FABCON 2026 Recap Part 2 - Ep.518 - Power BI tips

In Episode 518 of Explicit Measures, Mike Carlo and Tommy Puglia unpack the latest Power BI and Microsoft Fabric topics from the show. You’ll get a quick read on the episode’s biggest ideas, why they matter, and where to dig deeper in the full conversation.

News & Announcements

• No linked announcements were available in the episode description for this post.

Main Discussion

This episode covers the major themes, opinions, and practical lessons Mike and Tommy surfaced during the conversation. The transcript below captures the full verbatim discussion if you want the exact phrasing and context.

• Mike and Tommy react to the episode’s biggest Power BI and Fabric developments and explain what stood out to them.
• They connect product announcements to day-to-day practitioner decisions instead of treating the news as abstract roadmap chatter.
• The conversation highlights where teams can move quickly, where they should slow down, and what tradeoffs deserve attention.
• They share candid perspective from real project work, which gives the discussion more practical value than a headline recap alone.
• The episode mixes tactical advice, opinionated takes, and a few forward-looking predictions about what listeners should watch next.

Looking Forward

If this episode’s topics affect your current Power BI or Fabric plans, use the transcript and linked resources to identify one concrete change you can test with your team this week.

Episode Transcript

0:22 Good morning and welcome back to the explicit measures podcast. We are yet again talking about more Fabicon announcements. Surprise, surprise. A lot of announcements have happened. So that being said, that is our main topic for today. Welcome back, Tommy. Welcome back to the show. very much. It’s weird to do It’s weird to see your face now twice a week again. Did we do this a couple of times before? So So It’s been quite a while. We’ve had about six episodes or about three weeks separate or different because we’ve been

0:52 separate or different because we’ve been just doing different things. So We’ve been just all over the place. We’re back into the routine now. It’s just over It’s extremely overwhelming when you’re gone for so long, Tommy. There’s just so many emails and things to get through. Yeah. It’s a lot. So I feel very overwhelmed right now trying to get through all the projects and caught up on things and making sure the team’s still working on what they need to be working on. It’s just a lot. Anyways, Speaking of which actually, this is a little bit ice breaker here, Mike. Are you one to check

1:22 Are you one to check your favorite tools release notes or change logs on a frequent basis? Do you find yourself doing that ever? I think it’s changing slightly, honestly. There’s there’s some tools that I really pay attention to cuz I really want to understand what’s going on. Shockingly enough, I read a lot of the release notes on Databricks. They actually have like regular releases on what they’re coming out with. Mhm. And so I’ll actually go dig dig through a little bit and find features. I don’t probably read them entirely. But I definitely read sections of them.

1:53 But I definitely read sections of them. What do you How do you read your release notes religiously? [clears throat] Yeah, so I can be very much a nerd with this cuz I love release notes. I love change logs for my for my favorite tools. Oh, yeah. So VS Code would be I’m I’m going to assume that you read all the release notes for VS Code. Well, they they make it easy to find it. So their So their blog is is always ridiculous. So And And when you ever you update the program, the release notes are right there on the first page. Like so you open When you update your VS Code, it’s very easy to see. So I I like their

2:24 very easy to see. So I I like their release notes. I feel like they’re releasing more quickly. It feels like VS Code is releasing a lot faster now. They’re trying to catch up to Cursor. Really? Oh, yeah. How fast is Cursor releasing? it’s not so much how fast they’re doing, but where they’re at. In terms of like what they have available in the tool itself. Interesting. Well, Well, in general, I guess I feel like I was looking at VS Code it was like what

2:54 looking at VS Code it was like what twice a week is when they’re releasing code. It feels like they’re more like weekly now is what they’re doing. Weekly releases on things in VS Code. things, right? What was that? A lot of agentic things. Yes, I will agree. A lot of the development seems to be coming from the agentic space, which which is great. I absolutely love that. also, I’ve seen a lot of articles around around Anthropic releasing very fast as well. And they just came out with a new announcement on something else that seems very interesting. It’s It’s kind seems very interesting. It’s It’s hard I don’t know if you’re feeling

3:24 of hard I don’t know if you’re feeling having the same same feeling, Tommy. It’s hard to keep up with all this stuff that’s going on. Yes, a thousand percent. It It feels like we’re able Companies are able to build code much, much faster. And I was reading on X the other day about Anthropic and they were saying, “We can build code much faster and now our new bottleneck is like testing, QAing, and and that area.” And now they’re working on agents that can help with like the testing and the QA. Did you read these articles about Mythos that they’re this new Mythos this new

3:55 that they’re this new Mythos this new model that they they actually built and are not releasing because it’s too it’s too wild? Have you Have you heard about this? No, I’ve not I’ve not heard about this. Dude. Dude. All right, let’s get into that Let’s get that one in a second. Let’s finish your topic. Was there anything else that is on top of mind? no., just no., just do you do you check the Databricks change log then? Yeah, a little bit. Yeah, I’ll do that there. I will say I do read the one for VS Code. Like that one I do go through. I look for like what new features are they doing? Are

4:25 what new features are they doing? Are there any major bug fixes I should be aware of? Like that one I do read. I was trying to install an extension for VS Code recently and it was not a Microsoft developer. And I was trying to look at the release notes and it was just saying like it was like nothing. And to your point, Tommy, like do you read them? No, I usually don’t read them. Yeah. But I was extremely disappointed like, “What are you changing? What’s going on? I don’t understand what you’re doing.” And I was actually making an issue on the extension because I can’t see

4:55 on the extension because I can’t see your release notes. What are you releasing? What things that you fixed? What’s changing from version to version? That was annoying. Why do you bring up the question, Tommy? What’s the reason? reason? just because with all the things with Fabric like a little little side note here like for example, I’ve told you how much I love like tool called Notion. Mhm. Yes. Yes. So anytime there’s a new feature that comes out like they release custom agents in the tool. I use Perplexity and it’s not just Notion, but

5:25 Perplexity and it’s not just Notion, but any any tool like even Perplexity comes out with something new. And it’s like I ask Perplexity, I want to find every example on the known internet of how people have utilized this, what are templates, open source repos. Interesting. So and it’s like the when skills came out, right? When skills started being popular, it’s like, “Okay, what can I do with it with this, right?” Yeah. Yeah. So I Perplexity’s great for really going through the web and finding

5:55 really going through the web and finding every known piece of articles where I’m like, “I want not just an article about it, but I want to actually see the examples. I don’t care if it’s on GitHub, whatever.” But that’s that’s what I do with my tools. Like you use Miro, right? I use Miro, yes. That’s That’s my one of them, but I I I don’t probably use Perplexity the way you do, Tommy. Like Perplexity’s like again, from what you describe it, it’s like a browser or a search agent search engine for like agents on the internet. Is that how you would describe

6:25 Is that how you would describe it? What would you describe it as?, , it’s meant to scour, like scour information on the web, yes, but like structure very well. So it’s not just doing a search. Like I still use Google, obviously. But if I want to know Well, yeah. If I want to know all the different examples of Notion’s latest releases for custom skills because their blog doesn’t do a great job. It will go through blog articles and GitHub repos and Twitter feeds and basically curate

