Microsoft Ignite & November Recap – Ep. 482

Ignite season always lands like a firehose, and November didn’t slow down. In Episode 482, Mike and Tommy do a full recap across Microsoft Fabric, OneLake, and Power BI, then zoom out to talk about how they’re actually using AI day-to-day—meeting facilitation, project thinking, and developer workflows.

News & Announcements

• Advancing Data Integration: Innovations in Data Factory in Microsoft Fabric at Ignite 2025 — Microsoft continues positioning OneLake + Data Factory as the backbone for “less copy, less ETL” integration strategies. The theme is enterprise-grade integration at scale, with a focus on reducing redundant movement while making ingestion and orchestration more approachable.

• Fabric IQ: The Semantic Foundation for Enterprise AI — Fabric IQ reframes the enterprise AI playbook: meaning matters as much as data volume. Mike and Tommy connect this to the ontology conversation—AI agents can’t reliably act without consistent business semantics.

• Power BI November 2025 Feature Summary — November brought a set of platform-wide changes, including Copilot/AI enhancements and key product lifecycle updates. It’s a reminder that the “BI layer” is still evolving rapidly alongside Fabric.

• Fabric November 2025 Feature Summary — The Fabric update highlights continued maturation across core workloads. The release notes emphasize enterprise readiness and broader platform coverage rather than one flashy feature.

• What’s New in OneLake and the Fabric Platform — OneLake platform improvements keep trending toward “more sources + more security + more capacity control.” The big story is operational: making the platform easier to run at scale.

Main Discussion: Ignite Recap + How They’re Using AI

AI for Productive Use (Not Just Demos)

Mike and Tommy share how their AI usage is shifting from “try the new model” to repeatable productivity:

• Meeting facilitation — capturing action items, summarizing outcomes, and keeping discussions moving
• Project thinking — using AI as a conversational partner to stress-test architecture and decisions
• Developer workflows — experimenting with tools like Claude Code to accelerate real work (not just snippets)

Platform Velocity: Fabric + Power BI + OneLake

They tie the Ignite announcements to a broader trend: the semantic layer is becoming the organizing principle. Fabric IQ, model editing on the web, and notebook integrations all point to the same direction—semantic models sit in the middle while every other experience (pipelines, notebooks, reports, AI agents) routes through them.

Looking Forward

The recurring theme is “meaning at scale.” Ignite announcements aren’t just feature drops—they’re infrastructure for agentic workflows, governance, and cross-workload development. Mike and Tommy expect 2026 to be the year organizations discover whether their definitions, dimensional data, and governance habits are strong enough to support the AI wave.

Episode Transcript

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

0:13 You. That’s Good morning and welcome back to the Explicit Measures podcast with Tommy and Mike. Well, we made it back. There’s a little technical issues after these breaks, these long breaks. It takes us a bit to get all the computers resigned in and logged in. So, sorry we’re starting a little bit late, but that’s how it

0:46 Goes. Mike, I’m a little concerned because we haven’t done a live podcast in two and a half weeks. I don’t know if we know what we’re doing. We never knew what we were doing. We When did you tell me when we started to know what we were doing? Like that would that would be more helpful, I think. I haven’t figured it out yet. No. So, I think you’re right on that one. I don’t think we actually ever knew what we were doing to begin with. So, see your face. It’s good to see your face. There’s something about community getting together with people and yeah, it happens to be about data. So, it’s good to have be back.

1:17 I would agree with that as well. , so for those of you who are just rejoining again, we’re coming back off of Thanksgiving break. We’re getting back into the routine of doing some more episodes. , the holidays for us in the US, Thanksgiving is in November. We have Christmas in December. Usually, we have to do a lot more recording and juggling around different shows and episodes. So, bear with us while we do a mix of pre-recorded and actual episodes here for the end of the year. That being said, let’s talk about our main topic today and we’ll get into our news for this episode. The main topic today is going to be Microsoft Ignite and

1:50 Recapping the things that occurred inside November. There’s a lot of things that have been announced. There’s a lot of There’s actually two new really new features that are really interesting that I want to unpack with you, Tommy, like what does it mean? How are we going to use it? So, , ontology being one of them and Fabric IQ possibly being the second. So, interesting what these two new experiences or features are going to turn into. Be curious to see what how we unpack them and how they how useful they’ll be in our data. Anyways, that being said, Tommy, over to you for some news items. So, I have three beats from the streets

2:24 Here that I think you’re going to love and I think other listeners are too. And it’s really utilizing AI for productive use. And I have three different tools that all you probably use today. And I’m going to challenge people that some of these if you’re not doing it, I think you’re doing AI wrong. We’re going to talk about facilitator chat GPT and conversations and projects and claude code. So, let’s dive in. Yeah. So, I’m going to quickly jump in here. I think this has been the piece that’s been slightly missing here for AI for a while. It’s been useful in very specific use cases. I feel like AI has

2:57 Been very like equipped to do like code code generation and over the last I don’t know 3 to 6 months. It seems like that part of AI has been doing a really good job at building those things. Yeah. And Tommy, I’m going to add one more thing here to your quick news item. I’m going to I’m going to go Google Gemini at the end here. , I’ve been playing with that. I think the Google Gemini 3 is what I’ve been playing with. And, , I’ll have I have a story about that of what I did over break as well. So, okay, you’re going to get a lot of news stories today because Tommy and I’ve had

3:28 Some time off to like not do do work and we’ve been able to explore things. So, this is probably why we have a lot of like thoughts on the AI space. But yeah, in all in essence here, I would just say AI is very fascinating right now. It’s getting very interesting what it can do. And I think we’re starting to see now that AI has been out for a bit, I think we’re starting to really start seeing real tools appearing that are enabling you to do things a different way. And I heard someone say this on like a YouTube short or something like that. It’s we’re

3:59 Not prepared yet for the economy of the AI. Like the AI is a is a brand new market economy. It’s going to change how we do things. We’re in the middle of a of a really big shift from the way we used to work to the way we will be working moving forward. And I think that’s I starting I’m starting to feel that I feel like that’s occurring. We’re physically shifting what we do and we’re we’re going to be looking at a new economy of things that have AI in them. And it’s funny too because I’m sure

4:32 There’s some people listening who are a little dissenting like I thought this was a PowerBI podcast. Why do they talk about AI so much? Well, if you happen to be at Ignite or read anything from Ignite, that’s true. Unfortunately, we are in an integration with data runs or AI runs on data. And we’re realizing if anything, we are the most prepared for this avalanche of what AI is going to do at organizations. Yes. Because it runs on data. So, and with the semantic models later on top of that. That being said, we’re also very nerdy. So, there you go. All right.

5:04 All right. Awesome. So, let’s dive in. Yeah. So, you ever play a video game and there are like Easter eggs in it and you’re like, “Oh, I didn’t know about that.” Well, facilitator also has some Easter eggs, too, that can be incredibly helpful. Interesting. If you Facilitator, for those who are not aware, is a Microsoft Teams feature that simply takes your notes, will do the summary, we’ll do action items, really be that assistant during that call. So, you don’t have to worry about taking notes. Now facilitator always takes the same structure in terms of the output. There’s a summary.

