solid there. I really like that part of the product. I think and that’s really good. So I’m very much a fan of that part of the world for fabric. So that’s one that I really really enjoy

26:24 any so and I think so too when I’m looking at organizations there. Can you hear me? We’re all good.

26:30 Yeah, we’re good.

26:31 Okay, you got disconnected for a second. But any I’m also seeing too the the team aspect the git. I’m not having issues with git. I’m finding just the more preferred workflow.

26:43 I would I’d agree with that one. Git had felt really buggy in the beginning, like it would just break sometimes. I have found one weird thing around git that I don’t really love.

26:50 If you have a workspace with lots and lots and lots of items in it and there’s lots of files, your first commit is not very smooth. So that’s one bit of feedback around the git. Like if you’ve been building on a workspace that has just a ton of artifacts in it and the files there’s a lot of files and it’s just large in general, the literally git check-in has a limit on how much stuff you can check in. And so at that moment it says I can’t do it. I need a puke. And so what you need to be able to do on first commit is I’m going to commit not everything. I’m going to only commit

27:22 some things and not everything and then slow. So Microsoft needs to do a little bit better on that first commit on large workspaces. But other than that, Tommy, I’m with you. That experience is awesome. I really enjoy

27:33 Well, and too, I remember there you get the get error in the workspace like, hey, we can’t merge these things. You never knew why. I’m just finding the things that were not features that were completely foreign but things that yet we had to change our work. I think it’s the day-to-day tasks. How do you work?

27:52 And I find that’s the things that fundamentally shifting. So yeah, let’s go into the trust has definitely gone from we’re in this beta project that everything’s so new and just be careful with it to a point where not just because it’s the only thing available, but very much so, Mike, because we’re finding that it’s a better workflow. It’s a better experience and it’s a better way to use data. [snorts]

28:20 Let Yes, I would agree with that, Tommy. And I think this is we are fabric has blurred the line even more between data engineers and BI analysts. It it just keeps blurring the line more and more. You can give business users so much capability with bringing in files and shaping data. I think it’s a win. I I absolutely love it. One thing I’ll also observe Tommy when fabric SQL came out I have just now started building applications like literally a web application on top of fabric using Azure fabric SQL as the

28:53 database so moving my transactional system directly into fabric interesting I’m finding good success with it

29:02 usage seems fairly reasonable I can still use my SQL database on a smaller skew it’s not too bad so I’m monitoring things I’m doing some real testing right now of actually moving your transactional system into fabric and seeing what we get there. Really like this. One thing else I’ll observe, I really like what the Spark team is doing, Tommy. This whole pay as you go model with Spark, like it’s called Spark Autoscale.

29:26 I don’t like the name. The name is totally junk. Look, the name is wrong, but it’s pay as you go Spark is what it really is, where you link up your Spark environment and say, “We’re going to pay just for Spark and we’re going to pay for that through Azure billing.” So, you get to use as much Spark as you want, as big a cluster as you want, no issue. When you turn that on, you can use the Spark engine as much as you need to, but then there’s just an F2 that can run everything else. You have access to all the data in the lakehouse. Another pattern I’m finding really interesting, Tommy, is do the data engineering in an F2 or a smaller fabric skew, lakehouses,

30:01 pipelines, notebooks, all the fun things that we like to to work with, and then you import the data to a semantic model that was in a pro or premium per user workspace. You want you want to you talk about efficient modeling, efficient costs. For 150 bucks, 160 bucks a month, you can get an F2 skew, load in a bunch of data and you can make a semantic model that is up to a hundred gigs for 20 bucks per user. This is a great deal. Like the amount of that’s wow, I like this a lot. this is something that I

30:34 really am thinking here that’s going to be more organizations need to start thinking through., also Tommy, I’ll just maybe point out here. I feel like there’s I’m going through this and now where there’s a development pattern of just build it now make it work get it running find that we have value added to it and then come back and optimize later and then we bring out cost then we optimize and tune the solution I feel this makes it very easy in fabric I can easily get in build stuff that I want and then come back and optimize it for use and if something doesn’t work I can just throw it away and say I don’t need that process anymore let’s get rid of it that’s not

31:06 useful like that’s what we need speed to take ideas into real products, it’s interesting, Mike, because we’re talking about all these amazing new and advanced features, but that is the opposite though of PowerBI was because I think PowerBI’s biggest selling point even in the very beginning was its simplicity.