6:55 and Twitter feeds and basically curate that all together for me. Interesting. as a side project I have had to figure out my own way of collecting data anyways for our website. Tommy, we we’ve been hitting new numbers like never before. The amount of stuff that we’ve been putting out. Just big numbers these days. So one, I got to say thank you to the community, right? So this wouldn’t have happened not with our community here. We’re putting up some very large numbers across LinkedIn, YouTube, and X seem to

7:26 across LinkedIn, YouTube, and X seem to be our main platforms, but we’re doing thousands of views per episode nowadays and we’re doing two a week., it’s pretty impressive. Wow. Yeah. We We are It’s been fun. But to your point though, Tommy, about scraping the internet and getting all the things together, this has been a challenge for for me is how do you pull all these different social media platforms analytics together into one place? What does that look like? And it’s been very difficult to get something like that going, but I have built a vibe coded

7:56 a vibe coded agentically coded an app that lets me do this. And now I have an app that helps me collect all of the data for the podcast in one single place. Very nice. And with that, I’ve been able to find out who else is speaking in our space. Like so I look at a lot I watch a lot of other channels as well. So as a maybe a poll here back to the audience, would you want to see like a newsletter about all the hottest, latest, greatest blogs, posts, things like that? Yes. Would you rather have access to an app that you

8:26 rather have access to an app that you log into and see the things? I don’t I don’t really know what’s the the best thing. thing. I have enough apps. I have enough apps that I’m trying to learn, so Okay. Yeah. All right. Well, maybe Whatever you want to build, build it. Yeah. I I’ve got the I’ve got the data. It’s just a matter of how do I distribute it? So I’d actually be curious for those who are in chat or talking in the in the YouTube chat here, would you would you be interested in something like that? Would that be something that you’d want me to continue pursuing and building? So anyways, let me know in the chat if that’s something that interests you.

8:56 that interests you. Let’s move over to our main topic today, Tommy, unless you have any other like,, startup , startup news announcements or so. Speaking of major updates, let’s just go to some of the most major ones for Fabric. Okay. Let’s jump in there. let’s let’s talk about some some features that were announced at Fabicon. I think there’s one that I think will be extremely useful. I’m I’m really I’m really interested to dig in a lot more around this thing called the operations agent. Tommy, what what’s your impression of that one? What

9:26 what’s your impression of that one? What do you think about the operations agent? So it seems to be much more focused with real-time data rather than the data agent, which is the chatbot chatbot interface. The operations agent is really meant to monitor KPIs, automated workflows, recommendations to, to users with context and be grounded in a Fabric runtime environment with or a real-time data, excuse me. And operation agents

9:56 data, excuse me. And operation agents are really meant to run in the background where I’m not going to be necessarily talking to it or users are not really going to maybe not even know it’s really there, but it’s really meant to monitor. It’s a monitoring agent and specifically with real-time data. data. So So I want to unpack this a bit more. So this is one I think it’s I think a lot of people are talking around the operations agent. Have you seen really good use cases of where this is used yet, Tommy?

10:26 yet, Tommy? It feels like people It feels like people are excited about it, but I don’t really see like solid use cases of this. Yeah, I think it’s too early to tell. Like okay, well let me ask you this. How many good examples have you seen of data agents? I think data agents is a bit more so Well, let me say this so let me say it this way. Data agents, I’ve actually used them a number of times inside Foundry. I’m using them in leveraging them in testing them out a bit more. So I feel like I understand a bit more around what

10:56 like I understand a bit more around what the data agent is supposed to do.

10:58 the data agent is supposed to do. To your point, Tommy, the operations agent seems much more focused on the event house, real-time data, what’s going on there, and reasoning, reasoning on top of real-time data. Mhm. Mhm. I don’t know who’s using the real-time data portion, but I don’t think it’s like small or medium businesses with smaller FSKUs that are using real-time data. So data. So this is where I’m I’m looking at this going like okay, this feels very very

11:28 going like okay, this feels very very specific to the real-time story. Yes. Yes. You’ve got You’ve got trucks moving around. You’ve got,, things happening in real time. You need to make decisions. You want to know where things are at like right now immediately. Did something get delayed, right? Those are the things that it feels like the operations agent makes more sense in. I feel like there needs to be a general agent that can just do things, right? And maybe I’m just missing things. What’s What’s your initial read here, Tommy? Am I on Am I on point here? Is

11:58 Tommy? Am I on Am I on point here? Is this Is this how you’re feeling about this? I don’t know how much traction we’re going to get here when something so catered towards just a real-time piece. piece. I I’m I I’m I was hoping you wouldn’t say what you just said because that’s the feeling I was getting to. Okay. And I was like maybe I’m reading this wrong, but no. It really turns out that yeah, operational agents only work on event house data, which again is not everyone. And I feel like there’s so many other good use cases for what an operational agent could be. It doesn’t shouldn’t

12:29 agent could be. It doesn’t shouldn’t necessarily just have to be real-time data. Could be monitoring my fabric environment. It could be monitoring APIs. APIs. It could be monitoring semantic models, just background operation. Yeah, background operations, not necessarily real-time. So that’s a little frustrating to me because they’re trying to put it in par, with data agents in terms of hey, we have data agents and operational agents. So they work together, but they don’t. they if you have obviously

12:59 don’t. they if you have obviously the event house. So that’s a little frustrating because I feel like there’s so much potential for it. What I What I feel like I would like, Tommy, and maybe this is part real-time, but part but part When we talk When we had Chris on talking about real-time for us on the podcast Mhm. he he said something that was really that resonated with me, which was real-time doesn’t have to be like a full stream of lots of volume of data, right? Real-time can be

13:31 not random, but like event-driven actions. Let’s say it that way. Right? An event occurs, something happens, and then the agent takes over and does something or the and then the pipeline runs or and then something action happens. It doesn’t have to be real-time meaning some data event is occurring and then I need to do something on top of it. I feel like attaching this only to the event house seems like a point in my it seems like it’s a part of the solution.

14:02 solution. But if I look at like data activator that’s that’s what we want, right? Yeah. Like why wouldn’t the operations agent be able to attach yourself to a data activator when this file shows up, when this model refreshes, do something. When this, having events that are listened to at the data activator level and then when those events occur now trigger some action that the agent will do things with. And that’s I feel like that’s most of my use cases

14:32 I feel like that’s most of my use cases when I’m trying to apply agent level things things on various tasks, it’s that kind on various tasks, it’s that stuff, right? When this occurs of stuff, right? When this occurs randomly, not randomly, occasionally, like once per day, once per week, whenever that event triggers, that’s when I want something to pick up and do some agent agentic workload. Let me give you a Let me give you a very clear example, Tommy. On one of my websites, I’ve got a fill out form, a contact me form. And so I have people fill out some information, and I could have done that with like a Microsoft Forms, but Microsoft Forms doesn’t really let you customize it too

15:03 doesn’t really let you customize it too much. You can’t just send the Microsoft Forms the data. Sure. to basically iframe it. So because it doesn’t give you the ability to like customize the form in a good way you need to like build your own system. So I built my own. You go to this form, you fill the information, you hit send. Right now Right now what I really like about it is it sends the data over to NAN with a with a webhook. Yeah, yeah. Here’s the data, and then the webhook picks it up and says, “Oh, I’m going to