5:36 There’s what was talked about. There’s always the action items. Correct? You can prompt facilitator without actually saying facilitator with certain keywords during the conversation. I found out which completely changes the output and the per the the output the actual value of the output that you’re getting from facilitator. So during every call where I have facilitator on I’ll always end with okay so the action items if I’m hearing and I make sure I say the word yeah action items

6:08 Right and I’ll say what we cover you the way facilitator structure is and you’re going to get that exactly outputed in facilitator not miss anything I would agree say hey let’s add to the notes here if I even say those words too let’s add to the notes or let’s make sure we this is really important for the call or for the next steps that we do x y and yada yada yada. Yep. Going to be highlighted in the beginning. So these little I call them Easter eggs because they’re not really well known and it’s not like facilitator say you unlock something you

6:40 Know hey but it just feeds into the output. So every call I have I’m making sure that I’m speaking to facilitator without not actually referencing it. Now, cool thing, you can also reference facilitators like, “Hey, facilitator, make sure you get this.” , , this is going to be really important. We need to follow up on this. Yeah, that also works too. But just saying certain key words during the conversation is going to let facilitator know this is essential. We’re going to add it here. So, there’s Tommy, I I like this. I

7:14 Think this is very interesting. And facilitator, I would agree when you played for teams premium. So there’s like a teams you can buy regular teams and this is like a team’s premium feature that has the AI bolted onto it. They did a lot of like AI, , summary meeting like the team’s premium was like after the meeting occurred, it had like a summary and had like notes and and all the the transcription stuff completed and done. That was interesting. But for me, facilitators really has moved the needle for like if I’m going to be paying for extra money for teams,

7:45 I’m definitely going to want this thing called facilitator. And what I’m waiting for here is I need some tighter integration between what facilitator is doing and like other AIs or things that are occurring inside this. So one of the things I’m also using I use loop and so when facilitator takes notes it takes notes in loop which is a good thing for me because I’m already in that bandwagon of taking the loop notes which is great. It also does the action items at the end. I found I really like. And then also if you go to the action items at the end, you can even it puts in whose name it thinks it should be like

8:17 In the description, but you can go in and actually add the name of the user who’s supposed to go do the action. And because I work with a team, I like assigning people to those things. Also, a good meeting runner, manager, whatever, should also say, here’s the actions and here and you, the person doing them, tell me when you think you can get them done by and noting when they should be completed by. And what I found is when I do this at the end of meetings, the notes and the action items that come out of loop actually go on other people’s to-do lists. And if you use the to-do app, it automatically goes

8:51 Over there. And I heard someone else talk about this too, Tommy. I don’t know if you do this as well with facilitator. At the end of the meeting after everyone leaves, you can talk to facilitator at the end of the meeting and give it extra information. Well, at the end of meeting that makes sense. Yeah. So, you can actually So, I have done a couple times where I’ve gone through the the meeting. Thanks everyone. Bye. And then, , let everyone lead the meeting and then I’ve gone on there and talked a little bit more to facilitator. I’ve read through the notes. , I’ve also noticed when facilitator is making notes, it’s not just making

9:22 Notes linearly. It’s actually making notes in topics. And as we’re talking about things, we may revisit something earlier. Mhm. facilitator goes back to the notes area where we were talking about that and adds additional notes. So it’s not just this long list of it’s not like a long list of notes. It actually took it takes like topics. So when it does the summary or when it’s taking notes, it’s giving you bundled topics of things and showing you what happened in those. And I think those are really good especially if you’re one not in the meeting. But then going back to this at the end of the meeting

9:54 Experience at the end of the meeting you can read through the notes and then say facilitator add some additional notes around this topic in this section and it will go back and add them in the middle of that section. That way you have all your mental memory captured right inside the meeting which is cool which is a good exercise. My former mentor actually taught me that say hey if we’re on a call and we’re a big part of that stay on the call until everyone leaves. We’ll just fake everything out cash it out a little bit more at the end. Yeah. you can vent like I don’t think Jim’s going to do that. Facilitator, make sure you put that down. I’m concerned.

10:27 Well, just throw it out of the notes. There’s a there’s another little So, that’s what it does right now. And those things are I think are impressive. I agree, Tommy. This is a great hack like just using it more efficiently, making sure you have it on. I think this is a very good useful experience. One of the things I would like it to also do in addition to this is I needed to I need to be able to action other agents or other things to do something on top of this. Right. I I one thing I’d like it to do is I’d like to be able to say, look, there’s an agent that can read this loop document and I could trigger from facilitator. Okay, wrap up

11:02 These notes, put these notes or , take action item one, take action item two, take note this, take note this, and write an email to everyone in the and summarize it and put it in an email for everyone and put it in my drafts. So, I’d like to I’d like to prompt at the end of the meeting, Tommy, going back to that, I’d like to have some additional like prompting that I’m speaking to the AI in and having it do like the next step of things. Hey, for example, I’m, , I’m in take this information and add it to a dynamics new contact for this person,

11:34 This person, this person. Like, I should be able to trigger multiple actions that do other things that go off and start automating little pieces of actions around. So that way my workflow is take the meeting talk to the AI at the end and then from that a series of actions are able to be taken or handled from the next steps really just better integration with co-pilot studio. Yeah, just the idea is like hey facilitator make sure you get our PO agent make sure any outstanding PO agents in the northeast or POS in the northeast are sent to me so I know what’s there and

12:06 That should trigger that off. So we’re working on that. All right, I know we have so much to cover. This next one is probably going to be my hottest take I’ve said in a long time. Okay, but if you’re not using chat GPT’s conversation and project feature together, I think you’re doing AI wrong and let me allow I don’t use this one. So, this is one you have to inform me on a little bit of I’m using more cloud code and I’m doing a lot more code like for and I’ll Yeah. So, code’s different. So I I’ll give you I’ll try to be as general as possible here on the example, but I’m

12:38 Working with a company that’s helping my own company out in terms from leads and there’s basically a lot of training that has to be done. I have to provide the script, but there’s videos on videos on videos I have to watch. It’s about like this methodology from a better speaking and a lot of different concepts. I’m not a video person and there’s a lot that I have to output here like a script, , this big shift idea. It’s really cool. So what I did is I took all the transcripts and added it to a chat GPT project. So now it has all these VTT files and PDFs like the cheat

13:13 Sheet, the book, even the book. The book like the giant book. Yeah. And what I started is I just stood up, had my microphone up and began to thought have a think, , a thinking exercise with chatt. No chatting at no typing at all, Mike. And basically what I did was hey chat GPT we’re working on we got to build out this ERT script for lack of a better term and can you tell me what needs to be part of that ERT script. Okay it went through it. It’s like all right what I want to do with you I want to do a thought exercise with you. So

13:46 Take a look at all the components that are necessary to complete this and then ask me some questions. So I’ll give you my responses on my past experience. So and make sure you keep it in my tone. I Mike I did this for 45 minutes and just it was a the most fluid conversation I’ve really ever had with a chatbot but it was also in the context it had all the domain knowledge of this tool this was that you’re not going to find in Google because it was only for this company. Yeah. And we went through it’s like hey no I don’t like that. what about this you

14:19 Know napkin idea? did they do we have files for the napkin idea? They said no I don’t see that. Why don’t you go ahead and add that? got that file out of that end. It’s like, okay, tell me more about this. What what’s a part of that? So then it would ask me a question like, hey, give me an example of when you had this type of situation in your data, it already has the memory of me because the project custom instructions as a fabric consultant all this and for 45 minutes, Mike, I it spit out afterwards everything and it was a lot of phenomenal content in the format that

14:52 Was needed with each of the components. Take take in mind these were like two hours two and a half hours of videos I could have watched that I’m not a video person but more importantly I think we forget too when we’re talking with these chat bots typing is a lot different than having a conversation when you’re actually like when we do in the podcast like when I’m talking to you that conversation the value that I’m finding from this is been bar none I’m I’m not going to say that but it’s one of the top three things I’m finding

15:24 With AI in general that’s changing my workflow and I think if people aren’t doing this I think you’re missing half the half the value. This is interesting. So I’m I’m doing a little bit of while you’re saying this and trying to get things together. this whole concept of like memory and and documents and a lot of other referenced materials to like pull from. So there’s this there’s this idea of like I remember when we first came out with a lot of agents things. They’re like well agents and AI really shouldn’t be a search thing. It’s it’s good. It does a good job of searching