31:29 Five minutes to wow. It was literally a five by

31:32 they called it a five by five, right?

31:35 Five minutes to get started, five minutes to wow.

31:38 Yes, I agree. So this is I think this is an important distinction here because when you think about small businesses and when all the things we’re talking about with notebooks and the warehouse they Microsoft has made that I really to me as easy as possible to

31:54 jump in but at the same time I’m looking at honestly perfect example my wife’s company is a small design company

32:04 you cannot expect them like they it would make sense for them to use PowerBI I to do to get that data but let’s say I’m like hey you guys are going to spin up a fabric license you’re going to put things in a lake house and that is a lot more of like a harder in terms of a barrier to entry or just that ramp up time I could build for my for a small company right now a single report off of probably one system that they’re using using power

32:36 query done and it works. Yeah.

32:39 And and it’s not even a demo case., that’s what we did.

32:43 But correct.

32:44 If I try to do that in fabric, it’s like, well, you don’t have access to that. I can’t really share the file. U also using a Python notebook and we’re doing all these things. So that is one of the biggest things when it comes to I wouldn’t say as much anymore adoption, but the when you think of the flexibility of the product.

33:02 Agree. All right. This since flexibility is happening here, we’re going to be flexible with this show and we’re going to have a hard transition into something fun., I’m going to Microsoft FabricCon in Atlanta. Tommy, are you coming, too?

33:14 I am going to try my hardest because I believe you’re speaking, correct?

33:18 Well, I’m definitely speaking, so I’ll be there speaking. I want to invite everyone else to go there., so make sure you go check out fabriccon.com, which is Microsoft’s fabric community. I want to push you people to there. If you are not already going, I would highly recommend you go register. This is going to be a SQL conference and a fabric conference all rolled into one. I think it’s going to be a pretty big show and I’m looking forward to it. I’ll see you there. It’s going to be March 16th through the 20th in Atlanta. And in honor of that, I would love to also throw down a song here. one of the

33:50 things we’re going to do on the show here for a little bit, just because I’m having fun with it, I hope everyone else enjoys this too. I’ll give you a quick little interlude here in the middle of the show., this is in honor of traveling and things as well. I’ll throw this out here, Tommy. And,, we’ll we’ll get some music going here as well., we’ll do a little intro for our music. This is called Expense It from PowerBI tips.

34:25 [singing] [music]

34:29 I’m really just looking for Tommy’s reaction for a lot of this song. [singing]

34:59 dude. This is too much.

35:01 Isn’t it great?

35:04 I’m having so much fun with AI. This is one of the things that that I’m doing now. Who is

35:16 Hotel with a view king bed. So grand room service caviar on the company’s hand. Mini bar rated chocolates and snacks. It to the boss. No looking [singing] back. [music]

35:36 That’s my favorite line. Hit it. First class. Here I come. [music] Recyle expending everything [singing] mile after mile.

35:57 Bit at a loss. Mike,

36:00 it’s pretty wild. Meetings in the morning but even mine. Fancy spot treatments. Feeling so fine. Souvenir [singing and music] for the fam. Call them gift t-shirts and keychains. Expensing the lift. WHO GAY? WHO GAY?

36:11 WHO CARES HOW MUCH it [singing] cost? My boss will pay for it. [singing] [music]

36:25 Dude, what are you doing at FabCon?

36:27 Singing my song. underainment. [laughter] But wait a sec. That extra dieser [singing] wasn’t really divine. Nah, it’s all good in a policy book. Just staple [singing] receipts. [music] No second look.

36:48 Back at the office. Stack of paper so tall. Bos signs all blind. Approving it all. Next trip already booked. Can’t wait to go. Living the high life on the company’s door. [music]

36:57 Come on. Come on. It was pretty good.