15:34 picks it up and says, “Oh, I’m going to run an agent against this.” And then the agent runs through and says, “Okay, event shows up, process the image. Is this a real request or is this a fake request? Did someone just make something up or is this Did someone actually say, ‘Hey, I want to see a product demo.’ Right? So cuz I I don’t want to look at things that are not real. I don’t need it to send me every little thing. So the agent does some reasoning against the data that it receives and then decides what to do next, right? If it If it finds that it’s a real legit request request it routes it on to a second agent that

16:06 it routes it on to a second agent that does, ‘Hey, write my email. Go research the company. Go figure out what’s going on. Oh, and by the way, send Mike a Teams message so he knows that this customer reached out to him.’ So like all this is happening agentically. This is like a is like a maybe for lack of better term like a workflow that’s occurring. Yeah. That’s what I like When I think about that experience yes, it’s real-time in the fact that someone hit submit on a website, but that happens like once a week, every couple weeks. Like it’s not very frequent.

16:36 frequent. Same thing I’m thinking about like Power BI tips. BI tips. We used to have a mailing list where you would go in and sign up for the mailing list. Again, now that I have this data about who’s talking about cool stuff on the internet I’d like to revive the mailing list and say, “Hey, last week, here’s all the interesting things that came out this last week from different sources and who’s talking about different things.” Blogs, posts, videos. I think people would like to see that, an aggregated form of content around Power BI.

17:08 form of content around Power BI. Well Well there’s no way on my website to like That’s an event-driven thing. If someone signs up, I want to take that name and put it in a list. That feels very operations agent or something an agent should be able to build for me that I can make it very easy to do. So this is where I’m I’m getting a little a little bit more like I I don’t know. Like it feels useful, Tommy, but it feels like a very small use case of what a what a broader solution or tool could could could be doing. Yes, 100, 000%. honestly, this is how a

17:38 100, 000%. honestly, this is how a lot of if you were learning about AI, you’re learning the back end. That’s how they work. It doesn’t It’s not necessarily real-time. There’s triggers, sure, but Microsoft figured that out with Power Automate 10 years ago. Yeah. in terms of trying to wait for real-time information. It doesn’t have to be in a real-time database or an event house, which is, , it’s just odd to me that the data would live there. most Power Automate flows and also

18:08 flows and also like take any tools, Zapier, any of those tools like that, they actually run on triggers from databases, SQL databases. Yes. so it’s very strange to me. But I hope just like they released Copilot for everyone for F2 eventually that they will eventually look at this and maybe try to consider something else for data operational agents, but there’s a lot of potential there.

18:39 there’s a lot of potential there. Hm, interesting. Yeah, I think you’re right, Tommy. Anyways, I just want to unpack a bit more. bit more. I’m I’m feeling the same read on the operations agent announcement that you heard that I heard. it seems interesting only if you’re in the real-time space. Okay. Other major announcements from Fabric Con, Tommy. Is there something else that that resonated with you, something else you’d like to chat talk about or discuss or unpack? there’s so many, but I think it’s at least worth touching on Fabric I Q again. Not necessarily diving in, but

19:09 again. Not necessarily diving in, but just to keep the drumbeat alive of what Fabric I Q is going to be. They did start releasing a little more context around it. Because if you remember, they made this major announcement. Do you remember what month that was? that was? When they said introducing I Q, the I Q developer. And you’re like, “Woah, that’s going to be amazing.” And what is it? We don’t know yet. There’s a feature you can use in Fabric, but doesn’t go anywhere. Okay. Well it seems

19:40 doesn’t go anywhere. Okay. Well it seems like we’re getting a better idea of what Fabric I Q is, and it’s not necessarily its own tool, but it seems to be the layer of your semantic model, Fabric the graph, which was also released, released, data agents, and operational agents, and also built built all on this thing called an ontology layer, which is basically like the business semantic layer. Yes., so , so and I think that’s why we were all like, where where when does it get released?

20:10 where where when does it get released? It’s like, well, that’s not really what it is. it is. I really think this we’re still seeing just the beta version of what Microsoft’s planning to do with this because for me, Mike, I really liked their messaging and we actually I’m pretty sure we did an episode on Yep. the AI developer rather than necessarily the data developer and is that the future for us? Yes. So, let’s let’s talk about Fabric IQ.

20:40 So, let’s let’s talk about Fabric IQ. I believe Fabric IQ is really going to be is well, it’s focused around the ontology ontology feature, it seems like. So, when you when you read a Rune’s messaging here,, Fabric IQ ontologies will soon be accessible and also available through MPCs, MCP servers. apparently that the MCP server is currently in preview. It’ll allow agents to discover, understand, and act on the semantic layer of your data. Ontologies can also serve as contact source maps soon in operations agents in Fabric extending

21:11 operations agents in Fabric extending shared business context directly into your operation and decision making. Again, it it feels like a lot of the Fabric IQ piece is really heavily vested in the ontology space or the ontologies item that they created. and Tommy, I think you and I looked at the ontologies and we’re like, it’s limited. It doesn’t really do a lot. lot. Yeah. And it was difficult to build my own ontology by itself. I basically have come to the realization that the Fabric IQ ontologies piece isn’t for mere humans to play with. Like

21:42 isn’t for mere humans to play with. Like you’re not going to build ontologies using the UI. You’re going to create ontologies using like some code or agent or something. You’re going to let something else build them for you. I just again, this the messaging there

21:58 I just again, this the messaging there is interesting, but I’m again, I’m with The thing that’s missing for me in the Fabric IQ piece is where’s the value? What what what value does this bring long term? What’s that story there? Does it make my agents smarter by 10%? Does it make them access less data? Does it make it more efficient when it runs CUs? I’m not hearing any results of it, just the fact that it is you’re giving knowledge to your agent of your whole business. Well, I guess I still got to program the whole ontology.

22:29 still got to program the whole ontology. Yeah. So I I think that’s Is that Is that how you you’re viewing it too, Tommy? the ontology what they initially announced was like, hey, if you want to build your own ontology, just connect to the semantic model and we’ll build it for you. I really think that there’s going to be this huge part is going to be like you said an an agentic layer of AI is going to be doing a lot of the building. But again, Fabric IQ really isn’t a one particular tool in the in the toolbox. It’s like the all

23:00 the toolbox. It’s like the all the background things that have to do with planning, agents. so, I guess there it’s more marketing team than it is developer team on this one. So, He’s right. He’s there. Yeah, so this is more I feel like this was much more marketing winning than the developer team because Fabric IQ isn’t necessarily like a spot you go to in Fabric or connection, right? Well, who knows how long it’s going to be until we have IQ hub, right?

