Less Guessing? More Building! - Ep.525 - Power BI tips

In Episode 525 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

0:01 Dance to the [music] day to laugh in the mix. Fabric and A. I get your fix. Explicit measures. Drop the beat now. H feel the crowd. Explicit [music] measures. Explicit measures. Drop it loud. All right, Tommy. Welcome back to the show. We’re back here with another episode of Explicit Measures Podcast. Tommy, how you doing?

0:31 Tommy, how you doing? I’m good, man. Am am I next to There we go. That’s what I’m looking for. So, how you doing, my friend? Oh, boy. I’m doing well. We’re doing things are quite busy these days. We’ve got a lot of things going on. we have a really cool episode today for you. It’s called Less Guessing, More Building. I think you’re going to have a lot of fun with this episode. It’s going to be really very interesting. I think you’re going to enjoy the topic for today. But before we get into our main topic around less guessing and more building, Tommy, what news do you have for us?

1:02 Tommy, what news do you have for us? So, we have some cool ones, man. We have fabric that’s come out with two features that very different that I think you’re going to enjoy. So, the first one we’re going to be talking about today and this just by the way, did you notice the blog changed?, it’s now on the Yes. Yeah, I saw that there there’s a the Microsoft community site is now carrying the Microsoft blog these days. So now it’s a little bit a little of a different format now which broke all my RSS feed readers and I had to go that’s what happened to me too. Yeah, exactly. [laughter] I had to go update them all so I could

1:33 I had to go update them all so I could see what’s going on in the news. but yeah, that’s all now different. So if you go to the the old URL links would work. So you go to blog. fabric. com or sorry blog. parb. com it goes to the new site and if you go to blog. fabric. microsoft. com it now goes to the correct blog site. So those two now are new website links. Yeah. So we’ll get we’ll get used to that. We’ll get used to that. But anyway, so we have this great article that just came out is around discover

2:03 that just came out is around discover items across workspaces using one lake catalog. But the API MCP and CLI tool. So I can go through one lake catalog now and without actually even going to fabric. com. I can do this through REST APIs which is fine but then I think the real selling point here is the MCPS because the s the search supports free text matching display names or descriptions which we’ve talked about descriptions the importance there so

2:34 descriptions the importance there so it’s a lot easier to find items even when you’re not sure what the exact name is the everything’s going to be filtered by the item type metadata filters and what’s nice is because of the API you can build your own catalog experiences into your own apps So you can have a full catalog of things the way that you may want it for your own organization. But again, selling point here is the agentic AI integration here because it’s built actually already into the micro fabric core MCV server. Yep. So AI agents can locate fabric assets as

3:05 So AI agents can locate fabric assets as part of a broader workflow. they have already some prompts out there to look to,, lay related to the sales team, identifying workspaces with certain keywords and then obviously there’s a CLI tool. So, a ton of things here. here. I think the main thing they’re trying to highlight here is the search feature of this one. So, there’s a catalog search core API and they’re talking about this one being like this is this is really the new thing, right? So, now you can go scan for things. it says here exactly this the fabric CLI now supports a find

3:37 this the fabric CLI now supports a find command command which will go across workspace discovery right from the terminal. So when you’re using a CLI like a a GitHub copilot or you’re using the MCP server, that find command should be able to give us some more strength around looking for different items or assets directly inside fabric, which makes a lot of sense to go look for things. So pretty cool, dude. So I it’s one of those things where again a user can get started right now because they can use it with the MCP. Awesome. But I think there’s a the oneway catalog is a great

4:08 there’s a the oneway catalog is a great resource. But again, is that somewhere in the UI right now that I’m actively bringing people and I think we’re all have this holy grail idea of eventually having a center of excellence like experience with the one lake catalog but it’s got to be much more refined and it’s going to be interesting what the API can do because obviously the MCP is not something you’re necessarily going to roll out to everyone. everyone. Yeah, but the MCP is not something you roll off everyone tummy, but you build on top of it. The MCP is like a