37:07 Expensing everything mile after mile.

37:13 All right, dude. All right. All right. I You are Maybe you I think you’re going to have a new career.

37:19 Well, if the data thing doesn’t go well for me, at least I’ve got something going for me there.

37:26 All right, there we go.

37:27 Great. Okay,

37:28 so fun little interlude. I figured that was in fun honor of let’s go to FabCon together. I will admit though, Tommy, I have been waking up in the morning and been singing this song. It’s the chorus is what an earworm. It just gets stuck in my head. I I I keep thinking,, mile after mile. The lyrics are just so good., super fun. We’ll throw out a couple fun songs here and just mix it up here a bit. thank you all for enjoying and indulging our fun little song together. If you want to go listen to it yourself, we have a lot of other songs as well out on Spotify. Enjoy. This is just for fun. This is for

38:00 community. I tried to put some of these fun songs out on the fabric, the R fab the R the Reddit thread for PowerBI and it got deleted. I’m like, “Oh, come on, guys. Live a little. Have a little fun.” They felt it was too spammy, I guess. Whatever.

38:15 Really? Oh, yeah. like, forget that. I’m going to put on my own podcast. The world’s largest podcast. We’re going to put it out there. We’re just going to get some news out there about some fun music., all this should be fun. It’s in just hope you enjoy some of it as well. Let us know in the comments what genre you like. Maybe I’ll kick out a couple songs in a genre that you enjoy. That being said, let’s get back to it. Tommy, let’s go back into your conversation around trusting in Microsoft Fabric. You’re you’re lost. I literally lost you now. You’re like, I’m done. Let’s [laughter]

38:46 All right. Let’s Let’s do this. Let’s do this. Okay. where were we, Mike? Fab

38:51 picking up around. We’re we were talking about Fabcon, but we had gone into working in Excel’s documents using parts of fabric and just talking about the complexity, right? Going from PowerBI, which was supposed to be five minutes to wow, 5x5, and then moving into fabric, where we have a lot of other options and things to build.

39:08 Yeah. Yeah.

39:09 I was just maybe going to point out here, one point I wanted to make around this topic was I’m really interested in understanding if I was some listening to the podcast, right? If I’m someone else, what parts of fabric do you like? which parts do you trust? Because I don’t assume you have to like like all of it at once and your business or whatever you like to use. You’ve got certain experiences with different parts of the product. So Tommy, back to you. Which are your most solid parts of fabric that you trust?

39:37 So honestly, right now it’s the lakehouse. I would say that is probably the number one like area where I’m finding just,, source of truth. the quality of it. Not quality, but just doing direct lake building that way.

39:54 Is so is so nice. But I always say the lakehouse is just like that is now the core component of really what I’m building.

40:04 well, I’ll I’ll stop there. What about you? And I think I know what you’re going to say, but

40:10 No, I’m going to say I’m going to say it’s going to be Spark and Python, right? So Spark and Python is by far my go-to part of the solution. I love working in notebooks. There is some really neat experiences around like you have SQL databases, you have SQL data warehouses or the the data warehouse experience. Great. You can write SQL num. I just don’t like how I don’t like how the experiences work honestly. Like I don’t like having to write a single statement statement SQL statement on a line code editor and not be able to like save that statement, save the output and go to another cell and write something next to it. The amount of time I’m I’m writing multiple SQL statements side by

40:43 side and have different tables and I I just scroll up and scroll down. The notebook experience for me is just top-notch. Whether you’re writing TSQL, whether you’re writing Python, I absolutely love it. my only drawback from Fabric Tommy is this whole co-pilot experience. That’s the that’s the one piece I feel like we’re not quite dialing it in right. And every time I see co-pilot show up on top of the experiences, it doesn’t quite do what I want it to do. No,

41:09 it’s good for asking questions. It’s good for having context, a little bit of context about what’s happening in your workspace or whatever you’re working in. So again, one of the ones I use a lot is notebooks and Python. Well, I could either use the fabric one that charges me compute units or I could go to my browser and use the free one that’s part of my browser and ask it my questions around Copilot. Hey, what does this do? How do you write this function? And then I don’t use any CU units from the notebook. So the whole pricing licensing model of fabric co-pilot, how it gets

41:42 build or not build, it’s confusing to me. that part I think is just a bit rough right now. And I don’t I will say this, I definitely don’t really trust a lot of the things Cop’s coming out with. I’m using a lot of trust but verify with other tools to see to build and create things. I really think Microsoft needs to focus their attention on the creator experience. It’s really focusing on building that into their agentic experiences.

