BI in the Age of AI – Ep.550

The Magic Quadrant is out, Databricks appears for the first time, and Microsoft leads on execution while slipping to fourth on completeness of vision. Both hosts think the placement is fair and the cautions are mild. The more interesting argument is about ontologies, and about the marketing message the whole industry seems to have settled on.

News & Announcements

• Microsoft named a Leader in the 2026 Gartner Magic Quadrant for Analytics and BI Platforms — Microsoft’s announcement post, and the basis for the whole episode.

• The Gartner report itself — Worth reading directly. New mandatory criteria this year include agentic insights (agents autonomously surfacing anomalies, drivers, and clusters) and conversational analytics, alongside the familiar semantic modeling and analytical model creation requirements.

• Chicago Fabric user group — August 20 — Downtown, RSVP required. Tommy will demo starting a project from nothing, getting notebooks and code in, all from a local UI rather than the Fabric portal.

Main Discussion

Topic: Reading the 2026 Magic Quadrant, and what BI actually needs from AI

• Microsoft: first on execution, fourth on vision. Both hosts see the vision slip as real and probably useful pressure. Microsoft is setting the trail for where BI goes, so a nudge is healthy.

• Tableau is still higher than Mike expected. He assumed it would have lost ground by now. Tableau set the original trail, then Salesforce acquired it, and the recent output has been lackluster. Its community gallery, though, remains something Microsoft has never matched.

• Databricks’ inaugural year, and it shows. Both have spent time in the AI/BI dashboard space there and consider it well behind — visionary placement, weak execution.

• The three Gartner strengths ring true. Dominant market presence, productivity application synergy, and consolidated data and analytics. Mike’s version of the middle one: if you’re a Microsoft shop with Teams and Outlook, the integration is tight in a way no competitor matches.

• Real-time intelligence is a strength they don’t personally use. Microsoft messages RTI heavily. Mike does more real-time actioning than RTI work, and finds the capability feels industry-specific.

• Ontologies are the sore point. Mike has tried the current implementation three times. It’s slow to build in, won’t easily take multiple semantic models, and he can’t get a straight answer on what you actually do with one. He built Power Designer Lineage precisely because of that gap, and finds the Apache OSI open semantic interchange a more sensible direction.

• Nobody has explained what an ontology buys you. The direction is right — one metadata layer describing every table, business rule, SQL and DAX statement across the organization, with data still in OneLake. But does it reduce tokens? Improve answers? Mike doesn’t hear Microsoft asking what practitioners want it to be.

• The cautions are mild. Ecosystem lock-in (which Mike disputes — he doesn’t think Microsoft is closing the system), capacity management overhead, and workspace proliferation. His read on the licensing one: you can’t bring your own $100 Claude subscription to Fabric, you pay for Copilot. And Copilot is pay-per-use like everything else, so efficiency is your bill.

The message that’s missing

Mike’s closing argument, and the throughline of the last several episodes: the marketing across this entire quadrant is “throw AI at your data and ask it questions.” He calls that fleecing you for tokens. What agents should do is help you discover what you’re looking for, then turn that discovery into a reusable object you come back to — the creator agent. He doesn’t see that message anywhere in the report, and thinks it misses the mark.

Looking Forward

Whichever platform you’re on, expect the build experience to change substantially inside a year — less writing code, more reviewing and regulating what the agent produces.

Episode Transcript

0:29 Hello and welcome back to the Explicit Measures podcast with Tommy and Mike. You may have noticed our shirts are not changing very much these days because we’re recording multiple episodes per day. So this is a recorded episode, just FYI about this one. so that’s where we’re at. topic for today what is this our sixth this week? Oh jeez, yeah, it’s been a lot this week. Sorry, that was my that was my fault. I had some unexpected travel appear on my calendar and has pushed us to record a lot of extra episodes. So anyways, you’re going to get a very mixed of these next couple recordings

1:00 mixed of these next couple recordings around around somewhat somewhat awake, somewhat not awake, somewhat heavy worked Tommy and Mike’s throughout this week cuz we’ve been we’ve been pushing the episodes pretty hard. Anyways, Anyways, we’re trying to deliver still two episodes a week. That’s the goal here. Anyways, that being said, let’s talk about our main topic today, Tommy. Today our main topic is the Magic Quadrant for Analytics and Business Intelligence Platforms. And our topic here really is BI how is this changing in the era or the

1:31 how is this changing in the era or the age of AI nowadays? What is this looking like? How is this changing and and how do we expect this stuff to behave now moving forward? So I think this is a really good topic, very relevant and timely for what’s going on here. I think this is extremely useful for us. so yeah, just wanted to stop there. That’s that’s where we’re starting with the conversation today. All right, but before we do just a quick update, if you have not gotten your tickets there are no tickets, but the next Chicago Fabric User Group is coming up on August 20th. It’s going to be downtown p. m. Make sure to RSVP.