4:39 on top of it. The MCP is like a like a tooling element of something you may want to build. So, this is where,, I’m seeing a lot of this right now where some teenager in some house somewhere across the US is just building out an incredible replacement of a piece of software application and just really disrupting these large players in this market because they can just vibe code their own way into this one. I I was listening to a a YouTube short the other day and they’re talking about what’s the new business model going to market here with apps these days. The the advantage is not on the

5:10 days. The the advantage is not on the big the big existing software companies. The the advantage isn’t, you companies. The the advantage isn’t,, hey, I’m Salesforce and I can wrap know, hey, I’m Salesforce and I can wrap a a headless API layer on top of me. I a a headless API layer on top of me., yeah, maybe there’s some advantage mean, yeah, maybe there’s some advantage there, but like the small guys are nimble. the companies that are building ground up agent first from scratch, those are the companies that are really going to get an advantage here. So I think this this is another one of these opportunities where you have the capability now you actually have right to go build what you want. If you have

5:41 to go build what you want. If you have this better center of excellence idea in your head, Tommy, you just got to dream it up and now you can talk to it and and then the MCP just becomes part of that that you use to help you reason about data. agent first is going to be the way to build things moving forward, I think. So,, you’re not talking anything new in this on this article. Do you see yourself using this right now or just testing out with the MCP? I’ll probably test it out. I don’t really know if I have a complete application that needs this at this point.,, more of where I’m

6:12 point.,, more of where I’m feeling right now, Tommy, is the need is,, this this universal ontology of the organization is is kind ontology of the organization is is where I see the need right now. For of where I see the need right now. For me, I really want to be able to take all the tables, measures, and definitions from many semantic models and mash it all together. It really needs to be I really need a central location where I can put all that. That’s becoming increasingly important for me. I’m having real conversations with customers right now talking about that exact thing. And for us to be able to programmatically generate a timle or a

6:43 programmatically generate a timle or a model off of this unified whatever it is, semantics model would be really helpful, I think. And I think a lot of organizations will like that and we’ll we’ll actually have to figure out how to make that easier to to to build and create. So anyways, I don’t think the tooling exists there yet. That’s what that’s where I’m putting my eggs in that basket right now. now. That that’s where I’m at as as well. But again, they’re Microsoft is doing at the very least providing you with all the framework to do to build it out. This is there’s been a tipping point for me, Tommy. Ever since Fabcon, I was a

7:14 me, Tommy. Ever since Fabcon, I was a little bit more like h there’s so many great other tools that are not Microsoft that are being built right now with agents and agents attached to them and nothing integrates together with Microsoft. It just felt like everything was just very cobbled together. Since March, everything is starting be starting to feel more integrated. Mhm. Mhm. One thing I haven’t been able to test, I got to be blunt purely blunt around this one, Tommy, is I’m very excited about the idea. I just haven’t actually built anything with it. Not yet. is this idea of foundry hosted agents.

7:44 agents. Right? So I have right now a hosted agent on a computer on my desk over here that’s not my main computer. It’s a side computer. computer. But I think Microsoft is trying to provide you the same ability to host that agent inside the cloud inside Foundry. It has a file system. It has all these things you can tag like for example this solution, right? I can tag on to it the MCP server for file search and then I can go talk to it and it can go search for things, pull down definitions. Maybe I can ask it to pull

8:15 definitions. Maybe I can ask it to pull down three or four things. It can pull down the definition of multiple items and then write the files to disk, do some work with them and then push back what I want. Like this is this is what I’ve been waiting for. I’ve been waiting for Microsoft to catch up here a little bit and it feels like we’re finally gaining some momentum or at least I’m seeing the news about what they’re building and I’m starting to put the pieces together and thinking, okay, I think we’re finally getting on the right step. I just read another blog post by a VP of Microsoft. I think they’re calling it claw pilot is what they were

8:46 calling it claw pilot is what they were saying. saying. It was basically taking openclaw the agent and shoving it into copilot. So the same thing you’d have open claw on your local machine the same personality file system that structure that same element was allowed to be able to be pushed into co-pilot. I don’t know where the preview is. I only been reading about it on LinkedIn from this gentleman but it looks really interesting and I don’t know if anyone’s actually in the in the preview or the pilot. So I’d love to unpack that and see if there’s anything else going on. going on. So you’re hoping that person’s listening so they’ll send you that invite link.