42:09 So, that’s where I say I trust the the notebook. Tommy, I’m with you at the lakehouse. I I like it. Substantial. I’ll be honest, Tommy, I’ve been really trusting this SQL database inside Fabric. Pretty awesome. Works pretty well. Just operates like a normal SQL database. that’s another one I think I would add in my pocket of my Swiss Army knife of things that I like around fabric. That’s it’s it’s nice being able to have a real time transactional fast database that’s there because lakehouses are good but they’re not really good for transactional level data. They just don’t seem to fit that need very well.

42:39 Who what type of company do you think is the one that is going to have the hardest time trusting fabric? Because I think you and I are a little different here because

42:49 it’s a good question. What company would have the hardest time really stepping in and trusting fabric? That’s a great I’m I want to lean on my first reaction my initial reaction to your comment Tommy is a is a company that prides themselves on running everything through Excel has has created a lot of unintentional data silos. I think there’s a pretty big shift from working in Excel to working in notebooks, lakehouses and tables in the cloud. I I I I think there’s a pretty

43:25 sizable trust jump, [laughter] trust fall, whatever you want to call it, to get from doing that like I I can control it all. It’s all inside my notebook. I know I’m in my Excel file. I know exactly what I’m doing. That team that has built like all those macros

43:40 inside the Excel file, right? Do this, custom that, press this button, process all this data. Like they’ve they’ve spent the time and the rigor to go through that process. Not saying that you couldn’t rebuild that same process in fabric. I think you probably should, but I think there’s this resistance to release a little bit of control and put it in a program you’re not as familiar with. So that would be maybe the only thing I would say. I think if you’re a company that’s doing real-time data, fabric supports it.

44:04 If you’re a company that is doing lots of large volume, heavy volume of data, Fabric supports it. So that’s that’s maybe where I would put things. What do you think, Tommy? No, honestly, Mike, I think it’s really the whole cloud capacity environment where I think that you’re going to see small businesses, the companies of like 25 people that are probably going to have the hardest time just not saying they won’t use fabric at all. But again, I could run PowerBI on my local machine and I think going through

44:42 going through the different environment areas of fabric is where I think people are going to really lose trust with it because now their data is not on a device. And again, it’s just harder to scale up. And that’s something I’m going to carry with me is as great as fabric is for you and I who are already,, kneedeep into this, a company who’s trying to,, ramping up or trying to ramp up, there’s so much skill

45:07 it’s barely even worth the time.

45:10 Yeah, I I I want to Yeah, I think I think I’m okay with that that comment there. I would agree with that a little bit. What other parts of fabric? So Tommy, one of the things that you were talking there and something that came to mind here was looking at fabric in lie

45:24 what’s happening with not just the fabric space but like the real time space. I I really have liked a lot of like the the spark and the batch processing. So a lot of what a lot of my world is like batch processing stuff. Great. Super good. Then when you flip over to like the real time, it’s a there’s a couple areas that I have some friction around the real time space. When I look at other competitor products that do real time or event driven things, there’s nowhere in fabric where you can turn on a web hook anywhere, even with basic authentication. You

45:58 can’t do it. And so for me, I look at this going, if I’m going to build an application like a web app or something like that, it’s totally fine to have the web app running and we can secure it because fabric has secure connections into fabric. I get it. Fine. But this whole concept of like having a truly open endpoint that fabric can use to get data in userdefined functions you can you can use them but there requires authentication to get through the door to talk to them. So that’s one one of my things the real-time space is

46:29 interesting to me but I don’t think Microsoft has fully grasp this this concept of like for example I have like a web hook from PayPal. PayPal’s a system I use for doing tips plus management for people who want to buy power designer building templates and things or people are using the themes.p powerbtips site where we save your theme files right that stuff. We have that we have a payment system but I can’t send any of that payment web hook data to fabric without first putting in a f a Azure function.