2:01 downtown p. m. Make sure to RSVP. And what are we going through? We’re going to go soup to nuts through the Fabric Task Flow Studio. This is a forked version or a version of Alex Powers and Microsoft’s Fabric Task Flows that they built that allows you to agentically build your architecture in Fabric just with a simple prompt. And what we did or what I was able to do is build a UI around this with deployment phases, the ability to edit, and then continue the conversation. So, it gives a nice UI. And what we’re going to do, Mike, is go through just an initial stage if you had a project

2:32 an initial stage if you had a project you’re starting on. You want to get the notebooks, and you actually want to get code into there. And doing that all not in the user interface, doing it in a nice UI that’s running on your machine. Agreed. All right. So, check that out, Tommy. You have to sign up in order to get into the event. So, make sure you go to the the Meetup page. The Meetup page is in the description below. If you want to go check that out, make sure you go check it out there. And then with that being said, let’s get over to our main topic, Tommy. pull us in here, Tommy. What’s going on here with this,, one we have picked a great career to be in,

3:02 we have picked a great career to be in, Tommy. Tommy. Let me just say that. the Microsoft Fabric Community blog, which has been posted. That’s There’s a blog post around this one. Microsoft again has been named a leader in the 2026 Gartner Magic Quadrant for Analytics and Business Intelligence Platforms. Microsoft is standing alone in the far most up right corner of the graph that they have here. So, it’s they’re the most ability to execute. And they’re not the most visionary, but

3:33 they’re not the most visionary, but they’re very high on the visionary platform as well. So, they’re a combination of being a visionary and the ability to execute against that vision, they’re in the leader space there as well. well. One that I would throw that’s very interesting here, Tommy. When I look at the graph here on the post that comes out of Microsoft, there’s a couple of things that stick out to me here. one, I’m still surprised Tableau and Salesforce is as high as it is. Yeah. Yeah. Honestly. Honestly. I would have thought they would have lost some ground by now. They’re still in the visionary stage.

4:03 They’re still in the visionary stage. Yeah. Click. I’m not hearing anything new and cool coming from Click. So, I would think they would be much more on the niche player side at this point. One that kind player side at this point. One that surprised me now of surprised me now is Databricks. Databricks is a visionary, but very low on the ability to execute. And I’m kind on the ability to execute. And I’m surprised that Databricks has fallen of surprised that Databricks has fallen so far behind Microsoft in this area. So, this is interesting that Microsoft is getting a much bigger tick here than Databricks. I will say this, in the

4:33 Databricks. I will say this, in the reporting that I’ve played with, again, very little, I not very much, in the Databricks AI BI dashboard space, Databricks is way behind. They’re They’re not They’re They’re They’re not They could potentially be knocking on the door of Microsoft, but they’re not really pushing the move here. here. Well, we’ll we’ll go into Databricks, too, but this is actually their inaugural year on this Magic Quadrant, Mike, too. So, this is the first time that Databricks has actually been shown in the Analytics and BI Platforms, because of Genie. They’ve been on the

5:03 because of Genie. They’ve been on the other ones, but this is actually their first year on I assumed that they would always They’ve been there for a while because of the AI BI dashboard stuff. Yeah, they had things, but it wasn’t enough to even make the quadrant. So, and and Wow. Okay, so, they’re definitely playing catch-up here a little bit. Yeah, so, but, honestly, it’s not even just catch-up. It’s I think it’s also, too, Then I wouldn’t put them that high on the visionaries, either. I would drop them, maybe I’ll read the the report a bit more. So, let’s let’s go through it, then. They’re not visionary in this space at all. Like, they would be much farther

5:34 all. Like, they would be much farther back., I feel a bit They’re very immature in this space compared to what I see coming out from Microsoft at this point. point. Well, let’s read through it. And if this is your first time going, “Hey, you keep saying visionaries and leaders and quadrants and magic.” What are we talking about here? So, let’s probably maybe do a little back-end for our new listeners because we talk about this every year. what the heck are we talking about? So, Gartner a non-profit company that really looks at it industries across the board technology. they’re the one who actually came up

6:04 they’re the one who actually came up with the hype what is it called? The hype scale? and so they’ve been a really a great source for what what is the industry as of today? And they look at all different ones applications. One of their biggest ones is the business intelligence one. And I’ll read from here because I think this is probably giving it its proper due. Yeah. Yeah. Magic Quadrant assesses analytical and BI platform as the market moves towards agentic AI, governance semantics, and AI augmented decision support. Vendors are

6:34 augmented decision support. Vendors are differentiated by execution, the ecosystem alignment, and their ability to scale scale secure self-service analytics. And it goes through but one

6:43 analytics. And it goes through but one thing I wanted to talk about Mike is this does this does adapt every, year. , year. And they go through here on the benefits of leveraging AI but one of the ones that I want to talk about is their mandatory features that any BI platform must have must have Mhm. Mhm. in order to make the cut. Sure. Sure. So, analytics governance, the capabilities to ensure and can be compliant across analytical platforms, controlling access, data preparation. They even say supports drag and drop user-driven combinations of data from

7:14 user-driven combinations of data from different sources. Mhm. Mhm. Creation of analytical models, sets, groups, and hierarchies. Conversational analytics, I thought it was interesting it was on here. Semantic modeling, obviously, but then the ones that are probably new to this year is the agentic insights and conversational analytics that are mandatory to Gartner to make the cut here. Agentic insights that leverage agents to autonomously or semi-autonomously surface insights such as anomalies, drivers, clusters, and cast, conversational analytics to