15:55 And finding things, but then it also does a good job of like, , I remember us talking a lot about like rag initially when language models came out. And when you talk to Microsoft, they’re like, we really don’t want agents to just be like a big search engine. You’ve already got Bing and other things out there to search really well. Like, why do we need these large language models to be searching things? What it it feels like it’s starting to turn into is it’s more of like it’s less about search. It’s now more about idea aggregation. like here’s a bunch of materials that have a whole bunch of ideas and concepts in them. You need to the as the agent scan

16:28 Through these things and produce ideas or conceptual summaries of what’s happening. And that’s what we do, right? We humans, we can’t Tommy, we can’t read an entire or watch all the videos and sit down and know every single word on every single thing. We always take that information and like aggregate it down to something we understand, something we relate to. And then we can speak to those topics or speak to those concepts, but we’re not actually speaking to the actual individual words that were made in the slides, which I find is very interesting. I like this idea, Tommy. I

17:01 Think this is really good. And I think honestly this idea of a project in chat GPT is starting to feel very very similar to like the ontology in fabric that we’re seeing. So I think we’re starting to see like I cannot wait to get into concepts that are like connecting between different areas here is how I’m starting to see this thing. Yeah. Awesome. Very cool. And then the last one I’ll quickly go because like you said major announcements at ignite but cloud code if you’re a developer or Mike even if you’re doing git in workspaces.

17:35 Yeah, odd code is a concept and it’s in each one codeex for chat GPT I think they’re the only two major ones and then cursor has their own agents but this idea of I’m going to connect to a workspace or a folder on my machine and this is you at first sounds very similar to something you do in VS Code with an an ID IDE program. However, cloud code is now integrated in cloud code desktop, but I can also use it in the browser or I can use it in my terminal and it’s basically prompting

18:08 And basically go through everything. For example, I have a semantic model. I have all these repos or semantic model folders pirates. It’s like I want to add descriptions to everything. Before you add the descriptions, look at each of the measures. Look at the tables. I want you to evaluate yada yada yada. And what happens with each of these projects is I can do something else on my computer and it will just run. It is truly to me the definition of an agent of of the automation side. Now why would you do this and not just VS code and GitHub?

18:41 Well, GitHub is also pushing for this too. There are GitHub agents and the really to me the value here is when you open up an IDE like VS code or cursor well it takes a lot of a lot of runtime on your computer like it’s it’s pretty intensive you don’t want a lot of other things running with this Mike it is running running running running running and it could it is just in the background I am finding I am using this anytime I’m doing anything in a a semantic model repo or a PowerBI repo or my own projects when it comes to AI.

19:15 This is my default now. I I start with this. Yeah. I’ve been seeing a lot of people really pushing on cloud code. we use we don’t buy cloud code directly. So we don’t use the cloud code cloud service area because I think Tommy you buy a lot of different services. You’ve been buying like cloud code you’ve got chat GPT you’ve got co-pilot like there’s multiple agents that you’re using to run things. I think you even have GitHub copilot which is a GitHub. We all do. All MVPs do. Yeah. But I’m just saying like GitHub co-pilot is like an interesting test. Yeah. Yes.

19:47 Because it’s like it lets you have an it has a co-pilot but it’s not really a co-pilot. It’s like you can pick any a large language model you want to go with it. So we have chosen to like centralize around anything VS code related. So a lot of our payment paying features is against the VS code because a lot of our dev time is spent in VS Code and writing code stuff like so we’re not really spending a ton of time on Claude. , we’ve some of our employees have found some really good use around cursor. Cursor has been doing some really again. So, it’s like this I feel like it’s a three-horse race right now between like cursor, claude

20:20 Code, and then, , co-pilot and GitHub. The other co-pilots that I’m seeing not as effective. It’s the GitHub copilot that we really like because we can pick whatever model we want. Like, we can give the latest models. They’re Gemini, there’s, , cloud code, we can use grock, like anything we want. all the models exist and any one of those models can be used inside the concept of of GitHub VS code. so I like this idea Tommy I think this is really interesting that you’re finding this around cloud code. I would agree check it out if you’re a

20:51 Developer. One thing I will also want to throw in the mix here is I think we’re very very close to entering a new era of never writing code and being able to commoditize app development like we’ve never seen before. Oh my god. So, I had just started playing with Google Gemini 3 and there’s the Google AI Studio, which you can go use, which I’m having a vision of something here that I really like. So, I was okay. I’m I’m on vacation. I’m bored. Not bored. I’m just sitting around. I’m like

21:22 I’m looking at stuff. People are taking a rest or whatever. And I’m my mind is just going like, what can I what can I invent right now? I want to build something. And so, my family, my kids are playing some video games. And I’m thinking, yeah, playing video games is one thing, but being able to come up with the idea of how to build the video game that seems much more constructive just in general like Oh, I know. game design and game theory and like it just seems a bit more of like it’s okay. I find I’ve played a lot of video games in my day, but I’d actually like to what happens if I want to start building

21:54 Them? And so I went into Google Gemini 3, logged into AI Studio, started playing with it, and I started building my own strategy game. Really? So yes, and I was incredibly impressed how well it was able to like read the start the code base. So I made a map on the map. There were trees and there were mines and then I had to have a character show up. So the character would be on the map. I could click the character and then I would move them around the map. And what I was doing is I was trying to learn

22:25 What what information how descriptive do I need to be for the agent to understand what I want it to do. Right? So there’s this gap between like agents are super capable. They can do a lot of really rich things. But there’s a there’s a knowledge gap in what I know how to prompt it with. How do I give it instructions? What works really well? how do I design these things in a way that that helps me get like an output that I want? Right? So, , so I started just vibe coding this strategy game and in about, , maybe an hour, hour and a half, my son’s

22:58 Playing Minecraft and I’m over here going, “You want to see the game I built?” He’s like, “What?” I was like, “Yeah, while you were playing Minecraft and building a world in there. I had built my own game.” And so I showed him the game and immediately it was very enriching because it was a conversation around, well, could you add this? What about enemies? Right. Well, you’re it’s just a flat map with a green background. Can we make like other, , the creative side of it, right? The creative side, like the things that are like this is really interesting. So, it it immediately starts taking me away from just saying I’m just blindly

23:30 Following a game and I’m and I’m again I’m looking at what my family’s playing. They’re playing things like Block Blast. They’re playing things like these are games that would be like pretty easy to replicate and rebuild using like AI prompting. So all this to say is I think I’m getting to a place where it was one more one more comment and then I’m going to get to my idea. Okay. The other comment is when you were playing with this writing the code there was a code and prompting section and then there was a preview section and I was able to to your point Tommy around using your computer on one hand I could switch over on my desktop. I could have

24:05 The the code prompting section on the lefth hand side had a preview and on the right I could just give it commands. Hey, , add the character, make the character selectable. If I click the character a second time, it deselects them. Here’s how to add new characters. I was doing all these things to build additional characters. So, we we had this, , , the idea was I built a character, the character had to go mine trees, go mine gold, come back, build a hut or like a home base. Once they built the home base, you could have more workers being produced so I could get more workers to get more resources. Okay. So, then how do I use those

24:36 Resources to do more things? Well then, so Starcraft, basically very early versions of Warcraft or Starcraft, very early versions. I love those games. I thought those anyways, that being said, I stepped back from this and said, “Okay, what about an app that makes all the apps you would want to make?” It was fun creating the game, but what about the Mega app? Right. I’m in Google Gemini Studio and I have a project for the game that I’m making.