9:17 so they’ll send you that invite link. Yeah, I’m hoping. Yeah. Yeah. Hey, Microsoft, can I get into into what is it called? Claw pilot. I don’t know. I don’t know what they’re calling it right now., but I’ I’d love to test out this new feature that’s coming out. I’d love to really see that one. All right. Awesome. Awesome. Well, I think we’ve burned through enough time of news and announcements. How about I just move into our main topic, Tom? What do you say? Let’s do it. Okay. So, today we’ve got our main topic. topic. Well, I’ll hear that. Tommy, is that a call? I believe we have a guest today. So,

9:47 I believe we have a guest today. So, we’d love to bring on a guest today., the one, the only, the most amazing, Alex is with us today. Hello and welcome, sir. I love that phone call ring. It’s beautiful ringtone. We should be able to customize that on future streams here. Let’s do that. It must be like a Motorola one. I think it’s [laughter] one of the really classic. classic. It bring back our memories in the early days. days. Yeah. Yeah. I’m in my flip phone, my LG 250 ringtone. So, this brings us to our main topic today. Alex Powers is

10:18 our main topic today. Alex Powers is here today. He has been cooking and you can see by his background on if you’re watching live, he has been cooking some AI stuff., it’s fine. It’s fine. Alex has got it under control. He’s building a ton of AI. Really cool things. Alex, Alex, welcome to the show. Thank you. What have you been up to? Let’s get into it. it. Tokens, man. Everything is tokens right

10:38 Tokens, man. Everything is tokens right now. now. Tokens. at least until the GitHub co-pilot shift where it’s like ah they’re tokenbased billing and then it’s like [laughter] I need to go back and like revisit some of my systems and be like all right let’s let’s slim that one down a little bit. well that’s been that’s been the news of the hour recently. GitHub’s going co token based. Everyone’s doing tokens. It’s all usage based pricing now. Yeah. And I have a subscription outside of work too. So it’s like hey personal projects things that I want to own when I leave Microsoft. It’s like absolutely like like Yeah. subscribe. I think right now

11:10 Yeah. subscribe. I think right now though it’s a very interesting time. We hit that inflection point too of like ideas are starting to turn into like production. and I think the the new frontier now is like cool you got it to 80%. Can you get it to like 95 or 100? 100? Yeah. Yeah. And that’s really the gap that I’m seeing people like either they build it in a weekend and abandon it or it’s like hey I’m going to keep investing in it. And I think especially for like the topic today, the point I want to hit home on is like the key word you use was central.

11:40 the key word you use was central. We need central tooling and especially as a community, we need like a central repository for the things that we’re doing. so especially as we like talk about the task flows. I consider it like the spine like there’s a brain and it was like here’s the thing that holds us up, keeps the body moving, the arms, the legs, everything. So, how do I kind the legs, everything. So, how do I drum up some interest within the of drum up some interest within the community to invest in the central repository? Not necessarily the thing that I built, but the thing that’s behind it. Yes. Yes. Interesting. Talk about that.

12:10 Interesting. Talk about that. Well, Alex, that’s a great point you bring that up. Actually, we are really we’re trying to cook something that’s going to help with that. Exactly. As you describe, we we have a thing called collections that we’ve been noodling on. We’ve been slowly in the behind the scenes working on a workload that’s going to do exactly that thing. we’re looking at right now notebooks, UDFs, templates that you’re going to have from PowerBI files and like there’s a whole bunch of other things that need to be added to this central like something that’s community based like so people contribute to the community but also looking at this lens of to the spine

12:40 looking at this lens of to the spine that you’re you’re speaking of. I I really resonate with this story cuz I’m I’m actively trying to build it is we need the spine as both community and for your organization like people that create in your company where you collect the things for sure. for sure. What do you think Tommy? What do you think about the spine concept? honestly there’s that integration but the problem I’ve always had with that or at least there’s still that barrier though of though of I can start building something but then when you actually want to get it deployed right like and I think this is