7:44 cast, conversational analytics to interact with the data using natural language. language. Optional features this year are embedded analytics, insight delivery, and an analytics catalog. So, this is the how they are gauging or really judging and evaluating each of the different players in the industry. What they do as well is they look at the quadrant from the ability to execute, low to high, and then on the x-axis, you can imagine is their completeness of vision. So, Microsoft is the number one

8:14 vision. So, Microsoft is the number one with their ability to execute, which is considered a leader, and they’re about fourth here when it comes to being a completeness of vision. So, usually Microsoft has been one one. They have had the best and they like in terms of the best product and being the biggest visionary, but they fell behind with the visionary side, Mike. Yeah, I think so I think so a little bit. I think they’ve dropped the ball here a little bit. And I think it’s it’s good that they’re doing this. I think it’s good that Microsoft’s getting a push here, but also I think if

8:45 getting a push here, but also I think if you don’t, they’ve been a leader for quite a long time here in this magic quadrant, which is great., I’m very happy that that is occurring. year now. year now. Yeah, and so the fact that there’s so many years of regular investment on this, I think they’re they’re very well positioned to continue to stay here. However, this also means if you’re number one in a lot of these areas, areas, everyone else is everyone else is gunning for your position. Yeah. Yeah. You got to stay ahead of it. You got to work,, put your assets and your

9:15 infrastructure and your teams and allocate resources where it makes the most sense. Where you’re going to make the biggest bang for your buck. and so I think a lot of the direction of where BI is is expected to go, it’s got to be driven by Microsoft. Microsoft’s setting the trail here. I I honestly I think Tableau was setting the trail initially. Yeah. Yeah. They got bought out by Salesforce, and I’ve just seen a lot of lackluster stuff coming out of Tableau recently. So, it’s just a part of the Salesforce ecosystem. So, anyways, we’ll

9:45 Salesforce ecosystem. So, anyways, we’ll see we’ll see where this goes. What are your thoughts? Anything you picked up in the articles Tommy here that you felt were relevant? There’s a few I want to talk about when it came to AWS and especially Databricks just being a new player here, but I thought probably the best place to really jump in is looking at Microsoft, right? There we go. I don’t want to look at Domo. I really don’t want to look at Qlik and Zoho, sorry. So, Let me in my point. If anything, the only reason I look at them is how do I migrate them into Power BI? That’s exactly right. That’s the only time I touch the other systems is like, “Okay, we have a new

10:16 systems is like, “Okay, we have a new project, great. You’re using Domo, great. We’ll migrate you from Domo to Power BI. Hey, you’re in Qlik, great. What’s my project? Oh, yeah, we’re migrating from Qlik to Power BI. How do we get the equivalent experience?” The only time I hear the mic is how are we moving out? Yeah, exactly. right. right. No one moves to Cleveland, I guess. But, no. no. so, but I I thought again, just I wanted to offer what Gartner said here about Microsoft. And what they do is they give an explanation. They talk about the strengths and cautions or really the weaknesses. So,

10:46 cautions or really the weaknesses. So, let me give them the summary, and then I want to pick with you what you found interesting on their strengths and then we can do the cons here. So, Microsoft is a leader in the Magic Quadrant. It’s primary analytics offering, which is Power BI, which operates as a critical component of the Microsoft Fabric platform. Thought this was interesting right off the bat, Mike, but we’ll keep going. Operations are highly global and typical customer span all organizational sizes across nearly every industry, particularly those already invested in

11:16 particularly those already invested in the Azure ecosystem. Recent and future investments focus on the Microsoft IQ initiative to provide contextual data for AI agents on transitioning from standard semantic models to ontologies, enabling automated workflows and read white cap read write capabilities directly within the analytics environment. So, Mike, we’re hearing some buzzwords here, definitely, but Tons of buzzwords. Yeah, tons of buzzwords, especially ontology, but we just talked about a couple of weeks ago how they’re shifting

11:47 couple of weeks ago how they’re shifting Copilot, right? Where Copilot, they’re really having a different direction for Copilot in the Power BI ecosystem, which is the first step, but the one I want to actually touch on you first or get your take is this is talking really Power BI, not Microsoft Fabric. So, I am imagining that Fabric need has not been part of a quadrant yet for maybe data engineering, the other magic quadrant that that Microsoft has. And because that was really the key thing I got out of this is, “Hey,

12:18 key thing I got out of this is, “Hey, Microsoft Fabric is our tooling now, but this is looking just primarily at Power BI.” BI.” Mhm. Mhm. Yes. I I Yes, I think so. I But I But I believe the I think, Tommy, there I think there’s another another magic quadrant that Microsoft is doing, but when it comes to Fabric, though. So, I believe there’s Yeah. Yeah. Yeah, yes, yes. So, I think maybe what you’re describing here is there’s two magic quadrants that Microsoft is playing in. And