25:09 What if on your phone or any device for that matter, you just prompt it for what you want and it builds inside that app everything that you need. So basically you you basically it’s like a it’s a m it’s a it’s a single app. It’s like an AI app that helps you build many apps inside it and basically it becomes self- sustaining at that point. You don’t need any other apps on your phone. You don’t need any other things that you have going on. You can literally go in and say, “I’m bored. I have this idea for a game. I’m going to orchestrate or talk to it to build out this little mini

25:40 Game.” And you just play with it for 2 hours. And you just build out this thing. And then you’re done. And then imagine a whole economy that could be produced around this, which would be I’ve made an app. It’s good enough. I can share it out to a community that anyone else could just turn it on and run it directly inside the app. Or they can go buy it from me. like, “Hey, here’s an app that I made. I vibe coded this thing.” You could take the app, get a copy of it, you pay for it, and then you could even enhance it. Like, oh, this is a neat app, Tommy. You made this really cool thing. It does like 80% of

26:12 What I want. I want to make it do an extra 20% more. You go back and say, “Hey, add this extra thing to it.” Da da da da. And you and it you enhance it to what you want for what your needs are. And so now every app can start with like a base app and get customized per those users needs. And I thought to myself, we’re not far away from this. Like this is this is months away from where we’re going to start seeing the monolithic app like AI Studio being able to produce all these mini apps in line will chatting with it.

26:43 I would even go a step further, Mike, to say I think that’s possible now because the value of the agents of the cloud code and the codecs and GitHub is it’s not just so much I’m doing it on my local computer, Mike. It’s like so if I want to make a modification like so I have I’ve talked about this before. I have a local model manager for my all my local AI models and I categorize them. And what do you use for that? Because I want to call that one out specifically. well what program do you use to manage all your local models for? Is it is it notebook? No use. You’re using and then yeah. So

27:17 And all those like I categorize them I rate them. I like to add descriptions and also like the custom instructions how they feed team fit like you can use LM Studio whatever but the thing is I have this application that runs Mike it’s a GitHub repo it’s a GitHub repo so if I want to make a change to the application all I have to do I don’t have to open my computer I can literally go on my phone go to cloud code connect to that repo and say what I I I wish I could add my own custom logo for each for each thing

27:49 Custom image to whatever the case may be, , or a description or, , categorize it with a green blue and GitHub, it will run GitHub. It’s not running on my machine. It’s not running on my computer. It’s running the GitHub repo, creating a new branch, also learning that. So, now I have these branches to test things out. We’re getting to the point, Mike, where you can then in a sense launch that if you’re using a page subscription at this point, but launch that and I can I can do this on my iPad or my phone. I don’t need my computer to have these applications. And keep in mind this

28:21 Local application that’s being built is JavaScript, Python, HTML. I don’t know any like I know Python, but the other two I’ve never built any I’ve done HTML, , when it was back in the day. Yeah. but these applications are running and it becomes a web app, right? So it’s becomes an application on my phone to your point. So yeah, we are it’s ridiculous. And also I would encourage you with that thing you’re building Mike build in a GitHub repo, a private repo and then you can in a sense do all these

28:53 Enhancements without having to really even utilize your computer. Yeah, exactly right. And so you’re using cloud computing to do this one. And James is actually coming out some really good comments here. He’s like asking things like how would you how would user support work in your app? I’m like well it’s an agent. So the agent already knows it wrote the app to begin with. Mhm. You you just have the agent self-support it like you’d have in your in your initial app say hey agent add a add a feature in the bottom righth hand corner that says if you have any questions ask open a little chat window and let the user chat to

29:26 Yourself and then figure out the questions for it like I I think the agent it so that’s interesting and then James also asked well who pays for the tokens when you sell or purchase this app well my assumption here is everyone buys something like Google AI Studio or the the general app. So, you buy the general app and then you’re buying Google’s version of the agent. So, everyone basically is paying Google, but there’s a microeconomy running where I’m building things and I’m selling many apps in them,

29:56 Right? , I’m paying for Google AI Studio, Tommy, you’re paying for Google AI Studio, and both since both of us are paying for the same base product, I just make the app, I sell it to you, Tommy. You now have the app. You can start from there. sign into Google and you’re good and then you just sign into Google and then all the app lights up. You have all the the icons that you need. it could literally and then Tommy if you have any questions if you want to change things or you have any issues about the app you’re just using your tokens going back to your subscription anyway. So all it does is back to Google now they have everyone paying for Google AI studio

30:29 Subscriptions and inside there now they have a whole library of apps that they’ve been building or sharing or selling now becomes a microeconomy inside this larger Google AI studio. I think this really works. I’m not sure if this is actually something. So I think we’re getting something. But Mike I think we need to talk about fabric. All right let’s get let’s move on. We’ve been doing a lot of I love this lot of new stuff. We have major announcements. Well, let me let me just tie one thing in here really quick. Okay, I think let’s talk about fabric and what fabric is doing and all these concepts around what AI is occurring. So, a lot of the things that are interesting here,

31:01 Tommy and I are exploring a lot of like what AI can do in our workflows, how we integrate things with it. AI needs a lot of data to begin with. , in our data models, in our semantic models, we have a lot of ontology type things is what we’re going to be talking about from one of the announcements from Ignite. But right now, we haven’t quite seen the intersection occur yet. So I I think what’s happening here is there’s so much changing in the AI space. We’re getting like a very clear diver like it’s it’s not diverging, but we’re getting closer and closer to merging together. And let’s imagine my idea about vibe

31:34 Coding the app, right? If we’re vibe coding these apps, why can’t we vibe code vibe code fabric fabric? Why can’t we vibe code an environment? Why can’t we vibe? So, right now it’s not connected today, but I see really in the near term because of what we’re talking about here in the AI space of what AI is doing separately outside of fabric. I don’t think it’s going to be very long, Tommy, before we’re going to start really vibe coding a lot more directly inside our applications. And I think I saw an article recently from Brian Julius

32:06 Talking about how he likes MCPs, but now he’s moving away from MCPs and favoring clawed code directly because it’s more efficient and using less tokens. I agree. I 100% agree with that statement. So even now like MCPs were like the new hot thing. They’re interesting, but at some point they’re going to become obsolete and the next thing is going to occur. And so we’re going to keep evolving so much. Yeah. because when you run when you have the NCPs and I’ I’ve noticed this with cloud cloud code and the and VS code it takes a lot of resources and a lot of tokens on your computer and so

32:39 You have to be wary of that. So I agree with you completely on that and we’re we’re going to get there I think with Vive coding and fabric but and I yeah and here’s I think there’s some intersections here but Mike I we have 91 features that came out for Microsoft Ignite. Yeah. 91. Yes. Yeah. So for those who are unaware while we were on our vacations eating our turkeys and all those things, Microsoft Ignite also occurred and Microsoft Ignite is one the second biggest conference for us fabric people for us PowerBI people.