13:11 deployed right like and I think this is the big I always think that way it’s great when I have these tools for my company that I’ve in my company but guess what I have to deploy to one person or at the very most with you but I think the bigger thing is as we start building and I’ve been trying to be more conscious of this especially because how easy it is to have that vibe coding experience is really trying to be actively conscious of what I’m building to see whether it’s going to even help my own task flow. this actually is it something that can scale and I think that is the question right now we still have help with a lot

13:42 right now we still have help with a lot of these agentic tooling things that people are building with that I think Alex maybe you should introduce one of the projects that you’ve I think you’ve gotten it pushed to production it’s out it’s like it’s like the real deal you’ve gotten it past the 80% you got it to like what 90 95% you’re probably still tweaking it here and there and adding some things to it but like what’s the project you were recently working on and give me just got a got a like let’s land the idea here. We actually do have an intro video for it for that Alex wrote that we’re gonna

14:12 for that Alex wrote that we’re gonna we’re going to go through here as well. But Alex, give us give us land the idea here. What what is this thing called? What are we getting into? Yes. So, we’re going to go back in time a little bit to when task flows within Microsoft Fabric were announced. The community was in in a fervor. They were in love with task flows like this is the coolest thing in the world. And they’re like well what does it do? And we’re like well it doesn’t really do anything. it’s a diagram in your workspace and then everyone was annoyed and you’re like well why doesn’t it do stuff why doesn’t it deploy my items why doesn’t it have like the cross workspace lineage

14:43 have like the cross workspace lineage all these other cool things it’s like a lot of the the original ideas around task was like I still believe in so especially here within our project it came from a place of and this is more of like an AI thing for for people listening hey there’s these amazing things I just want to deploy items in fabric easily why can’t I do that., I went out and I took screenshots of the task flows. I threw those into a folder and I said, “Hey AI, go look at these images.”

15:13 “Hey AI, go look at these images.” Converted into,, the markdown, converted into the item dependencies, and it went out and built that for me. So, I don’t want anyone to be like, you need to spend a lot of time and resources here in in the brain. No, take screenshots, dump it in a folder, make sense of this for me. And then from there, Mike, it then turns into a agent that you talk to and at the very end, it provides an architectural recommendation. So to look at the different patterns, look at your problem statement and say, “All right, you want to build an

15:43 “All right, you want to build an application, you want to have a chatbot at the very end. Cool. Let’s do like a SQL database and we’ll also use like a data agent.” so especially as the situations get more complex, it then starts building out the full architecture for you. here’s medallion, here’s like lambda, here’s all these different things. And of course, if there’s like a very custom scenario, it does have the ability to go out and build its own recommendation. So, it’s not limited to like the 13 task flows that exist. It says, “Cool, I’ll build everything to

16:13 says, “Cool, I’ll build everything to your specification.” Now, this might be scary for some people. people. Yeah. Yeah. In the sense of like, well, I’m a consultant. I’m an architect. Like, that’s my job., I get paid to whiteboard and do all these things. [laughter], we’ll watch the trailer. We’ll talk about a little bit. Yeah. Yeah. And I’ll hit home on some of my key points and why I think that this should be a part of everyone’s projects moving forward. forward. Oh, I love it. Awesome. Excellent. All right. So, let’s roll the trailer., and then we’ll talk more about this in a second. For those of you who are you can listen to this. There’s there’s some words to it, but it’s going to be more of a graphical thing. So, this is a

16:44 of a graphical thing. So, this is a little bit different for our podcast than we normally do., but here we go. Rolling the trailer. More building. Sorry. Less guessing, more building. A messy whiteboard and a simple question.

17:16 What can we build with fabric? [music] But what if you didn’t need to know fabric, What if after a few prompts, the whole solution was mapped, items deployed, and everyone could see how the pieces connect. connect. So you can start building and stop [music] guessing.

17:53 With the fabric task flows assistant, you describe the problem and the fabric advisor walks you [music] from discovery to design to [music] a CI/CD ready deployment in a matter of minutes.