23:30 be until we have IQ hub, right? [laughter] This is going to be This is going to be another hub., I’m more excited about the database hub because I think that makes sense, right? I’m starting to I’m starting to experiment in Fabric with other databases. It’s not just Sure., I’m I’m not just doing a lakehouse anymore. I’m I’m using a lakehouse in connection with like a SQL database. That that makes sense to me. Like I like that connection point. How’s that going to work with Postgres? They’ve announced that one. How’s it going to work with MySQL? How’s it going to work with Cosmos? Does Does the operational side of data

24:01 operational side of data make it easy to get the analytical data down to lakehouses or other storage methods that makes it easy for me to get that data out, right? That’s the information that I’m really really interested in now is how how is that going to change? So So Yeah, and I I guess that’s what you would say Fabric IQ is. It’s more or less a process than it is a tool. So, at least for now it feels like it. It doesn’t It doesn’t feel too solid to me yet. yet. Which is very frustrating because it sounds so cool and like I said, their

24:31 sounds so cool and like I said, their marketing team did an excellent job. But that’s like me going up to my team and going, hey guys, I have a really major announcement. We are really excited to release release putting your stuff away for Puly BI consulting. Yeah. Which is basically the process that we do because it really is. It’s again, it’s using the ontology, using planner, planner, the fabric graph, the data agent, and semantic models. That’s Fabric IQ. Yes, I would agree with that one as well. well. Okay, let’s move on from that one. So,

25:03 Okay, let’s move on from that one. So, let me let me just step back here. One of the the main threads of excitement that I saw throughout the entire conference was just this whole emphasis on agents in general. Where do you apply them? And my opinion here, Tommy, is the places where the most excitement has been coming and even recently I’ve been retweeting and reposting on LinkedIn a lot of news articles or a lot of articles. There’s a ton of people aggressively building on top of agents

25:34 aggressively building on top of agents Mhm. that let you create reports. And I think there’s a large amount of need from the community that is clamoring to build reports with agents. agents. Let me just pause right there. What What is your What is your sentiment on this, Tommy? Are you seeing the same intensity, excitement? Like I don’t know what I don’t know what you want to call it, but are you seeing the same thing that I’m seeing here? Yeah, I’m not just seeing it for a report building even though that’s always been the

26:04 though that’s always been the sore spot. I’ve just been seeing it for any especially with Fabric any way you can connect to Fabric now, there seems to be a skill or a really neat way to connect to it and more than just a single connection. there’s a ton of skills now that people are skilled for notebooks, right? Unpack that. What do you mean What does that mean to you? So, what I’m seeing right now is a lot of agentic tools for interacting with your Fabric environment. Such as? Like What do you mean you say that? Notebooks. Yeah, so if I want to start

26:35 Notebooks. Yeah, so if I want to start actually interact, see what my notebook’s doing from Claude, I can do that now. If I actually want to run something from my command line or update something from Claude code and update something in Fabric, I can do that now. So, all these just these additional like almost MCP like mini MCPs just around the Fabric environment is seem to be where I’ve seen the biggest focus, but and and then on PBIR, obviously, there’s just been just because it’s so good for

27:07 just been just because it’s so good for anything it’s developers. I’m not going to say that I was right here, but I feel like I’ve hit the right I feel like I saw this way earlier than other people did, right? Tommy, we’ve been talking about this for a while now, which is Microsoft keeps pushing agentic things on top of the here’s a pile of data, agent go find the answer for me and then ask the agent for the answer and it does something and returns a result. result. I have just found that experience that

27:37 I have just found that experience that, offloading the search and reasoning about the data directly to the agent isn’t as effective. It just doesn’t seem as effective to me. For for use of my time, for use of getting the answer back, why would I why would I always come back to an agent and ask it the same question week after week to get me the answers, right? right? I think the real and this is how I’ve personally developed with agents or AI things. Yeah. The strength of AI is creating things you don’t have time to

28:08 creating things you don’t have time to create. That that’s the that’s the strength. It’s the creator story. How do you use an agentic system to help you create? That to me, that’s that’s the part. That’s the thing I really want to get into is like we need to build that. And I feel for now for the first time after the Fabric conference, people have been emboldened to oh, to oh, we can point agents at the PBIR. We can make a PBI tools CLI that directly talks

28:41 make a PBI tools CLI that directly talks to to an agent. It’s almost like bring your own agent. I think I actually said this in the podcast a while ago, Tommy. Like what I want is bring your own agent and build with Fabric. Like that makes sense to me. If you’re paying for Claude code or Anthropic or whatever the Gemini, whatever Google product is out there, it doesn’t matter. It it’s more of what agent do you have and I’m going to use a term here, Tommy, that I heard a lot at the conference. When I say the word harness, the agent’s

29:13 When I say the word harness, the agent’s harness. harness. The agent’s harness. Do you Do what like is that I just got introduced to that. I say the word and people are like, huh? Like Do you under like when I say the, the agent harness, what do you What does that interpret to you? Because I want to just unpack that a little bit because I think I think this is what we’re getting into is people are building different harnesses for AI to help you build with Fabric. And I think that’s where the real investment’s coming from. I’m [sighs and gasps] I’m going to let you I think

29:43 I’m going to let you I think you’re going to do a better job than I am because I’m just beginning to see this being get give me your current understanding of what harness is and then I’ll unpack it from there because I’ve I’ve been feeling like I’ve been spending a lot of time time unpacking what is the harness and how does it work? Yeah, to me my best take at it is is the model, the context, the tooling that it has and in your own environment. So Okay. Yep. Yeah. So I’m going to I’m

30:13 Okay. Yep. Yeah. So I’m going to I’m going to change your words slightly there. there. But I believe so there’s you’re right. The model is like I I saw someone do a really good like a diagram on X the other day. They had like a computer, right? The analogy let’s think about a computer. A computer has a hard disk. A computer has a CPU. A computer has, RAM memory, right? the computer has like input output mechanisms, the keyboard, the screens, the mice, right? So there’s all these different things that contain a computer.

30:43 computer. And we can basically say an agent is an equivalent to a computer. In in in many degrees. You can abstract it to that level. And And there is there is there is some fundamental things like the CPU on your computer. Without that, nothing would run. Right? It does It does doesn’t work. The operating system, Windows, Mac, Linux, whatever the operating system is like the harness. Right? So if the CPU of your computer is like the large language model, the harness is the

31:14 language model, the harness is the software that contains what the large language model can do. Literally all you do to the the model, all you do is you send text to it or an image to it, right? So you you send physically an a model the large language model, you just send it some information. And all it does is it returns data back to you. That’s it. That’s all that it does. If I send it code and ask it to fix it, it will return code and say here’s what I’ve corrected, right? It’ll It’ll respond with just text back to you or an image or or a generated image. That’s

31:44 image or or a generated image. That’s it. It the model isn’t that great in and of itself. What I think is happening now is the harness is the Claude code application you download on your desktop. The harness is VS Code and the chat window that you use to talk to the agent, right? The harness is how do you swap out which model’s being used at the right time when you’re chatting or discussing something with with an agent, right? Those kinds of elements are now part of the harness. What we’re

32:14 are now part of the harness. What we’re seeing right now I I did you I’m sure you saw it, Tommy. I thought I think Claude code Claude code just released some agent management. Did you see this? Shoot, that one I actually missed. So I I think it’s I it’s like scaling managed agents or something like that. Something along this line. It’s yeah, it’s called introducing Claude managed agents.