12:49 And, if you if you look at the scope of other magic quadrants, there’s like a data engineering or another quadrant that they’re a part of, I think. I think. and I think that’s another one that There’s a just a separate quadrant that they’re talking about here. Like they’re leading in a lot of little areas. Like in, in the global industrial IoT platforms, Microsoft is a leader, right? In data engineering, I think they’re one of the leaders there as well. I’d have to double-check that a little bit to spot-check my my knowledge there. there. but they are playing in the space

13:20 but they are playing in the space here where they’re doing a really good job. job. And it’s interesting here because you and I work and live and breathe in

13:28 you and I work and live and breathe in Microsoft Fabric and I think it’ll be hard for us now to basically differentiate the two, right? Where if you’re doing Power BI, at least you and I are making the assumption you’re doing Fabric and I think the other way it goes the other way. So, that’s an interesting thing that they make this separation here and then obviously obviously you can see how much Gartner’s also pushing themselves, this buzz of the agentic AI experience, the agentic Yes. They they firmly believe that this is coming through and this is what’s

13:58 is coming through and this is what’s happening in the future. sorry, I was doing a little bit of Googling here on the side Tommy to kind Googling here on the side Tommy to come back to some of this, right? So, of come back to some of this, right? So, Microsoft’s listed in many different Magic Quadrants that Microsoft has. Data Integration Tools is the one I think I was thinking of, right? So, Data Integration Tools, there’s Informatica, which is like the leader and I would argue really expensive to use pure Informatica, but Microsoft is a good secondary option there, which also plays with Oracle, IBM and Amazon Web Services. So, they’re all like this is like a a race across those

14:28 this is like a a race across those industry leaders. This is like the 2004 article that I pulled up here, but there’s a lot 2004 2004 to look at. Yeah, the ones I care about, right, are the Business Intelligence and I care about the Data Integration Tools. Like those are the two sets of technology stacks that I’ve basically built in and now there’s an AI space, data science, right? And so, Microsoft’s also trying to lead in that way with like Foundry and other AI Integration Tools into their BI solutions. I’m curious, Mike, do you feel

15:00 I’m curious, Mike, do you feel I I think there’s going to be multiple AI Magic Quadrants. I don’t think it’s going to just fit in a single bucket. but why don’t we actually dive into the strengths here when it comes to what they talked about? So, Gartner gives three strengths to Microsoft or three items that they said, “This is why Microsoft is so,, important, such a leader in this space.” Yep. Yep. They talked about the dominant market presence. How are you guys widely deployed? Productivity application synergy. What a What a phrase here. What a buzzword.

15:30 What a buzzword. Three buzzwords. If we’re playing buzzword bingo, this one’s like on the winner list. Right? You just did a tic-tac-toe all in one sweep. Yeah, cuz Yes. Yes. Synergy is worth two, by the way. So, the and what they’re basically talking about here, what the heck does that mean, is simply integrated with the apps like Teams and Excel, which helps organizations standardize Microsoft infrastructure. And then consolidate, which I found interesting that’s a major strength that they’re putting. Well,, I I do think, if if you look at compare this to other tools,

16:00 you look at compare this to other tools, right? Compare this to Tableau, to compare Compare this to like Amazon Like so so my business works in If you’re a Microsoft shop and you work in Microsoft products, right? You have Teams, you have Outlook, like that’s the You’re going to choose like a platform to stick with, right? The integration between Fabric reporting, BI reporting, and then Teams directly, pretty tight. Like I don’t see Again, from an equivalent standpoint, I don’t see the same level of tight integration

16:30 see the same level of tight integration between Google and Looker, which Looker is their version of analytics and Google Workforce or Google Chat is their version of chat. There’s like almost no integration between the two systems, right? You can’t You can’t go Maybe you can, but I This doesn’t feel like it’s a very smooth experience to go from chat in Google into a Looker report that’s there. You can share links, yeah, but it’s not like It’s not like I’m going to like what I do in in Teams, which is there’s a dedicated Power BI extension I

17:02 there’s a dedicated Power BI extension I can click on and I’m already signed in and in my report. It just works. It just works. So, the last strength, and I really want to get your take here, because we just said the things about Power BI. Consolidated data and analytics. Microsoft Fabric unifies data engineering, warehousing, and BI’s in one SA in one SA SaaS platform, reducing integration across fragmented stacks. The capacity also provides access to the broader environment, including real-time intelligence in Fabric IQ. So, we were just saying, “Oh, no, this is just Power

17:32 just saying, “Oh, no, this is just Power BI.” But now they’re changing it. Especially when you look at the caution here, here, where one of the things that they mentioned as one of their, you mentioned as one of their,, negatives of Power BI is know, negatives of Power BI is dependence on broader Fabric environment. Power BI’s full value is increasingly tied to the Fabric capacity and the broader platform adoption. So, Mike, if I’m Microsoft reading reading those strengths, would I be pleased, satisfied, upset, or fine with just reading what the strengths are? Would I agree with that if I was,, a

18:03 agree with that if I was,, a runner someone Yeah. That’s a good question. And I said, “Yeah.” I’ll let you think about that. And when I was reading through this, I was trying to take it from that mindset. And I think they’re hit I think these messages that I’m hearing here in these strengths are in fact the right messages that I’m seeing Microsoft talent, right? I would agree with Yeah, they are market dominance in They already have a lot of other business software integration. Microsoft has really taken a storm by Power BI has been a growth monster for