33:13 The other one occurring at Fabcon no would be Fabcon would be the first and then Ignite is the second basically it’s our second biggest holiday on the lurggical fabric calendar. that’s true. , so and it comes with a lot of features. There’s a ton. There’s going to be I think two you and I dive into. But what do you think of this? Why don’t I run through some preview features and I just want to run through them. I think this is going to be a lot of our conversation for the next, , few episodes. Yeah. But there’s a few previews, a few gas. Again, 91 to go through and I think a

33:46 Lot of them are really cool. Why don’t we talk about quickly some of the just new things that came out just what is it what I think shotgun it or machine gun it then we’ll go into I think the one everyone wants to hear about. Okay. Yeah. Let’s start there Tommy. All right. So fabric and agents fabric IQ. We’re going to talk about that in this ontology thing. A ton of things with a data agent such as data agent unstructured data support so you can read over documents and PDFs which is freaking pretty going to be good because

34:18 I’ve been asking for this query fabric data directly from M365 apps the NCP server for data agents and prep for AI support. The speaking MCP we also have remote PowerBI server and preview PowerBI modeling MCCP server and preview and standalone co-pilot and mobile and the service that are getting this is this is our point earlier like this MCP thing has gotten very hot and right now people are finding immense amount of value from it. I’m also seeing a lot of LinkedIn and Twitter and comments around the MCP showing up and

34:51 Being able to do MCP things to a semantic model from your VS code instance directly inside right desktop right so this is changing the model in real time using the MCP server I’ve also seen a number of open source community projects show up there around MCPS for PowerBI PowerBI desktop so there’s a lot of things happening very quickly here where MCPs are like proliferating fabric MCP ridiculous data warehouse MCP P, , PowerBI desktop, MCP, there’s all these different servers that are showing up. And again, I’m not sure they’re always going to exist, but this is like a step

35:23 A stepping stone into something else that may happen. Completely I completely agree with that. It’s a pillar to get to somewhere else, but they are important. Now, let’s give PowerBI some love because we have a ton with PowerBI. I think how many do we have? And there are good ones here. It’s not just, , we updated you can now make your page gray for PowerBI. We have 17. , so we talked about those NCP ones. we have a few with our also card visual is now GA the hero images and they made updates to the image visual enhancements. There’s an update for our matrix. Our

35:55 Matrix can grow to fit. It can auto expand columns to fit available space. Okay, just pause right there. Let’s give a clap. Let’s give a round of applause for like auto sizing columns correctly. Single clap. Only a single clap. Okay, that’s that’s something has been needed for quite a while and has been quite a pain and very painful. you. Yeah. Okay, cool. Because what? It should have been out earlier. So that’s should been out a lot earlier, but I’m very happy that it’s here now and and about time. So very very pleased about that one. Semantic model version history is GA. The Tim Delvia’s code extension is GA

36:28 And transolitical task flows in the one lake catalog are going to be in preview. That’s another conversation for another day. And the importance it’s a very specific need. What I think Microsoft needs to do like that that one is very powerful. What Microsoft needs to solve is Microsoft needs to build another visual that is a table, right? Or a matrix that is legit rightback. Like just we have all the pieces. We have functions. We have the visuals. Like someone needs to come up with like

37:01 The open- source the version of the visual for rightback from a table where I could enter multiple cells of data. Hit submit sends it back. Yeah. If they haven’t worked on it for the last 10, 11 years, they’re not going to do it anytime soon. They’re not going to do anything. I know. Yeah, that dream’s gone. And for you heavy developers out there, V R and Python visual deprecation for embedded, which only means that Python and R are going away too at some point. To me, if they’re not doing it in embedded, that just means they’re just reducing some of the codes. That means no one’s using it. Honestly, this is again one of these features of like if no one’s using it, they’re going to get they’re going

37:33 To get rid of it at some point. for you heavy heavy heavy heavy heavy heavy data engineers out there. DBT is going to be available in data factory. There’s the fusion engine coming in 2016. there’s a ton of things with dbt if you’re not aware with that. it’s a just simply it’s very developer centric codeheavy centric way of doing data engineering. export query results from power query desktop. We have some more function transform for AI intellisense and power query online. It’s available. So, excellent there. Let’s move on to

38:06 Databases. Few claps for databases. Now, GA, which I haven’t seen a lot of major updates with databases, but I’m happy to see this. So, I think it’s it’s happening kind not subtly, but I I feel like databases are so I feel like there’s been a community when we talk about data platform for like fabric Azure data platform pieces. One of the things that have been left out has been like the SQL DBA like where do you fit where do where do you live inside in lie of the space of fabric and but SQL in my opinion has been more of that transactional system it’s like the system that runs an app it’s a

38:39 System that stands up and and does things one of the things that I’m really excited about now is I like fabric becoming a mix of reporting services but also transactional services right and so I I feel like Cosmos DB SQL databases These are transactional type systems. Even custotob for that matter is a bit more real time and and like transactional in nature a little bit. But I see like a vision here where Microsoft is going to continue to add features that are going to allow you to

39:12 Build your transactional system directly inside fabric and do all your reporting inside one single ecosystem. I think it makes sense. I’m not sure if the community or the broader PowerBI community at large has adopted this is the way we should build things. , but I think it’s coming. Cosmo DB is now G& fabric. Interesting. I just had a a round table discussion with a few CIOS yesterday. Like the amount of cost we saved getting away from Cosmos. Incredible. So that’s cool, I guess. What do you mean by that? You mean that they had it originally and then they moved it over to

39:45 They move it to? I we didn’t go into the details on the act what they went over to but they’re like we had to make the migration because of the cost just get off of the system. Yes. Just get off the system. Yeah. What I found so we use Cosmos DB for things and it’s very good on extremely fast point lookups. Right. So if you’re if you’re doing like aggregating or collecting lots of records to report on it. So where I find Cosmos DB falls apart. Cosmodb is not efficient when it comes to grabbing lots of reporting data and getting it out. That’s not the efficiency of it. Cosmos

40:19 DB is extremely efficient when you’re doing looking up a single record from a very wide database. And we we use a graph database on top of Cosmos DB. So we can do a lot of very intelligent traverses and like add relationships. It works really well when how to design a system around it. What I feel like you have to know how to design it. you got to know how to design it around. And I think that’s sometimes some of the misnomer here. If you’re building an app, it’s really nice to be able to write and read data like point lookups inside Cosmos DB. Super efficient and with minimal effort, you

40:52 Can get stuff in and out of there. But when it gets to like big scale stuff and you’re trying to aggregate thousands or millions of records, Cosmos DB isn’t the most efficient thing. That’s when you’re going to want to land things like in other systems and aggregate them and report on So I think there’s just a the design mentality that needs to be accommodated for when you’re building Cosmos DB 100%. So we’ll let’s give some love to Chris Schmidt who we just had on for real time anomaly detection preview for real time AIdriven anomaly detection. We he did mention this in our in our episodes. This is now available in

41:24 Preview. Some user data functions UDS activator integration. Love that. which is now available. and then we’ll go on to the one lake platform. There’s a lot of mirroring updates in terms of being things being DA GA and to me this makes sense. you want to focus if I’m going to say what are my major focuses on one lake right now it’s complete the mirroring and complete the shortcut feature integration because that’s where the the value is a ton of developer or our security and then when we go into developers modeling TMVS

41:58 Code we talked about projects that’s a we’ll save that conversation about PBR becoming the default format for its own episode tell a lot of chatter on LinkedIn about that so for and then AI functions available. Yeah, we were very forward thinking on that one, Tommy, cuz right about the time everyone started chatting on LinkedIn about it. I’ve been seeing a lot of conversations. I think we started a trend here because when we talked about we talked a little bit around is the PBX file dead like and then we had an episode on that. We had an episode on that and then literally within two weeks I saw a ton of people on LinkedIn just posting PBX

42:32 File is dead. PBX file is dead. I’m like whoa whoa time out. Okay, that was a bit of a a catchy title that we used in our episode, but it’s not dead. It’s just shifting. It’s just changing. So, I think a lot of people globbed on to like our language that we were using there initially, which was the PBX file is no longer, but it is transforming into something new. And so, not bad for two people don’t know what they’re doing. So, well, I’ve been trying to like every time I find a conversation like, “Hey, we talked about this on our podcast.” Like, , there’s more nuance than it we’re we’re killing the PBX file. There’s a lot more going on there.