32:44 introducing Claude managed agents. Tommy, I’m telling you like think I’m having a hard time I’m literally like feeling like I’m I’m I feel like I’m behind a fast car and there’s a rope on it and I’m holding on for dear life because I’m

32:56 I’m holding on for dear life because I’m like just getting dragged across the dirt or it’s like one of those old cowboy movies like, someone’s the guy fell off the horse and he’s got something stuck on his foot. He’s just the horse is just running and I’m just getting dragged behind the horse. Something. Every other day something is coming out that’s like game changing and revolutionary. Manage like so another part of this is how do you have a harness to manage multiple agents, right? How do I have many different agents doing different things? How do I set up custom agents? The harness is all of this the code or software around the large

33:26 code or software around the large language model that supports this. Okay? Mhm. Okay. Does it Does this Does this making sense? Is this like making sense to you? is basically Yeah, yeah, yeah. So So when we unpack the concept of the harness harness you can now step back and say, well every tool has a harness at some level, right? Cursor has its own app, right? VS Code has its own harness. Every CLI has a a semblance of a harness. Like what are the commands? What are the skills? How do you interact with them?

33:56 skills? How do you interact with them? what is displayed on the screen? Like all of the the code that makes it work is part of the harness. So So that is my framework. And so when I look at like, okay, what is Fabric doing? What is the harness of Fabric? And how And I’m I’m comparing that harness to other tools that I’m used to. And to me this is where the gap’s widening. It It’s getting better. Microsoft is I think it’s finally identifying that the harness in Fabric is not sufficient

34:27 harness in Fabric is not sufficient to keep up with what everyone else is building. Does that make sense what I’m saying? saying? 100% 100%. Because you’re seeing the community do this really Microsoft’s just doing a few things. You but you’re not talking about Microsoft releases, correct? So I I I feel like what’s happening right now in the Fabric market arena is Microsoft is focusing on we need to build better APIs, we need more control around what people people talk to talk to and giving so remote MCP servers.

35:00 and giving so remote MCP servers. Th- This is I think a play to bring your own agent. Right? Hey, we’re going to run the MCP server. Your agent can just talk to our remote server. Great. I like this. This is a I think a big win. But now we’re getting this like fractured platform maybe, right? So now there’s a Power BI data modeling MCP server. There’s going to be a lakehouse MCP server that was announced at at FabCon. There’s going to be a real-time intelligence MCP like so every So now

35:31 intelligence MCP like so every So now every there’s an ontology MCP server that sounds like that’s coming. Like as I read these articles, there’s a lot of things that are coming down the pipe and it’s all it’s what we’re what we’re seeing is a lot of surface area that we’re giving agents ability to to talk to. to. And so the harness is going to continue to become the most important thing that you work on or the most important thing that’s going to actually enable you to adequately leverage the agent or the the large language model. model. Okay. Okay. And that’s where I see the community being excited. I think the community’s

36:01 being excited. I think the community’s building their own harnesses and like so Kurt Buhler just made a project with somebody else. I don’t remember the guy Maxim I think is his name. I don’t remember the name. but it’s called PBIR tools. And it’s focused They basically built their own harness around large language models. Use any model you want and talk PBIR and build PBIR things directly with their agent. Th- This This is what I’ve been asking for for months. This is why I was extremely frustrated

36:32 This is why I was extremely frustrated with the way the the state of things in January January because I had nothing that I could use an agent with to actually do things in Fabric. It just felt so limited. Oh, yeah. yeah. And I I feel like after FabCon we’ve had this acceleration or this wind of accessibility and openness and allowing me to do more things in agents. there’s an extension Tommy that I’m using right now from VS Code that I think is very much worth it for anyone to try. to try. Especially if you’re building notebooks. Tommy, you build notebooks inside I know

37:02 Tommy, you build notebooks inside I know what you’re going to referring to. It’s the Fabric data engineering extension in VS Code. It’s pretty stinking sweet. And there’s a lot of I’m in the preview. I’m looking at the I use the preview version of it, right? I’m using the Fabric data engineering extension. You sign in, you connect to your Fabric subscription, it logs you in. You can see all your notebooks across all your workspaces. You can click on the notebook. You can you can open the notebook up remotely. So the notebook is

37:32 notebook up remotely. So the notebook is essentially from the service opened on your VS Code. And I can then leverage This is why I want to go I want to talk about the harness again, right? This simple extension allows me to bring my agent that I’m already paying for somewhere else into that notebook. Th- That’s to me that’s huge. Because now I can build skills. I can have examples of things. I can have, how I work with my agent, what agent I like to work with. It’s now bring your own agent to notebooks.

38:05 bring your own agent to notebooks. This is where I think we’re going to see a lot of major up- These are the minor things though too and I as much as I love it, Mike, when you think of Do you expect that Okay, let me actually I know how I’m going to ask this. Do you expect anyone who’s building notebooks or using notebooks should be using this tool? Because here’s the thing like let’s say you’re a Fabric developer or you Mhm. Good question. Yeah. And I don’t know where we’re getting to the point where this is

38:36 we’re getting to the point where this is like, no, you should you’re expected to know this. You’re expected to be using this to some capacity. Not like, hey, if you want to use an AI tool. Because yeah. Yeah, let me let me let me give two I have two thoughts on this one, Tommy. First thought is I’m willing to use VS Code as much as possible as much as humanly possible with the condition that it it connects well to Fabric and doesn’t break the connection very often. It It needs to be

39:06 connection very often. It It needs to be very reliable. So for example if I connect my notebook from VS Code on my local machine and I’m running a kernel in Fabric that’s letting me like execute the notebook and returning results results the session between my VS Code and the service better be rock solid. Same way that like Power BI Desktop is. In Power BI Desktop, I can go edit a direct lake model using Desktop. It runs really well. It doesn’t break

39:36 It runs really well. It doesn’t break the connection. I can make edits. The edits go right to the service. Absolutely love it. Like this is the the fact that I can get a local machine of something to work really well with a remote server or remote model, awesome. Absolutely love this. That’s the same expectation I’m going to have on every single item in Fabric at some point, I think. It’s going to be where’s the VS Code extension for the equivalent item? And if I’m a pro developer and I’m stitching multiple agents together, I need the ability for me to connect into the system securely

40:08 me to connect into the system securely and do an edit what I need to and review the output with my tooling. Does that make sense? Sure. Yeah. Oh, yeah, I would agree. So with So that is my number one point. The second point is the the So the first one was,, I got to be able to connect to it and it has to have a secure connection. The second condition for me to really move towards this direction will be when I’m I need enough controls in

40:38 controls in VS Code VS Code to turn on and turn off or interact with the stuff in the service. So one thing I’ve been playing around with this Fabric Data Engineering extension. Again, I’m in the I’m using the early pre-release. So I’m trying to get all the latest features out and it it’s awesome. I’m actually loving it. I can go in, I can go pull down notebooks. I’m opening notebooks up. I’m running them on my VS Code but through the service. service. I’m opening my agent. I’m asking it to fix things. The notebook experience in