18:34 Power BI has been a growth monster for power for Microsoft. It’s continually growing. It’s been gaining users, monthly active users. They’ve got tens of millions, if not hundreds of millions, of users each month interacting directly with Power BI reports. So, from that perspective, they have come in and made a big splash in the screen, even though they were late to the game in this analytics online reporting thing. The interesting between around consolidation between data and analytics, right? So, they’re talking about RTI. And one of the messages I’m

19:04 about RTI. And one of the messages I’m hearing a lot from Microsoft is a lot of information or messaging around how good real-time intelligence is with Microsoft and in Fabric. Fabric. Right. Yeah. The interesting part here is that’s a strength of Microsoft while I as a business user or I’m working with companies, I’m not doing a lot of work in the RTI space specifically. It feels industry specific. I’m starting to do a lot more work around real-time actioning, right?

19:35 around real-time actioning, right? Something happens, a model refreshes, something changes, and then I need to action off that information or data or notifying and spinning out specific alerts from there. But,, what Microsoft demos is real-time intelligence feels a lot more of a narrow use case versus a broad experience. I will say Fabric and Power BI’s integration with Fabric IQ and now Work IQ, I think is interesting. So, one thing I didn’t see called out here was the Work

20:05 didn’t see called out here was the Work IQ integration, which I do think is part of is relevant here.

20:09 of is relevant here. Yeah, I think that goes over the bigger umbrella of Microsoft IQ, right? You’re probably right. But, Fabric IQ is trying to hook into that experience. Right. Right. Yeah, so And I’m happy they didn’t say the word ontology in this dumb thing. They did. Oh, they did. They talked about in in the strengths,. No, No, well, here’s the one that I think this is It’s not a strength. Sorry, guys. Sorry, team. Sorry. You’re not You’re not hitting the mark on ontologies at this point. You still got a miss there. I But, here’s the thing though. I don’t

20:40 I But, here’s the thing though. I don’t know if you read this in the summary, I feel like this is almost more concerning. And again, maybe they I I’m going to assume Gartner is not understanding Microsoft correctly. But, the initiative to provide contextual data for AI and and transitioning transitioning from standard semantic models to ontologies. That’s in the summary right near the bottom, enabling automated workflows and read-write capabilities. I don’t know if they’re talking about the ontology,, in terms of the Fabric product

21:11 , in terms of the Fabric product right now, but but I I’m not too concerned about Yeah. Just because there’s an item called an ontology doesn’t mean the term actually is important and does exist, right? So, I think Microsoft is clearly recognizing that the ontology of my entire business does make sense. That’s That’s legit. Whether or not the item in Fabric is actually the correct way to represent an ontology, ontology, debatable. Don’t like how it works,

21:42 debatable. Don’t like how it works, can’t use it very well, it’s too slow to get stuff built in there, it doesn’t automatically upload multiple semantic models and make it easy to work with. So, the current flavor of ontologies, the way it’s being built today, there is a graph database, you can use that. You do have the ability of making items in there, but in general, it’s difficult to work with. I don’t like it. I’ve tried it three times, everyone keeps saying check it out, I try it. It’s just not hitting the mark for me. This is why I built Power Designer Lineage because it’s the

22:12 Power Designer Lineage because it’s the same thing. It’s literally giving you like lineage of stuff and pulling items together. And to me, that’s a lot more relevant to what our business wants to see, which is I actually need to see tables, columns, and the lineage items of our ecosystem. Well, the gap for me, Mike, is really when it comes to what’s the purpose of the ontology right now, it’s agent, it’s an operational agent, right? Like, what can I use an ontology for is where the story I think is still there’s a their

22:43 story I think is still there’s a their road map is not clear. I don’t think the road map’s clear. I also don’t think they’re listening. It doesn’t feel like to me like Microsoft’s listening about like what we actually want an ontology to be and how it wants to work or function. I’m not hearing asking questions, I’m not hearing from Microsoft like what what do we want an ontology to really like look like. And again, I keep going back to okay, we say we have them. We say we can get them loaded in and hydrated with some information and data, but like what do you do with them? Right. Right. How How does that ontology actually reduce my number of tokens? How does

23:14 reduce my number of tokens? How does that ontology share my business logic more to across my team members? How does the ontology help me define a single metric that can be used in SQL or in Power BI? Right. How do I share that same language? Not there. Does not exist. exist. So, for me, I look at this going okay, well, this is all fine and dandy, but like we’re missing some core elements here that are I think more in alignment with the what now is the Apache OSI standard, open semantic

23:44 Apache OSI standard, open semantic interchange, which makes a lot more sense to me as a as an ontology layer. So, regardless, Microsoft’s moving toward the right direction, which is an ontology that describes all your tables, all your business logic, SQL statements, DAX statements, everything together. That’s what it is. It’s basically a semantic model for your entire organization. Right. Right. Except there’s no data stored in it. The data’s stored in one lake tables. The data’s stored in one lake tables. It’s all the metadata, right? And this is one of the things I think Microsoft

24:15 is one of the things I think Microsoft screwed up a little bit with the messaging here in the word semantic model. Because a semantic model isn’t really a semantic model. It is a semantic model and all the data that belongs in the semantic model. And I think that’s Yeah. Yeah. It should be the the semantic model should really be renamed to a data semantic model, right? The data’s inside the semantic model piece. So, we really do need an enterprise or organizational holistic view of a semantic model.