18:54 Okay., first off, why so serious? I know. Feels Bladeunner inspired. Yeah. Yeah. So good. It’s so good. When’s the movie coming out? Exactly. It feels just like a movie. It’s so good. Very, very in fabric flows. So, this is awesome. Give us the introduction. What’s going on here? What are we looking at? Yeah. the very beginning you sit down and you type in a problem statement. I don’t want you to type in fabric words. I don’t want you to know anything about fabric of like what’s a translitical, what version of UDF are we

19:24 translitical, what version of UDF are we talking about? Like lighthouse versus warehouse. Like I don’t care about any of those things. Yeah. Yeah. I work at a teleco company. We have a bunch of IoT devices. We’re trying to get better with our data. trying to make faster decisions from those like because we’re experts on this call like we can infer different pieces right faster all right so maybe real time I want to make better decisions all right possibly data science telco all right they’re having a lot of communications a lot of lines so this is like very huge in terms of like scale like we don’t necessarily

19:55 of like scale like we don’t necessarily need to say well I need a warehouse to do xyz do xyz so I really wanted to start it from a place of you sit down with the customer or even you are a customer internally. Here’s my current department’s problem. It starts asking you questions. So, it could be a scenario of like, all right, cool. You said CRM. What type? Is it data versse? Is like Salesforce? These are the great things about LLMs is they can go through, they can ask questions, keep asking questions. So, getting that confidence score up and

20:26 getting that confidence score up and then at a certain gate, it’s like, cool, I looked at your four V’s., so what is the volume? what’s the size of your data what’s the velocity so is it batch is it real time what’s the variety so is it real time is it more of the CRM Excel spreadsheets etc and then my last one is always my favorite versatility which is how versatile is your team your team am I just going to say hey you should write a bunch of Python or pispark and you’re like well I just use low code no

20:56 you’re like well I just use low code no code tools like power query or I just use drag and drops like the powerbi then it starts mapping out based on all these scenarios, based on the skill sets, based on the volume. Here’s what I recommend that you go do. So, it’s giving you that gate check here., from the back end, I have a agent system and skills. So, the first

21:16 system and skills. So, the first one is just going to be,, discovery. So, sitting down with someone, ask them probing questions. Next one is going to be the architect and tester., so they both debate a little bit back and forth because is what you’re recommending actually humanly feasible within the system?, I look through the documentation, look through all the different capabilities, and at a certain point I say, “Cool, here’s the customer checkpoint. Here’s what I’m recommending. Is there anything that you would like me to update?” Go through three loops. If they reject, reject, reject,,

21:47 they reject, reject, reject,, we just start over. Otherwise, we’re ready to rock and roll., from there, we develop a fabric CI/CD. template and then you also just run a script and you’re deployed in fabric. So imagine you sit down with a customer 5 10 minutes, you go through discovery asking these probing questions and then you run the Python script at the very end you’re deployed. Yeah, Yeah, with all the items they’re empty items, but we’ll talk about that. I I think this is a good point. All

22:17 I I think this is a good point. All right. My initial reaction to this one is yeah like I love this and then also like okay love that it’s stitching a lot of this stuff together some of the hurdle of just fabric every time I step into fabric there’s always a question of there’s so many different things y y which one should I be picking which one should I be doing and that’s that’s also where Tommy and I’s knowledge come into this right also if if the agent is good at you stitching

22:48 if if the agent is good at you stitching these requirements together and be able to pump out the creation of those items where I think still to your to your point Alex earlier was like I think consultants are still around in this space. This is not like a like a absorption of like okay well now no consultant needs to be around this but this now means consultants or people can now build three or four separate architectures altogether and like now you can build into that that architecture to a point where like you get stuck and you’re like h that wasn’t that like the the idea of

23:18 h that wasn’t that like the the idea of throwaway throwaway work now is now it becomes more like a commodity at some level now once you have notebooks built and the the internals inside these objects that are inside your your pipeline or whatever you’re building here, it’s going to be more resistant to like quickly throw all that away and start over again, right? Once you have data moving into these lakehouses, you’re not going to just delete it and start over., but the idea of like having multiple prototypes running at the same time at the very beginning, that’s that’s the