41:08 fix things. The notebook experience in VS Code is better than what’s in the service. So I guess all I’m saying is I would rather swap out the harness of powerbi. com or sorry, fabric. com with VS Code harness. Like that’s what I’m saying. I’m looking to swap to the harness harness because the harness I get with VS Code is way more capable. Does that make sense what I’m describing there? there? Yeah, yeah, yeah. No, I I completely agree. So So I guess then your conclusion then that’s you. In a wider team like for normal people who just get paid to build notebooks in Fabric, what’s your then your

41:39 Fabric, what’s your then your expectation there? Yeah, it’s a good It’s a good question, Tommy. Tommy. I I’m going to argue if if you already if what you’re doing, giving agents to people who know what they’re doing only emboldens them more and helps them build better stuff faster, period. Mhm. So I’m going to recommend if you already build notebooks in Fabric, the harness that you get with AI in Fabric. And again, also, Tommy,

42:09 AI in Fabric. And again, also, Tommy, when you use the Copilot in Fabric, you’re charging yourself CUs to use that in Fabric. Oh, yeah. So if I use VS Code, it’s a bring your own model. I pay for that entirely different. So So May- maybe the maybe the point is this, Tommy. I’m going to pay for my developers, my my premium developers, right? All of them are going to have some agent. agent. It It seems like right now businesses

42:40 It It seems like right now businesses are directly buying licenses for developers or engineers in companies. Even not even them. Even like middle tier. tier. We already have people buying subscriptions for them across different platforms. What’s the best way to maximize that investment? Give them tooling that allows them to bring that agent in whatever they’re building. building. So So if if our organization is already pushing for and I I’m in a lot of organizations right now, all of them are pushing for some form of we need to buy

43:12 pushing for some form of we need to buy a Perplexity, we need to buy Claude Code, we’re going to buy Cursor, we’re going to buy something from OpenAI. They’re Everyone’s buying something. It’s going to be bought. The question is how do you use what you’ve already invested in inside the Fabric experience? And where I see it working well right now is the creator experience leaving Fabric and using these agentic tools that you’ve already paid for in VS Code or other harnesses or now building your own harness. Again,

43:42 or now building your own harness. Again, Kurt Buhler built a Power BI CLI for PBI-R tools. Awesome. Awesome. But now that doesn’t have to live in Fabric. You’re You’re not doing thing You’re downloading PBIP files and doing

43:54 You’re downloading PBIP files and doing them locally. This is This is the harness has moved away from Fabric, right? Because it’s just not capable enough. enough. And I don’t Me personally, I don’t want to pay Tommy 40 bucks to have a GitHub Enterprise Copilot subscription and then pay CUs to run a Copilot in the service. I don’t want that. I completely Yeah, yeah, no, I I know what you’re I don’t want to pay twice.

44:26 Right? Yeah. All right. All right. All right. So So what do you think? What do you What’s your reaction, Tommy? It is weird the disconnect between, you It is weird the disconnect between,, living in the tool and then just know, living in the tool and then just what we can do in like You said living in the tool. Living So back in the day when if you were in wanted to build in Power BI, you’d live in app. powerbi. com, you’d live in the desktop application, right? Like Or if you were in Databricks or if you were in Azure, right? Like you lived on those websites or that platform to do

44:57 websites or that platform to do everything. Yeah, I would agree. If I’m if I’m using Databricks, everything I did was in Databricks. Like that was the tool I lived in. But I think this agentic integration is what’s pushing people to other tooling because the agentic integration that we want doesn’t exist. Yeah. No, 100%. That’s That’s why we’re seeing the prolific prolific Kurt Buhlers of the world and and Marco Russo generating these open source skills and CLI CLI tools and all these other extensions. That’s another whole ball of wax you

45:27 That’s another whole ball of wax you need to unpack is like the skills. Like how do you even distribute skills to our team? Like that’s a whole another mess in and of itself right now. Like how’s that going to work? Let’s Let’s say that for another day, but I know we’ve we’ve obviously dived into the AI. Mike, I did want to ask you though about we didn’t really give a lot of time to three things yet. We haven’t talked about the Fabric Runtime 2. 0. We haven’t talked a lot about the database hub.

45:57 database hub. And we haven’t talked about Tim B View on the web, which was also talked about. about. Okay, so let’s let’s get out of this agent thing for a little bit here. I think we’ve gotten a little bit down a rabbit trail. Let’s pick up Which one of those topics, Tommy, do you feel would be most relevant to talk about next? I’m excited So to me, I’m excited I’m excited about it. Tim B View is great. I love it. It’s got to get there to the web. I think you’re going to need it there. there. Totally dig it. So that’s that’s just got to happen. Cuz then it matches more of what desktop is already doing. Yes, yes, yes, yes. We

46:29 is already doing. Yes, yes, yes, yes. We know. I know you love Tim B View more than anyone. [laughter] So So It’s so good, Tommy. It’s It’s like I don’t want to build models any other way now. Like I I only want to use Tim B for everything. It’s crazy. That’s crazy. I I Tim B is fine. What do you mean Tim B is fine? It’s fine. It’s fine. It’s fine. It’s great. You like writing the BIM file better? Oh, yeah, that’s what I do. No, I actually write the the old the old Binary? You just write it in binary? You take I write ones and zeros, man. So

46:59 take I write ones and zeros, man. So [laughter] But But What’s What’s your hang-up on the Tim B format everywhere? I No, I have no hang-up on Tim B because obviously I rely on Tim B with Git. Sure. I’m like you, I’m not necessarily going in like a crazy person and going line by line to edit the Tim B file myself. Oh, I’m not editing it line by line. I’m using it with agents. I’m telling my agents to go do stuff. Like there’s there’s already a skill I love agents because I also then love JavaScript and Python, too. [laughter] That’s true, Tommy. So

47:31 That’s true, Tommy. So the the I like also that it’s very human readable, right? So I will say this, too. It’s also extremely readable as well. If I need to make quick changes, it’s awesome and the way to go. Tommy, I don’t use Tabular Editor C# scripting anymore. It’s It’s literally dead to me. Like I don’t want to write it. I don’t want to mess with it. If I’m If I’m building automation, it’s building automation with skills and agents now. And if I do need to write a C# script for Tabular Editor, it’s right to an agent

48:01 Tabular Editor, it’s right to an agent and tell it to build what I want Just have it build the thing, right? I don’t I don’t want to I don’t want to muck with it anymore. So I’m writing less and less code everywhere I look now. Everywhere I look, it’s less and less code. code. I guess, Tommy, this will be a very bold statement. How long, Tommy, until we stop writing DAX? How How long until that disappears? Like we have measures in models right now today. I I want to say about a month and a half ago. Well,

48:31 Well, it’s not It’s not difficult to train an agent on some DAX. It’s getting better already. already. they’re getting them the higher-end models are reasoning really well about DAX. DAX. And now that I can send it pages and web articles and links to things, I can I can send it to these different places. So Well Is DAX dead, Tommy? It’s not dead, but It’s going to be used, but I’m Do I write it? Like do Do I need to understand it or do I just make my agent understand it? Let me give you a pretty direct example here. Okay.