46:57 So I’m going to my agent I’m telling So there’s this concept of like I have to still stand up like Azure can do it. I can make a web hook that just talks to that and then filter and clean my data directly from fab from from Azure. Why can’t I do the equivalent inside fabric? And so to me, that’s something that’s a miss there. So I feel like some of the other things around that is a bit weak., one of the areas that maybe it’s not trust Tommy, but maybe it’s just high friction for me right now. It’s high friction around the service principle. When you talk about item ownership in fabric, everything has been built around the principle of individual users owning

47:31 items. That’s okay until you have a team of people trying to figure out how to who owns the lakehouse, who created the should a person have created it or should a service principle or the workspace identity have created the item.

47:49 So from one stance I’d look at this going I don’t want to create anything me personally I want the service principle of the workspace to create and own the items. I want the service principle to own the connections. there’s two things I think maybe I don’t trust or maybe I just have extra friction. It’s connections and that service principal area. Like I want to be able to create things without people involved, but people can then edit and manipulate them. That’s a given. It’s got to happen.

48:15 Let me pause there. What do you think, Tommy, about that?

48:17 That is such a good one, too, because that’s been the the biggest thing in terms of from the connections point of view., and again that is something unless what you’re doing that’s just not going to happen. So

48:32 what is funny because fabric is it’s building us up but at the same time all those little shortcuts we had in PowerBI

48:41 are and which I listen this is why we have had a jobs for so long. I don’t care how big a company is. you will find a lot of things under the rocks even with the central BI team of just getting data in like you can’t do that with fabric right like with that random CSV file that’s my reference table here that I’m just using because I don’t know how to do the connection or whatever the case may be

49:07 that I could go past

49:12 with fabric it’s like you have to do everything right you have to have it configured right and I think that’s where a lot of people are going to be left behind where it’s like our systems are not set up that way or we’re we don’t have that maturity there. We could export the data before and get into PowerBI and it worked great.

49:33 Well now we have to really change our process. So I I want to ask you who do you think best right now is fabric serving and who do you think either individuals or companies are being left behind?

49:47 It’s a good question. I would say the data curious teams the teams that are really trying to be federated right when I say federated we talked about the adoption roadmap of Microsoft the adoption road talks about different personas right the data engineering team the the business analyst that’s like the two personas Microsoft identifies I think there’s more than what Microsoft identifies but let’s let’s use their example starting with that example I think the that blending of those two roles that that persona there. Teams that are

50:20 already doing that, teams that have leadership that understands we’ve got to move quickly. Data is a strategic asset for our company and we need to allow people to build what they need to build and give proper data stewardship. Organizations that are doing that, I think you’re going to find an immense amount of success with having teams work together to build common stuff like that. That to me is just a win. two disciplines coming together to build value through the data., so that’s what I would say the team is going to be

50:52 finding the most reward or benefit from using fabric., and they can trust it the most. Right? You’re already you’re the culture is already there. It’s established. That was the first part of your question. Your other question was what about the teams that are going to struggle? Right? Seems that that are not going to like this. I think there’s a bit of skepticism for some companies around should we be using the cloud. we have a lot of on-rem things that we’ve already like there’s this whole maybe sunk cost mentality we’ve already done all the work it’s already here it’s fine

51:24 we’re not going to move forward and there’s a lot of cultural problems around this is my data I’m not going to share it we’re not going to be easily distributing this information out to the team that’s a team I think is going to struggle a bit more in the fabric space you want to push them more towards a collaborative experience and that team that can tries tries to over control, right? Everything has to come from central IT. Okay, I get it. But is your central IT team big enough to handle all the requests, building all the reports, managing all the things that the business needs? And so I I

52:00 worry about teams that have that super central, very closed mentality without any responsibility being given to the business. They can’t handle it. They can’t do it. It’s just not not capable. Because I think what that does, I think that actually prohibits you or the company from really being competitive and what happens is you get a lot of shadow IT. You get shadow Excel documents. You get these monolithic things that are built at standalone because the business needs to do something with their data and you’re not given the tools.