23:48 the very beginning, that’s that’s the advantage of having agents build things for you. You don’t have to land on something immediately. You can test out a couple designs very rapidly. I love this. Tommy, what’s your reaction? No, my first reaction to was the message in the in the video for those who are audio. Again, apologize, but is you the video shows the user asking a question or basically request, but the thing that really struck out stuck out to me was my Alex in there had anything to do with fabric at all. Not at all. Or had it didn’t have hey like our data

24:19 Or had it didn’t have hey like our data is in this thing,, or even remotely technical at all. It was literally just a user like, “Hey, our business has actually grown through expansion. Our data hasn’t kept up.” That’s the only time you really mention the data in spreadsheets. And he said, “We’re really trying to do this. We need to break this,, break it out. Where where do we go?” And everything is built from that. And to me, that is the part here that blew me away. I’m like, “Huh?”, so a user could a CEO could go and

24:49 user could a CEO could go and can just go or a team member go man our products are we’re trying to manage them across places everyone has this problem trying to update products okay that’s all you have to ask and it was able to build out this solid architecture obviously I love the part that it just deploys all that too rather than just the recommendation so but to me it was that first part of even if the tool just provided you with the framework that would be enough to me but

25:19 framework that would be enough to me but the fact that it deploys it too. So I I’m really those are where my two places are also I I’ll pause there like you see that question that was in the video. video. What sticks out to you there or do you have any concerns about that type of question that is so easy now to ask that and get an architecture back? Oh, so I’m going to I’m going to re like this is not in many ways to me a barrier for things if nothing else like Okay, let’s I’m going to put my consulting hat on

25:49 I’m going to put my consulting hat on here for a little bit, right? Yeah. Yeah. Why don’t I play with this tool? Why don’t I talk to it and even mock up here’s seven eight different businesses that I my company works in, right? Why wouldn’t I use this tool to already build the deployment scripts and have them just on the shelf like ready to go? I can tweak them if I need to. I can refine them. Like the whole idea of this is like it’s a first pass at something, right? It’s not going to have exactly what the business user needs, right? What h this is just

26:20 right? What h this is just infrastructure deployment. It’s not going to know what to go in the notebook. It’s not going to do an incremental refresh inside the notebook part, right?, there’s other things on top of this that you should probably use agent experiences to even do that next. So, this is only step one of I think multiple steps with an engagement with a client. And if you can go and say, look, I’ve got all these automatic deployment scripts and by the end of a meeting, you’ve got a representative. Here’s an architecture. We were just talking, I was just having something go here. And you, again, you don’t have to have it,

26:50 you, again, you don’t have to have it, you don’t have to go through the agent with the customer and show them it’s there, but you could have eight or nine different architectures that you like. describe what you want to it like I have patterns that I use. Tommy, why don’t we go through and say talk to the agent and have it build it and have the deployment script. So,, here’s pattern one. Here’s pattern two here. One of the big things I disliked about task flows in general was it was always closed. I think I think just recently, Alex, we’re now just getting to the point where the fabric

27:20 where the fabric you have import export now. So, like that was the one piece we never really had before. And on the right hand side there’s all these like templates and things that exist there. And this was pre- agent era right I wanted to have my template. I wanted to go in I wanted to build something. I wanted to export my template and just have it on the shelf and that way when I needed to I could put on a GitHub page. I could build some automatic deployment something. This just does all that. So now in this era of agents,

27:51 now in this era of agents, Microsoft I think and and Alex I think you would agree here is moving way more towards everything with an API. Everything agentic everything can be talked to with like some you talked to with like some agent behind the scenes. And what know agent behind the scenes. And what that does is you now don’t need templates for anything. You don’t need to have a library of this stuff. Microsoft doesn’t need to maintain this stuff. You just build the APIs and then we the creators the community can decide what’s the right pattern and how to build these things. So this is what this is exactly what this is ripe for. We need something like this tool. Use this