43:04 Yeah. , so those are again that’s a shotcut but Mike we have 20 minutes and I know this is not going to be the last time we talk about it but I would like to dedicate the rest of the episode because it deserves it to the two major features really two in one I guess you could really one and the same. Yes. Is this new I thing called Fabric IQ. Yes. And the amount of thoughts I have on this. Okay. Are pretty incredible. But let’s let’s back up. Let’s unpack fabric IQ and then I think because ontology is part of the fabric IQ space. Oh, I have thoughts. Yeah. Yeah. Okay.

43:36 So, Fabric IQ, what is Fabric IQ? It’s really this transformation Microsoft’s doing from and this is a big statement I’m about to make. This is from Microsoft, not this is not a Tommy hottake. This is a Microsoft hot take. Okay. They are changing really fabric and their idea of it from a data platform to now become an intelligence platform. This is really, if you wanted to have the elevator pitch of what Fabric IQ is, it’s really changing what Fabric’s all about. It’s no longer a data platform. It’s going to be an intelligence

44:09 Platform. It’s built around this idea called the ontology layer, which is really this idea of the semantic foundation within fabric where simply the ability to unify the data, the meaning of data, and the actions into a single semantic layer. It’s helping AI agents and business users reason and automate. The ontology is the core item here because it’s going to connect the people, process, systems, actions, rules, and definitions into a unified model. So, we’re really if you the way I’m thinking about this or the way you think about is if semantic model was a

44:42 2D picture, the ontology later is making that 3D. Yeah. Because we’re Yeah. And I think that’s I just thought of that just now and I think I’m actually proud of that. , it’s adding another I’ll say it another way. It’s adding another dimension to existing data you already have, right? And I like this idea of like the ontology part is interesting to me and I got to really put my head around like what does this mean because there’s lots of parts of your business that you can describe in like general terms, right? But then you have data that attaches to

45:14 Those things. And I think also Tommy, this seems like a preparation for like an AI agent, right? So we can’t to your point Tommy earlier which is we can’t supply all the information in every prompt every single time right sometimes the agent needs to make a decision based on okay we’re talking about customers what information or data is relevant to customers and so you may have to have something say well customers buy products and so if your question to an agent is what products are most important to these customers in this area the agent has to

45:47 Go and discover okay I need to learn what this user is talking about let’s go figure out what the customers are doing. Let’s go over here. Okay, customers buy products. Okay, this is how the relation. So, a customer buys the product. Here’s the products that we offer. And then in each of those areas, we’ll have data about customers and we’ll have data about products and product sales. All of that needs to like this is the ontology space. Like this is the making relationships between tables of factual and dimensional data that you have in order to be able to like answer a question.

46:21 I have never been more excited and more prepared and yet at the same time more concerned about a feature in PowerBI until this. You sure about that? metric sets. No. No. Even more you were pretty excited about metric sets. I was excited but what I’m saying here is I’ve never been not just excited Mike. There’s I don’t think there’s been a bigger impact that I feel like personally I’m a prepared for and that’s going to make an impact but at the same time concerned about and I think it’s important here let’s let’s invoke one of our former host Seth here where and

46:55 There’s never been a more better time to say this words have meaning and I think a lot of us are asking right now well what’s ontology well let’s define this conceptually like we have to unpack it yeah and I this is where the concern concern and the excitement comes from it’s really one and the same like oil and water. Let’s let’s break down ontology here and this is where Mike this is this is a big before we go there let’s let’s finish your thoughts on IQ because I I interjected a slightly there. Yeah. Yeah. Yeah. Yeah. So fabric IQ what you were describing was Tommy is

47:26 Like this like knowledge or information or intelligence now as a part of fabric which I think this makes sense. This is a natural progression. I think it was business intelligence when we were just doing reporting, but now we’re talking intelligence performance now that we’re adding agents and other things on top of it. So I think the addition of agents changes how we would want to perceive what a business intelligence platform should be doing. And I think this is what Microsoft is trying to push for in the same way that Microsoft tried to commoditize semantic models by making it free in

47:58 Desktop by making it easy to build by giving you a UI. Right? If you make it easy to create the thing, you’ll have lots of them being created. I think this is the same mentality they’re trying to go after with agents or yeah, intelligence is let me make it easy for you to build relationships between things and now everyone can create intelligentlike solutions. So, do you remember when you started searching Google and you would search for like Christopher Columbus or someone and all of a sudden it would show a picture, a description? This was like about eight years ago or actually more.

48:31 It was probably 12 years ago and it was Google’s graph. I was working at a digital marketing agency and this was a big part of it and the concept for Google at the time in a sense a from a universal point of view if you’re searching for a person place or thing well Google needs to understand it is a person place or thing now these are all the natural world obviously historical events and really what this fabric IQ is is in a sense doing that but for the business.

49:04 Yes. In in the context of like fabric context of the business of fabric information. Now fabric IQ also and this is to me one of the bigger deals here too or underrated features is this is also connecting to your shareepoint information your one drive information. Yes. business too where I I I was speaking I was doing a webinar on AI and one of the big things we’re talking about is this disconnect between that hard data which I you consider your table semantic model and a company soft data the shareepoint the word documents that are essential and it’s still

49:37 Considered data because it is information but there’s a disconnect there and fabric’s now connecting this where I have a semantic model I have a planes or whatever products well they have images they have, , the Excel files, they have and they may live in those things live in other places. Those are in SharePoint, some things are PDF, some things are one lake. Yes. Exactly. Right. Right. So, it’s basically graphing the concepts around your hard data, the semantic model where I have product, , product X. Well, it has all these

50:09 Other additional attributes. And this is Mike for me. I’m personally excited for this because my background is philosophy, , theology and just sociology, which is a actually now actually I’m happy I did it. It took 15 20 years, but now it’s going to make a little sense. Working in data and this forefront of AI, you’re realizing that this is now becoming the essential piece. This is I think what we’ve been looking for. I don’t know if this is the exact platform or in sense the user

50:43 Interface but this is on that right track. yes and and I want to add to your comment there like around like we’re moving away from having to write lines of code and understand codew writing and we’re able to do orchestration or directing and I was watching another YouTube show talking about some other people I think it’s the startup podcast which is one I really enjoy talking and thinking through it’s a lot of like startup ideas for businesses and one of the things that they’re they’re really touting right now is like we’re in the age of

51:15 Directing like directing agents to do things, directing things to do stuff is conductors area. Yeah. Yeah. We’re we’re we’re telling it what we want to design and now it’s how in our vision, in our mind, how closely can we get to it’s the ideas area. Like we’re we’re in the ideas world right now where anyone who comes up with an idea, we’ve commoditized the ability for anyone to make that idea and and make it come true. And so, , in this fabric IQ pace space, we’re looking at a couple different features here, and I just put the link in the chat as well in case you want to read up a bit more

51:47 About it. There’s items in the fabric IQ space. One is this thing called ontology currently in preview. Fabric data agents that’s been around for a little bit. We’ve been playing with this now, but data agents in fabric is a thing. Then there’s the Microsoft graph in fabric. So now there’s a graph item or graph in Microsoft fabric also in preview. The operations agent which is now there. It lets you create a real-time AI agent to monitor data in real time and then action on that real-time data doing some something with an agent in place there. And then you

52:19 Also have now our tried andrue the semantic model. So all these things are like housing looking at touching you being able to access data and all of them are adding additional information enrichments on top of them and one of them I’d argue very heavily here is the semantic model right this has been a a staple for most of the system which is here’s our data these are our dimensions these are our facts these are the relationships between them this makes a ton of sense to have these things related what was missing is more metadata more information that describes