52:31 What do you think, Tommy? Are those in your book or you have two different companies you’re thinking of? No,

52:36 honestly. Honestly, for me, I’m I’m I’m going to go back to the small guy that’s going to probably have the most difficulty, but I think you we’re we’re seeing right now with us where it’s going to be that selling part, right? So I think if I’m central BI and we’re looking at fabric

52:55 and even we’re seeing the conversations today there are so many places in terms of where to start but [clears throat] it’s just a few selling points in terms of the questions that we’re always going to ask do we have to reorganize the entire team do we we talked about this with managed self- service like how much has to change

53:17 where I think for an organization like we don’t h we can do this incrementally and not disrupt things. where like you’re going to see those midsize companies to say, “Hey, we have some big projects around or you like we have things that we’re trying to put into our olds SMS and our old legacy data. Let’s just try it in fabric and start incrementally doing things that they’re already doing, right? It’s not like especially those more middlesized companies who they have legacy u systems already probably Microsoft ones. So

53:52 there can be incremental gains of also learning like on my last company they were still doing job what were they called jobs and functions in SSIS. Yeah. And that was still a thing going on.

54:06 Yeah. I don’t want to I don’t really want to work in SSIS anymore. I’m sorry. I’m I’m done with that part of the world. I want to work in something that’s a bit more flexible and easier to maintain.

54:15 I would completely agree with that.

54:17 The deployment pattern of how to get SSIS stuff out the door was just atrocious. And I all the I’m done. If I had to work in Visual Studio, not just Visual Studio Code, Visual Studio Code, I like Visual Studio, I’m done. I don’t it’s too the program is too bloated. I don’t need any of that. Like nowadays, if I can’t if I can’t build it in visual studio code, any app or software, I like I don’t want to I don’t want to mess with it,

54:42 dude. A thousand%.

54:43 Not for me. But but again this is something that is occurring in more than still happening

54:52 but I think it’s getting I think it’s getting diminished. I think it’s shrinking. I feel like that size of the world is getting smaller and smaller and maybe it’s just the people that I engage or attract or the companies that I’m talking to. It’s more we’re in the cloud. We’ve got servers in the cloud. We’re doing migrations. We’re we’re becoming more efficient. We’re doubling down on fabric is our business management tool tool. We’re going to stay there and work on things.

55:16 All right, Tommy, we’re just about at time here. We’re right here at the end. I want to This is a good conversation. I like the articles we brought up today. I think this is good. My thesis here is you can trust fabric. You can you you can do it., jump in. There’s a price point for any organization at any scale, any size. But you got to learn it. You do have to take some time to do and think about how you can improve your business by thinking about your business differently and leveraging fabric to help with that. That being said, Tommy, where else can you find the podcast?

55:45 You can find us on Apple, Spotify, or wherever you get your podcast. Make sure to describe subscribe and leave a rating. It helps us out a ton. Do you have a question, idea, or topic that you want us to talk about in a future episode? Head over to PowerBI tipsodcast. Leave your name and your great question. And finally, join us live every Tuesday and Thursday, 7:30 a.m. Central on all PowerBI tips social media channels.

56:12 Let us know in the comments. We did the we did a test today of throwing out some music that we’ve been producing for just fun things around data. Did you like it? Did you not like it? Let us know in the comments or down in the comments on the video. Let us know if you like the song. Would you like to hear more songs? randomly, of course. It probably won’t be every episode, but just every so often we’ll throw some more songs in there just to make it fun and mix it up a bit. That being said, thank you all so much and we’ll see you next time. Measures [music] pump it up and lighting up the sky. Dance to the day in the mix

56:46 I get your explicit [music] measures. Drop the beat now. Feel the crowd. Explicit [music] measures.

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.