24:45 holistic view of a semantic model. Right., again, words have meaning. This the definition of semantic is,, is,, relating to the meaning of language and logic. Concerned with what words and phrases and symbols mean as opposed to how they’re structured. Meanings. Yeah, the the data. Exactly. Yes. Semantic doesn’t mean the raw data that comes from the Delta table stored into your semantic model., that’s that’s that’s Right. Right. And maybe the model part of it is supposed to be the carrying some of that, but I maybe maybe it’s semantic data model is what it should be, right? That cuz

25:15 it should be, right? That cuz there really is like physically stored data inside the semantic elements. Yeah. Right now, ontologies feel like a iPad mini with no touchscreen support. Like I Okay, why when I use When would I use this? Well, what’s the purpose here? We’re like, okay, I guess I can use it with a keyboard and mouse, but why? And I that’s how I feel right now with the ontology story. Yeah, I think I I think I feel similar. I think right now, I I feel like I can understand the ontology story when I have a single semantic model and I go from the

25:46 semantic model and I go from the semantic model and say import to ontology. ontology. What for what? I understand how to get it in, right? So, the friction for me is okay, once I have a semantic model there, how do I load a second semantic model? How do I take two tables from two different semantic models and identify well, they’re actually the same table? Yeah. Yeah. Semantically, right? When I say the word date, date, I’m going to have 10 different date tables across different models. All of them will be somewhat different.

26:16 All of them will be somewhat different. Different columns, different pieces. That’s okay. But what makes a person want to use a certain date table in this situation? That’s where the description comes from. It says, this date table has these kinds of properties in it for this reasoning, right? The columns in them have very specific decisions, right? I should be able to go to And this is where I think ontologies do make sense in this context is let’s pull in 10 semantic models. Let’s look at all the related date tables as a collection, as a grouped element. Okay, now that all those date

26:47 element. Okay, now that all those date tables are grouped together, what is the description of each of those tables? What does every column mean? And that’s

26:53 What does every column mean? And that’s where I should go to the AI and say, “Hey AI, I’ve got five date tables, and they’ve all been described on how they’re used from each of their individual semantic models. Tell me, can I consolidate these? What is going on here? Why do we have so many different date tables? That’s where the ontology makes sense. Yeah. Yeah. Because you’re able to generalize and bring together all of the tagging or relations between these various node objects that are all date tables. And it’s going to be interesting on what

27:23 And it’s going to be interesting on what the Gartner quadrant looks like, and I think more importantly, how Microsoft positioned this, because you can see that the strength and the summary is banking on Fabric I Q, and banking on Fabric. So,, I’m I’m not putting pressure on Microsoft, but they’re, I think there’s a there’s a lot of investment here. Both from the cost, but also from where they’re positioning themselves., , let’s go to little cautions, and then I want to get your final thoughts here. So, there are three things that

27:53 So, there are three things that Microsoft or the Gartner is saying, “Yeah, we’re not sure about yet.” What two I don’t know how much I agree with, so I’ll get your take. So, let me read them out. First, dependence on the broader Fabric environment. Power BI’s full value is increasingly tied to the Fabric capacity and broader platform adoption. adoption. Duh. Duh. Duh. Duh. Totally get that. Yes. Everything’s a CU. Bingo. Like, this is strategic. I get it. get it. Well, why is that a con, though?, that I feel is the nice part of it. It’s like, I don’t have to go to 18 places

28:24 like, I don’t have to go to 18 places for the resources, and if you want this data integration, and you could still run Power BI. It’s not like importing has gotten worse. You can land your Databricks tables inside OneLake if you want. No, but I’m not even saying that. I’m saying, if you were just using Power BI, I understand what they’re talking about, but yeah, it’s a it’s a What I think they’re trying to say here is Microsoft is starting to build a bit of a closed system, and I would disagree with that statement. I don’t think that they’re actually building a closed system in any way. I think Microsoft is trying to keep the standards and the system open as much as possible, so that

28:54 system open as much as possible, so that it’s as as flexible as you can have it for any user. So, If you wanted to use Yeah. I understand. If you wanted to use Power BI today like you did in 2020, back when we partied in 2015, you still could. Still connect to data sources and APIs in Power BI Desktop. Just run that whole thing. You could do your governance that way. You don’t need the other things here. So, I disagree with that. Now, I think I think this is a caution with Fabric. I would say this is a caution cuz if you want to bring in like things like DBT, if you want to bring in other data engineering tools like they’re not