52:54 Like the ontology part which is can ontology tell us what semantic model to use what tables are there where they come from the ontology piece of this is going to do a bit more of the intelligence side of things that are going to be it’s basically downloading what you and I already built Tommy like we know how to build this with pipelines and data transformations and building it the ontology is giving more context around what we already know and I would say maybe democratizing it or making it accessible or or giving more ability for people who are non technical builders to

53:26 Go interact with the data and see how it relates together. And here’s the crux of it Mike. So if if you have anything more there let’s go on to the definition of okay this is where my concern comes in not so much on the product itself but let’s let’s give the definition of ontology and this is where the goosebumps happen. Ontology in the original definition is the branch of metaph metaphysics and that studies the nature of being and existence and reality. It simply asks what actually exists? What categories of things are

54:00 There? In so many words, Mike, if you were to break down the philosophical side and is what makes a chair a chair? Does it have, , four legs? Right. But that’s this is what we’re talking about here. and but what makes a chair a chair? Does it have four legs in a back? My chair doesn’t have four legs. It has six wheels, right? So what makes something what it is now? So I find it really interesting Microsoft’s using this term to be a little more or less in the clouds but again you have to bring

54:33 Philosophy into this is the formal and this is Tom Gruber 1993 a formal explicit specification of a shared concept. What is something clearly defined? What are the concepts and constraints of something? What are the shared agreed upon basically rules of that by a community or organization and an abstract way of how something works in the real world. Okay. So that’s ontology and this is what Microsoft’s trying to accomplish. Mike, the reason I am concerned about this is because as

55:08 Essential as this is to AI, and it is, I think Microsoft’s right to be putting this in the business, I would agree. What has been the biggest problem with just doing data in general with PowerBI? my the constant struggle at every organization without even bringing AI into the mix has been the shared idea of definitions your inventory how do we actually categorize things and this is that’s not even metaphysical I’d agree

55:40 With that so so I’m going to argue right here and right now if you are not an expert or have the training or the experience on building data governance this you are not going to succeed here in Fact, Mike, I don’t think we’ve ever been at a point now or we’ve never been more at a point where data governance has been more essential for an organization. H interesting comments. I I guess where I would go with this one would be maybe two two areas here, right? So

56:12 When we look at like the going back to the ontology piece of this one, , look, I put another link in the documentation here for other things around ontology core concept definitions of the ontology itself. entity types, an entity instance, the property, and the relationships. And it goes through in the documentation a bit more. It says, “What is an entity type?” An entity type is like a shipment, a product, a sensor. The instance, which I’m not quite sure I’m clear on yet, but it’s, , it’s a the occurrence of the entity type. , and it says like a semantic growth. They don’t really give a good

56:44 Example of the entity instance, something that could be true or false. I think in my mental model I’m thinking of an entity instance like a customer but a customer has an address and that address may change over time. So we have an instance of this customer this address and here’s what it is today and it this is what it was for November but in December it changed and now there’s a new instance of that entity. So it’s still the same customer but now it has like a different version to it or there’s there’s some data that’s being changed about it. They talk about the

57:16 Properties of it like customers have accounts, customers have addresses, customers have a scale or how good a customer is a rating on them or something like that. And then there’s relationships. How does customers relate to products and what things they buy? So the ontology concept I think it’s just more about talking about bringing what we understand as people, the relationships between objects and giving it into a computer that can actually handle that differently. So I I think to your point Tommy I’m going to go back to what you were saying which was like the data

57:49 Governance part of this. I think this is going to increase the the number of conversations about making sure businesses identify how they do business. I can give you a really clear example. Time intelligence is quite complicated for people to get their head around what are they trying to actually do. So, some businesses or some teams will get in the mindset of when we measure things, we measure things week over week. , if I want the same week last year, I don’t really care about the dates. I

58:20 Just need the relative dates and time. And again, calendars and dates and times, it’s very different because yeah, as you think about these things, let’s let’s again, let’s imagine Christmas. It’s on December 25th, but that day of the week shifts year to year. And sometimes it’s on a Wednesday, sometimes it’s on a Sunday. it it changes throughout time. And so you have to think about what when you’re talking about like a marketing effort and what is impacting when people buy or acquire things, these moving shifting periods of time matter. And so how your team looks at

58:55 And analyzes the data may be slightly different from my team. But we have to sit back as like that governance piece and align on what is really important for us, right? , for example, I know like Black Friday is like the same weekend all the time no matter what. It’s always Thursday, Friday, but the day of it actually shifts. So, being able to build a calendar that allows you to shift periods of time. Hey, here’s where Thanksgiving occurred in this year, prior year, and other year, you change how you think about the the time in

59:29 The calendar and say, “Okay, Thanksgiving is day zero. The day before Thanksgiving is day one. Day two days before would be day two. And so you have to build out other intelligent calendars so you can compare the proper data year over year as to what was your performance on years. You’re not actually using the dates themselves. Right? So there’s all these really weird definitions that you have to say. We need to talk about what this means so that when we talk as a company, a business, we’re talking about the same information. And everything you’re saying, Mike, is all the hard data. That’s just PowerBI

60:03 That we’ve already struggled with most a lot of organizations struggle with where that success or failure of business intelligence is is occurring and has occurred for the last 10 years outside of now giving something in a sense of meaning like you’re talking metrics. We’re talking metrics here where like what makes a lead a lead. Right. Sure. Yes. And and Right. So, and that’s already something most companies if you talk to them don’t have a good handle on or they have different definitions depending on what team you’re looking at. Like, and again,

60:34 The point of what you’re saying here is like, do businesses actually talk about that or do businesses just go along the road and just say, “Well, Tommy, you define it one way for what a customer is for you. Mike, you’re in a different department. You’re going to define a customer differently for you.” And instead of having those two teams come together and talk about what should we define as a customer and we may have two different definitions for customer, that’s fine. You just can’t call them both customers. You need like more information like, hey, this is a, you

61:06 Know, subscription customer. That’s how Michael understands a customer. Tommy, you’re talking about a paying customer. That’s a different type. You have to give them different words and they have to have a different The biggest part meanings. Words have meaning. And if this is going to work and again we’re talking still the 2D version of this we’re not talking 3D IQ correct there has to be a shared and agreed upon definition for any data governance or data literacy to work literacy to work you could throw out the data part here man like just the idea of this this

61:39 Conceptual side on on my 2D art right now we are not everything you just said I want to be very clear has nothing to do with fabric IQ this is business intelligence frustrations in in a nutshell. So, and we’re still I see Mike this is most of my projects still around this data governance and coming in and this is the thing this is why I started consulting but I’m like oh this is everywhere and you see it too I you have the data governance workshops that you do this is obviously a big need now we’re

62:13 Introducing here to your to what I love that you said a layered another dimension to it and I think we’re having a lot of things here where I have concern about like the slowdown that companies are going to have but it’s almost a shift away from metrics right because everything we were talking about were metrics and now it’s going to be this conversation around these entities Mike I think we need to really dive into this because the data governance side of this and I and I think we’re just scratching the surface on what this

62:46 What’s going to make this successful right because the technology ology. I think Microsoft is putting the technology in place where we actually have a platform and the tools to make it work, right? Because I really do like how they’re like I’m looking at the user interface just diving in and this is still early on. That makes sense. It’s a logical way that in a sense I would try to do it from a technological point of view, from a machine point of view, but that’s maybe 20% of the battle of this working. Would you agree with that?