49:03 you a pretty direct example here. Okay. So again, I told you I I love Notion. Notion’s an awesome tool and what really has made it go out of the park for me was this idea of custom agents. Anyways, the idea is all when I manage when I work on a project and I use Notion to manage the notes, I use it to manage it does transcripts, I resources, etc. So Notion’s like a hub of all the information. I’m talking to the Notion agent. I’m like, all right. We just had a meeting. they want to do,, the everything in the world. I just

49:34 the everything in the world. I just talked to the tech team on what what’s available. So what I want you to do is basically generate what are our blockers right now? Like what data do we need because I honestly first time I’ve ever looking at these data sources and, it’s very we are going from old Excel things that came from out of a application and now we want to build this and make a database. So like what data do we need? What’s the what are the keys? We’re going to be building a power bi. bi. So it spits it out. I’m like all right, here’s what I want you to do. I want you to write some instructions

50:04 to write some instructions to that will be sent to Claude. Claude is going to connect to this and it’s going to basically tell him once we get the files from the client basically where to start what, like what is he looking to do? Sure enough it generates not just a single instruction page but in notion you can link pages together. So it’s like here’s the project. Here’s the tech specs. Here’s our concerns. Here’s the end report result. Claude I tell Claude because it’s connected with the MC MCV server.

50:36 connected with the MC MCV server. Like all right, I have power bi desktop open open and I’ve connected some of the base files that the client had. Based on that I barely didn’t barely tell him what to do. I’m like just read notions instructions for you and let’s start building. Yeah. Yeah. Literally Literally not just building Mike but also hey, I noticed that, this is an ambiguous column. We can’t create a relationship. Okay, I’m going to go right to notion and we’re going to put

51:07 right to notion and we’re going to put that on the concerns area. I’m going to also put that in your agenda that you need to talk to the client about, what’s X Y and Z. Sure. Sure. So it wasn’t just building an agent. It wasn’t just or building a report not doing tax. It was doing a better job than I ever would have monitoring and managing all that of those sources. And then sure enough we had our weekly call with the team and like hey, what do you want to bring up? I’m like yeah, I just had a few questions and they I sound like the smartest person in the room.

51:38 like the smartest person in the room. Anyways Anyways to answer your question the smartest person we are Tommy you are the smartest person. no absolutely not. Absolutely [laughter] not. But it was like these very in point questions that there’s no way I could possibly know what those things are just on my second week of knowing the company and what they do. Right? So So I’m let me let me put some language language into what you’re saying here and see if you think this is fitting. I’m interpreting what you’re saying. I’m hearing what you’re saying. No So it sounds to me like notion is this

52:09 sounds to me like notion is this interesting area. area. It sounds like you’re saying notion is a harness for an AI because it’s that’s a great way to put it. Right? So so this is this is why I think the term you’re going to hear a lot on any social media people are starting to get it their head around this is it’s the harness is is so important, right? So going back to what we were talking about earlier, your notion experience is harnessing large language models in a way that lets them operate alongside you with you and taking notes, doing

52:40 with you and taking notes, doing actions, making pages and basically the notion experience is part of your memory for a for a client. Like here’s the client. Here’s the things we talk about. You can funnel to notion all the all the transcripts, notes, agendas, next meeting and you can just command notion to do hey, set up meeting notes for next next our next meeting on Tuesday. Let me ask this question Tommy. Can you through notion tell notion to schedule a

53:12 through notion tell notion to schedule a teams meeting for you through notion and then take notes on how to get it started and like what needs to be there for the the teams meeting? Is that something that exists today? today? you could. I know that I can interact with it with slack so I’m pretty sure there’s a way to do that. Okay. So there’s probably an integration that that would let it schedule things or even outlook for that matter or your email, right? It could it could get access and leverage those things to talk to them. to them. So So I guess what I’m hearing you say Tommy is

53:42 is notion is just part of your consulting experience that harnesses part of the the AI for you. It’s one of those harnesses that you’re that you’re leveraging. Yes. 1000% 1000% [sighs and gasps] I like what you’re I really like what you’re saying here Tommy. Tommy. My challenge my my issue here is where’s the notion from Microsoft? Where’s the equivalent of that? They’ve already got all this this is where I’m like getting a little bit

54:12 where I’m like getting a little bit frustrated sometimes because it’s not it’s not like one particular product. It’s not like fabric or power bi specifically. It’s like the integration of tools. Why are other companies integrating things better than what Microsoft’s doing? Microsoft has all the products. We have loop. It’s supposed to be agentic anyways. anyways. Where’s the integration like it does with notion? It feels like somewhat of the equivalent product. Where’s all the integration hooks? Why can’t I go into a a loop document and do the same thing you’re doing here Tommy? Like hey, go research

54:42 doing here Tommy? Like hey, go research this and the agent just says yep, no problem and it goes to do it, right? This is This is I get frustrated because I see all these other tools out innovating Microsoft and I know Microsoft will get there but it

54:53 I know Microsoft will get there but it just feels like we’re behind. Well, I think the difference is literally notions one job is to be this tool and Microsoft has to manage a thousand other products at a time. So Yes but with the with the world of agenticness now like Anthropic’s literally writing programs that can talk to Cobol and like taking away, billions of dollars of stock revenue because oh by the way

55:23 by the way every person who uses Claude code can now write old programming languages that typically IBM had, right? Developers that that were there. And and the fact that Claude code and Anthropic is just ripping out new code pieces that are just solving big problems over and every week. week. Where’s the same level of speed development coming out of the Microsoft stack? I don’t see it. I don’t know any

55:54 Do we even talk about Yeah, yeah, yeah, yeah, yeah, yeah. So okay. All right, Tim Doll we just finished Tim Doll. Let’s move on. We want we That was Tim Doll? It’s like everything we talk about it like funnels back to like agents and harnesses. Like okay, let’s move on. What other what other final topic? We have enough item here for maybe one more announcement that you want to unpack. unpack. let’s talk about the let’s talk about the database hub because I because who doesn’t love hubs? So Well, I mean who doesn’t love hubs? So Well, I mean who doesn’t love hubs? So Well, at some point you’re going to be mean at some point you’re going to be hub bound. Microsoft definitely does not Microsoft loves hubs. Yeah, everything’s a hub now. love hubs. Yes. So all right, what’s the

56:26 love hubs. Yes. So all right, what’s the database hub? It’s a simply it is an agentic agentic control plane for the an organization’s entire data estate. really a single interface for managing, monitoring and trying to govern all the databases and whether they’re on premise or not. Databases that are included is Azure obviously Cosmos database for SQL, MySQL, SQL server, Microsoft fabric database. And what you have is a unified

56:56 database. And what you have is a unified inventory. So all discover all the databases across AI assisted management recommended actions for schema changes agentic operations so maybe there’s a where operation agent comes in. You probably got to run on an event house. Cross database search and then you can even connect with Microsoft purview for lineage and classification. So this is a all databases play.

57:28 I think this is like the way this is where they need to go. This is This is if you’re going to bring all the trans transactional data into fabric you’re going to have to have some way to manage it all. I think I think this is this is plain and simple. I can’t have seven different SQL databases roaming around different workspaces and a post grass and Cosmos DB and something with real time. It makes sense to pull this up together into the same spot. 1000% The only concern I have here is how do we do auto scale billing on this?