29:25 data engineering tools like they’re not like Talend, like there’s other tools that you may be paying for that exist that may do a better job of what they’re doing in that tooling space. Right. Right. You can get somewhat creative in getting those tools to work with Fabric. It does take a little bit more thought and effort, I think, to get these other engineering tools to work inside Fabric. And I think that’s maybe what they’re calling out is, “Look, if you’re going to use pipelines and notebooks, Fabric’s got you covered, but you’re going to be in the Fabric ecosystem doing it.” Sure. Sure. So, for the companies that have a lot of

29:55 So, for the companies that have a lot of legacy stuff laying around, or maybe you’re a Databricks data engineering shop, and you want the BI reporting shop side of Power BI, Right. Right. you’re going to have a little bit more friction. It’s getting better, but it’s still not super seamless. And maybe that’s what they’re saying. I And my point I’m saying is Power BI hasn’t gotten worse., it’s not like they limited features because of Fabric. But, let’s go to the other two. Shared capacity management, Rick. Copilot usage is centralized on a designated Fabric capacity, requiring active user assignment and monitoring to

30:25 active user assignment and monitoring to manage consumption and maintain performance. Fail. Fail. So, they just they Yeah. They just talked about this is a pain point. is This is a pain point. Yeah. Yeah. This is a pain point. But, what this does not consider, this does not consider consider Mike Carlo’s Power BI Desktop Designer extension where you can bring your own agent, right? So, but, to their point here, what they’re calling Here’s how I read this one. If I buy a Claude Code $100 subscription, I can’t bring it to Fabric

30:55 subscription, I can’t bring it to Fabric easily. easily. I’m building special things that I can work well. So, there is no concept of bring your own agent. My workload does do that. So, one of the things I think is a miss here is if you look at the scope of what Fabric is doing, there is this heavy reliance on Copilot, and all that usage comes from CUs. of changes though, too, with Microsoft’s article article Tuesday, July I think it was 22nd, because this is saying the old way of you needed a Fabric license, the separate Fabric license or Copilot

31:26 separate Fabric license or Copilot license in Fabric. And what the Remember what came out? We just talked about it where where they’re doing your data entering questions. Power BI’s been discovered all the big truth is now Copilot can answer questions by querying Power BI directly. And it sounds like that you’re not going to need another license, because to this end, the Power BI standalone, they’re not talking about the licensing, will be upgraded to the rich 365 experience. So, I don’t know where it’s going to go from a licensing point of view, but it sounds like they read this and went,

31:56 sounds like they read this and went, “Okay, this needs to change, because we can easily do something about this.” Now, I don’t know, but it’s sounding that way based on the article that came. It definitely sounds like a reaction. Someone’s losing their shirt on the amount of Copilot money that’s being spent, and the licensing pattern pattern needs to change. Right. Right. Right. This is what this is. This is is someone is figuring out, “Hey, we need to change all of our billing to go towards a not a flat fee, it’s pay-per-use.” So,

32:26 not a flat fee, it’s pay-per-use.” So, Copilot and M365 Copilot co-work is following the same pattern as every other agent related thing that you’re going to find with Claude, GitHub, and everyone else. They’re all billing as much as you use it. So, if you use it efficiently, you get a great deal. If you use it inefficiently, Microsoft makes their money. And here’s the last one. I can’t wait to get your thoughts here. Workspace proliferation risk. The ease of workspace creation and content publishing Well,, if you do it wrong, frequently leads to

32:56 wrong, frequently leads to duplication of dashboards and reports, as well as fragmented and semantic models, requiring administrators to implement strict life cycle management to prevent conflicting data definitions. Well, that’s when you call Mike Murray. More importantly, this is how you set it up, too. This is part of the appeal to Power BI, Mike, too. This So, this is where I’m confused. You can’t have your cake and eat it, too. too. Right. Do you want only the guy who’s in IT to build reports? Because then you won’t get duplication that way, but you’re also going to not get your reports. reports. How are other teams managing this? It’s

33:27 How are other teams managing this? It’s the same problem across any other platform. You either give all their ans- You give a lit- You make it easy for people to create stuff, or you don’t. Like, so

33:35 Like, so I understand the the caution here, but I think the caution here really is I guess the caution to organizations are you better stink and have a central BI intel- a central BI team, right? To me, me, that that that line item reads like you better go read the Fabric adoption roadmap, and you better have a COE, a BI COE. COE. If you don’t have that, this stuff will be a risk to your company, and you’re going to let it run wild, and you’re going to have a lot of extra costs

34:06 going to have a lot of extra costs because you’re not really managing it. Mike, if I’m Microsoft reading the cons there, I’m going, “Sweet.” Because those aren’t bad at all. I To me, I thought it could have been a lot worse. Like, They were pulling hard to find some kind They were pulling hard to find some caution to actually throw to the wind of caution to actually throw to the wind here. here. 100%. 100%. So, I I completely agree with that, and I think looking at the whole picture here, because I know we’re getting your time. You got Salesforce, to Google, Tableau. If you look at actually Tableau in this, in this, where you’re actually looking at, if I

34:37 where you’re actually looking at, if I can spell Tableau correctly, Tableau had some interesting cons, and I want to get your They actually have a really good pro as well. as well. Tableau has probably one of the most vibrant communities around building data things, and they’ve got a really good Tableau online experience, which is second to none. So, one thing I will call out is Tableau’s community gallery of stuff Yeah. Yeah. is is awesome. And Microsoft has no response to a build a free Power BI report in a community area, and everything’s public,