63:21 Yeah, I Yes, I do agree with that. I I just I keep going back. So I’m excited about ontology, but I’m a bit skeptical right now. I I don’t want to go by way of the dodo bird on other areas like the investment just gets removed from it because no one’s understands how it works and can’t we can’t align on it. So the ontology piece is going to have to be flexible enough to add value right out of the gate. And so I’ve just barely turned it on in my tenant. I’m just starting now to figure out how to play with it. Like, , the the initial UI was a bit

63:52 Confusing. You have to attach it to like a SQL database and then to get things going. So, it the setup of it is starting to figure itself out. I need to get my heads around it a little bit more. So, at the end of the day, like I understand what they’re trying to do, I think with Ontology, and I think the comment here as well, , thanks, , , chat here. , fabric IQ is looking a lot like Palunteer’s ontology, which is a very similar, again, it’s a similar concept. It’s just allowing you to provide relationships between information. I think this needs to be

64:24 Used. I think the the proof will be in the pudding here, right? Can we really get teams to start talking about defining things the way they need to define them? is it is this going to help us get on paper a bit more of like what does our business do? How does our business run? And does that really effectively move the needle forward? And that’s that’s where I’m I’m holding my breath here a little bit around ontology. I like the idea of it. It seems to make sense, but if I can’t get agents and AI things to help me build it and make it quicker and go f like I don’t want to click a

64:56 Bunch of buttons, I don’t want to spend a lot of time on it. It needs to start helping me like figuring things out. Why can’t we build ontologies based on semantic models? Why can’t we build ontologies backwards and saying here’s the pool of information that I have. Go find relationships and build out what you think is the right ontology and then I can correct it. There is there’s one concept here Tommy that we talked about I think recently on the podcast which you you look at your con your company your company already has a huge amount of semantic models already built out. Those are already parts of ontologies

65:30 That live in very small areas. How do we get from those hundreds, thousands, tens of thousands of semantic models? How do we get that back into what we’re trying to build, which is a companywide ontology? That’s where I think some of this value will become is there’s this idea of like taking the top down approach, Tommy, which I think is what you’re speaking to is this is what we define. These are our business objects. This is how we build our business. But I think there’s also a value in going from the reverse side. Here’s what we’ve built. Here’s the things that we have created and working

66:03 Our way backwards into what does the ontology look like. So I think there’s actually another reverse pattern here as well, Tommy, where we’re going backwards from here’s the items you’ve created in your organization, walking your way back up and saying this is how we run our business and then looking at the ontology that’s being produced and saying hm this doesn’t make sense. I think we need to have a conversation here around products because it seems to be very very confusing on how we’re defining them. Right? There may be things in our business that are very clear and don’t need to be

66:35 Defined. , maybe customer is very clear across all business units on how they’re building it and what they’re doing. The ontology should clearly show us, oh, this is a very well-known item. Everything links back to this single table that it’s being like whatever that is. We should be able to see that in the ontology and then we should be able to work our way back and say these other areas and I think that also provides again looking backwards Tommy through the ontology is when we see lots of random tables around the products tables because we don’t the company doesn’t understand them you

67:08 Would see a large fracturing a large variety of the same data over and over again and that would just tell me it’s confusing it’s not being thought about correctly and so someone needs to come in and organize it a bit better so that we all have the same understanding around what that data is doing. Does that make sense what I’m describing there? I So, as you’re talking, I was writing some notes and I’m realizing that we have like seven more topics right now for like that we’re going to talk about for an hour because I I I briefly say the idea of working backwards. My initial reaction is I don’t know if that will work. And I

67:40 Think that’s an episode in itself because well, here’s the thing. Let’s let let’s be for real here. We are just touching the surface here. We are just beginning to in a sense conceptualize what this is going the impact what this means. So this is new for everyone and let’s be very clear that first off if you get a message from someone that says I’m going to be your expert on ontology. We’ve been doing it for years. this scam this is going to be new for everyone and we’re all scratching this. We’re all just scratching the surface

68:11 Here. So Mike, I really want to dive in this with you and I think this needs deserves conversation around because here’s the thing, Mike. We’re finally getting into some we’re getting we’re diving into philosophy and data and I’ve never been more happy in my life. So I think with that I think my closing thoughts because I know we’re well past time. I am very excited about what happened with Ignite. Outside of the 91 features, this Fabric IQ can has the potential has the potential

68:45 To be one of the greatest impacts to my career. that’s really happened since PowerBI came out. I’m not saying it is right now, but it absolutely has the potential if AI continues to be a integrated part of my workflow, my career, my my life in an organization and the influence of the data there, then this has the potential to really be where in a sense my focus is where I’m going to say 10 years from now, yeah, bar charts used to be important. That’s what I used to do.

69:19 Are we there yet? There are so many other factors to consider that we have not even thought about yet. And I think for a lot of you people listening regardless is start getting your expertise, start learning about data governance if you’ve not done so. To me that is going to be the essential piece here outside of the technology. Final thoughts on this one. I’m very optimistic on this whole ontology thing. I think this is a great move. I think there’s more relationships across your data than we than we can

69:51 Experience. This idea of like one semantic model for your entire business, I think is a concept. I think conceptually it exists. Do I want to make one semantic model for my entire business? No, I do not. I do not care to do that. however the metadata the data that describes where our data comes from I do think I think there is a ontology of your business one that should exist for everything you do 100%. That is that is a single item that should exist inside your fabric experience or your your fabric space.

70:24 There is something everyone could sit down and on a whiteboard draw out your business in in its entirety. What what gets confusing is when you start giving different teams or different areas their particular area. There’s a whole bunch of nuances to every single area. , shipping, marketing, production, , distribution, all these things have a lot of nuances in their area. And those team members are experienced in that area. I think about at the end of the day it it’s allowing people to be flexible to figure out how

70:56 To be very nuanced with their data for their department but also be able to relate that back to the broader part of the organization and how you fit in. So yeah, I guess I guess I’ll lean on the on the fact again, , if these things are not I guess this probably could be the same thing we’ve said with reporting Tommy, right? If this ontology item isn’t saving your company money or making your company money, you shouldn’t use it. So, if it’s if it’s saving you time, if it’s saving you money with the ontology piece, so good.

71:26 I’m not sure you should you should incorporate it into your business flow. Same thing with reports. If you report you’re building isn’t making you money or saving you money, I’m not sure it’s valuable. So, I’m going to end that bombshell there. We’ll leave it at that. Tommy, I know you’re going to have lots more comments. So from this conversation, make sure , hey facilitator, make sure you add an action item for us to talk about more of these concept. We should really have facilitator on our podcast. I think it’ll help us remember things. It’s very true. And have it generate new topics from episodes. So anyways, I’ll

71:58 Just throw that in there now. Maybe we should have a new section of the podcast called facilitator. , hey, hey facilitator. That’s a great idea actually. Anyways, that being said, thank you all very much for listening. We do appreciate the chat. Thank you chat very much for participating and adding your thoughts there. We really do appreciate you jumping in and listening to us there as well. , thank you so much. We do love this podcast and we have a lot of fun on this as well. Please make sure you subscribe if you want to have episodes as they come out as soon as they’re published. Make sure you become a member. , members will get all

72:30 Podcast episodes as they are produced and also all episodes will come with no ads if you become a member of our YouTube channel. So, if you don’t like ads and you just want to listen to the podcast straight up, we’d highly recommend you become a member on the channel as well. That being said, Tommy, where else can you find the podcast? You can find us on Apple, Spotify, wherever you get your podcast. Make sure to subscribe and leave a rating. It really helps us out a ton. And do you have a qu I’m sure people have questions from this conversation today. And if you have a question, idea, or topic that you want us to talk about in a future

73:02 Episode, well, do it now. Head over to powerba.tips/mpodcast. Leave your name and a great question. Quick note on the mailbag, Mike. We’ve been getting a ton in and we are just trying to comb through them. So, please leave them. We love them. Join us live every Tuesday and Thursday, 7:30 a.m. Central, and join the conversation on all of PowerB.tips social media channels. Thank you all so much. And we really appreciate you joining us and we’ll see you next time. down.

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.