57:58 how do we do auto scale billing on this? This is my only concern right now is if I’m going to bring SQL into fabric I don’t necessarily want to pay CU pricing for that. I want to pay for exactly what I use. So this is where I’m I’m a bit more pro on in Azure let me give you by comparison, right Tommy? If I have a production load, if I have something that’s running, I don’t I do care about optimizing it. I do care about it running efficiently but I don’t want a throttled threshold

58:28 but I don’t want a throttled threshold from fabric shutting down my SQL database or making things fail. Does that make sense? Yes. I don’t I don’t want that. So how do we leverage the database hub in a way or how do we leverage this in a way that lets me auto scale bill just for those items? And I I keep going back to like the example that Spark has been. Spark is amazing inside fabric and auto scale billing for Spark is awesome because

58:58 billing for Spark is awesome because whatever I pay in fabric the equivalent pricing is automatically applied back to Azure if I want to go to auto scale Spark and I only pay for what I use. I use. That’s amazing. Like They’ve set they’ve set the bar of like how I want to have this run. I want to deploy in fabric. I want it to be managed in fabric but I want it to be billed through an Azure billing cycle not attached to compute units. That’s that’s what I want. I don’t I don’t know if we’re going to ever get if

59:28 don’t know if we’re going to ever get if we’re ever really going to get a true Yeah, I’m pretty sure right now that they’re don’t ask about pricing yet. So Yes, there hasn’t been a lot of unpacking what that looks like. I I will say this though Tommy. I’ve been putting some fabric sequel databases in my F2 and it it and it it the the way the sequel database turns on and it’s utilizing it’s immediately grabbing like all the the V cores all the time and immediately maxing out my back end processing on an F2. It it’s

59:58 back end processing on an F2. It it’s not designed it’s not designed to go small enough to fit on an F2. It needs to be on something bigger than that size which Oh, that’s weird. If I’m if I’m doing a very lightweight thing like I want to turn on the database load some parameters run a meta driven pipeline and then turn it off. I should be able to do that. It it shouldn’t be this really expensive like spike every time I want to turn it on. So I there needs to be better scaling there. So I think hyper scaling for fabric sequel will probably help us out a little bit there.

60:29 probably help us out a little bit there. I think you definitely spot on that. So awesome dude. Well, I think we’re going to be talking more about the database hub. again still in the early stages just a major announcement. There are still I believe if more than 20 things we haven’t talked about yet. Oh, there’s a lot there’s a lot left. Yeah, there’s a ton left. So for example just a few things I’m going to just quickly highlight and we’ll just run

61:00 quickly highlight and we’ll just run through because again we’ll probably have episodes for it in the future coming future. So what we also have is the obviously the fabric remote MCP composite semantic models off of direct lake. That’s crazy. fabric runtime 2. 0 data flows and data factory gen flow ones or gen data flows gen one going away. migration assistance coming out more and more.

61:31 we have obviously graph and Microsoft fabric. fabric. What Custom sequel pool Yeah, what? What data flows gen one is is being deprecated? Yeah, there’s a blog article on it. Where’s that one at? on the Microsoft blog. I’ll have to go find that one. I I did not know that one’s going to be that one’s being deprecated. That that’ll be interesting to if they’re actually trying to get rid of it. What are they going to replace it with? I feel like there’s a lot of I feel there’s a lot of people on well, not a lot. I feel there’s a good amount of people on data flows gen one. That’s going to be a quite a migration bumpy

62:02 going to be a quite a migration bumpy path potentially. they they haven’t supported it in a long time. I will say this. They’re not making any new features. But did in the blog post Tommy did they announce when it’s stopping when it’s end of life is happening? I’m sure they always did., like listen this isn’t metric still actually giving it its its due. So [laughter] So let me see if I can find that. Yeah. Okay. Okay. We’ll go after that one a little bit. All right, anyways. That being said, there’s a lot of things

62:32 That being said, there’s a lot of things to unpack. I’m still excited. Everything feels like it’s going towards the agentic side. I feel like we are finally getting to a place where the creator experience with fabric is being developed correctly and I feel like we this this whole fab con for me really moved away from doesn’t work we’re missing features to oh, we’re getting new features. We’re getting interesting stuff now that’s going to actually really help me. I think we’ve really turned a corner here and and and shifted the the culture

63:03 and and and shifted the the culture the vibe the the feeling of the conference was there’s a lot of net new things coming. It’s exciting to be developing inside fabric these days. So very excited to see where they’re going to go. there’s a lot of new experiences that are coming. I’m very excited to see these, new features coming out in particularly custom totals on tables. Like that’s a winner. let me give you your due here because on the blog it’s a it’s this is like the save the date that we kids do with their weddings now. Sure. So we’re like when are you

63:33 now. Sure. So we’re like when are you getting married? We don’t know. It’s like cool. So existing gen one data flows will continue to work for the foreseeable future. However However, retirement dates are being finalized. So that’s where we’re at. Okay, so so nothing’s been officially announced. It’s starting to be talked about. Right. Yeah, exactly. Oh boy. Interesting. Oh well, yeah, I get it. I I don’t know we’ll see. That’ll be interesting to see if they can actually get people off of data flows gen one. That’ll be a hard move I think. I think there’s a lot of people using it.

64:04 there’s a lot of people using it. A lot of people do rely on it. And that pushes you into fabric immediately. People are going to want to be like well, let’s just turn on fabric for everything immediately. We’ve already got power BI that’s working for us. We don’t want to stay there. I don’t know. This is interesting Tommy. Hm. Okay, excellent. Well, that being said, thank you very much for listening to podcast. Hope you found some value from this. Hope you were enjoying the news and the news articles. There’s a ton of news from the fabric articles. Check us out in the description the full list of all the features that have come out have been announced. The blog post

64:35 out have been announced. The blog post in March of all the announcements is also in the description below. So if you want to check out these features in your own, please we encourage you to go down and look at the description. Go check those out yourself and you’ll you’ll find out a lot of great features and this is all in the March fabric release summary. I’ll also put that in the chat window as well for those who want to see it there. it there. All right, Tommy. Where else can you find the podcast? All right, I’m flying blind here. You can find us in Apple, Spotify, wherever your podcast. Make sure to subscribe and leave a rating. It helps us out a ton. Do you have a question idea or topic that you want

65:07 question idea or topic that you want us to talk about in a future episode? Head over to powerbi. tips/podcast. Leave your name and a great question. And finally join us live every Tuesday and Thursday a. m. Central on all powerbi. tips social media channels. I should know this by heart after 5 years. Right Tommy. You’ve said this like hundreds of times now. Anyways, thank you all so much and we’ll see you next time. time. Explicit Explicit measures just pump it up be hot. [music] Tommy and Mike lighting up the sky. Dance to the data laughs in the mix. Fabric and AI get your fears. Explicit

65:38 Fabric and AI get your fears. Explicit measures measures drop the beat now. Hub and Kings feel [music] the crowd. Explicit measures Explicit [music] measures Drop it.

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.