35:08 community area, and everything’s public, and it has to be shared, right? They don’t have it. Right. Right. There’s nothing there. They have a galleries inside the inside the Power BI blog. They’re okay at best. They’re not anywhere near what Tableau’s doing for a community engagement. Right. And cost and licensing is always really the non-starter. Their stack competency on the agentic things, and it doesn’t really work for operational analytics. And though to me those are really like those are deal breakers. If your cost and licensing is already out of control, and you can’t do operational analytics, I think it’s a huge thing. You can look at Google in the same way,

35:39 You can look at Google in the same way, where they said it’s a leader, but one, it’s very developer-centric. This is This is your You can’t eat your cake and have it, too, or have your cake and eat it, too. Yep. Yep. You want it to be pure developer, where then you have bottlenecked the distribution of content. You want it to be more widely available. Well, without the right,, statutes and data governance in place, you’re going to lead to dupli- duplication. I don’t think that’s a Power BI problem, to a to a point. Agreed. Yep. 100% agree with you there. All right, Mike, what’s your final takes looking at this, and where

36:09 final takes looking at this, and where we are with BI in the age of AI in 2026? Well, I think I think there’s a lot of good things coming. I I really do think that the creator agent needs to be front and center. center. Mhm. Mhm. A lot of A lot of messaging I’m seeing coming out from Microsoft is, “Hey, you don’t need to build a bunch of complicated things. Just go talk to these agents and get answers from them.” That is just fleecing you for tokens. It’s just taking It’s just taking your

36:39 It’s just taking It’s just taking your tokens for money. That is not a good way. AI should be used in unknown areas, places where we need to discover things, situations where the the answer is not yet known. But as soon as we need to figure out what the answer should be, hardcode it in. Make a deterministic solution. Build a script. Build a tool. Build a something. Build a report. Agents should help you discover stuff. And then from the discovery, they should immediately be getting you into building reports and creating that discovery item

37:09 reports and creating that discovery item into a reusable object you can come back to over and over again. That’s where agents are going to excel. So, I don’t really hear a lot of language even in much of this magic quadrant. I don’t see a lot of that messaging happening. It really feels like the the main marketing message is, “Throw AI at your data. Ask it questions. It gives you answers.” And that’s not the right story. It’s It’s missing the mark in many ways. It needs to be have AI help you discover what you’re looking for. Have AI help you build the

37:41 looking for. Have AI help you build the complicated DAX that you don’t know how to build. Have AI help you build a good-looking report. what you want to build. You can build the visuals. visuals. How do you stitch all that together? Make it a good-looking report. Let AI help you with that. That’s where AI is good. It’s taking this semi- known item with discovery and then from there, it’s helping you build out the actual output of what you care about. And let’s not forget at the end of the day, if people don’t trust their data and have an easy way to access it, it doesn’t matter. You can have all the

38:11 doesn’t matter. You can have all the chatbots you want and these nice UIs, but this is what Power BI to me has always been, again, biased maybe, but it’s been what it is for the last 10 years. When Power BI first came out, the ease of use for development, but more importantly, just the seamless if you’re a Microsoft shop. shop. Yeah. Yeah. This is why 498 500 companies Fortune 500 companies use Power BI. This is why we still do what we do because it is an incredible tool, especially if you want to govern. If you want to get a

38:42 you want to govern. If you want to get a source of truth, that’s where you need to go. So, Mike, great conversation. I’m interested to see where 2027 looks like and how much the emphasis is going to be on AI. on AI. I would totally agree with this, Tommy. I think this is very much going to be an interesting place where we’re going to go go and build. I Regardless, I think we’ve made the right choice in our careers. I will say this, though, whatever you do, all of this will be directly influenced by AI. It will be It will be changed with it. So, if you’re building things today in these systems, either Databricks, Tableau, it doesn’t matter

39:13 Databricks, Tableau, it doesn’t matter what it is, honestly, all of them are going to change how you build in the next 6 months, 1 year, 2 years in the future, greatly shifting how much time we spend writing code and building things. It’ll be a lot more of discussing and reviewing things with AI-related stuff to help you build regulated items that you can reuse. So, I think that’s going to be the future. Awesome, Mike. Thanks, Tommy. Appreciate it. Where else can they find the podcast? Well, you can find us on Apple, Spotify, wherever you get podcasts. Make sure to subscribe and leave a rating. It helps us out a ton. Do you have a question,

39:44 us out a ton. Do you have a question, idea, or topic that you want us to talk about on a future episode? Head over Head over to powerbi. tips and leave your name and a great question. And finally, join us live every Tuesday and Thursday a. m. Central on all of Power BI Tips social media channels. Thank you all so much, and we’ll see you next time. Hey! Explicit Measures, pump it up, yeah. yeah. Tommy and Mike lighting up the sky. Dance to the data laughs in the mix. Fabricam AI, get your fix. Explicit Measures, Measures, drop the beat now.

40:15 drop the beat now. Power BI Geeks, feel the crowd. Explicit measures. yeah.

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.