28:22 need something like this tool. Use this to build out your first architecture of different elements inside fabric and then it’s up to us to hydrate what stuff’s in them. Alex, over to you. What do what do you think? Is this are we hitting the right notes here of what you’re thinking about? The banner statement is fabric is hard. Yeah. Yeah. So a lot of people get stuck there as opposed to like well like ignore fabric. Like please stop. Like what is your business problem? Cool. let’s start from there. so especially as you start entering the business problem going through like that back and forth loop

28:52 through like that back and forth loop and then it’s like hey my team does TSQL. TSQL. so back in the the back end here what I was talking about the central repository is a JSON file that has all the items has all the metadata about them. so let’s start with a public preview versus GA versus GA that’s a field in the JSON file. Mhm. My company can only use general availability. Cool. Cool. The Python script is going to rip through that JSON file and start eliminating items that say, “All right, this is public preview. I can’t use

29:22 this is public preview. I can’t use that.” that.” Yep. Yep. My team is TSQL. All right. We’re going to go through all the different item types that support TSQL. TSQL. I’m now going to look at lakeouses, warehouses. I’ve now eliminated event houses., you houses.,, Python notebooks. know, Python notebooks.,, Python notebooks are an item type. I’m starting from the data storage area. Oh, okay. Sorry. Yes. Okay. Now that I’m at storage, now I can move up a layer to say, “All right, cool. For the warehouse, I support notebooks.” Okay? Okay? Or because I’m in lakehouse, I support notebooks and all these other things.

29:52 notebooks and all these other things. so it’s really building that item dependency map to say based on all these things, here’s all the items I’ve eliminated based on your skill set. I’ve got low code LLC or I’ve got code for CF codes. So I’m now just starting to build that architecture as we go. the part for me is that this is built around fabric CI/CD. So the Python package package at the very end you get a full architectural design review document that you and the customer could then sit there and read together. Here are all

30:23 there and read together. Here are all the trade-offs. Here are all the decisions that were made. here’s why we chose this based on all the different factors. And then you have the YAML file and you have the Python script. Mhm. So, at the very end of an a let’s say a 30 minute, 60-minute engagement with a customer, you rightclick that project folder, you zip it up, and you send it to them and go, “Cool, this is what we’re building in your environment. Deploy it.” They see it right then and there in their side. The hurdle of fabric is hard is now eliminated in the sense of here’s the problem. Here’s how

30:55 sense of here’s the problem. Here’s how I think we should go design it. Let’s start filling in these items. Mhm. The the point that I’m crowing at with this project and this why I’m like waiting for the community and doing these events to like drum up interest. What’s a copy job? It’s a connection string and it’s a list of tables. Mhm. Mhm. There’s no reason I can’t take this project further to say, well, here’s the copy job file. I don’t want to feed in those values to the to the the LMS itself, but I’m happy to have those in a file format I can parse through and then start building

31:25 parse through and then start building that dynamically., so why couldn’t I build a full end to end solution with a customer in 30 minutes? This is where we’re going. You got to charge them for an hour though. though., project based deliveries, right? [laughter] No, it’s fine. 100 I dude, I’m completely with you on that. So, Alex, you you want you talk about the community here and one of the things that also stuck out to me because my head’s spinning now on the way that to your point where it can be expanded

31:54 to your point where it can be expanded or that like with my own workflow here, I noticed that it’s co-pilot the CLL for copilot only. And I was looking through the directory here and it’s like, okay, it’s it’s very similar to a claude plugin for cloud for sure. That’s all it is. Yeah. or if if anything the same. So because I think especially if you want to think about the interaction and I love using the terminal on the terminal all the time but there’s a lot of use cases here from a

32:25 there’s a lot of use cases here from a simplified point of view where why couldn’t this be in cloud where it’s already plugged in with the plugin and also connected to your other resources. Again you’re starting right now with a simple question it from having all the context being provided to that how much can I do there if I wanted to give the the tool right now like the context of here’s our current environment these are some of the example tables or if I had I just give it a data dump of a schema from

32:56 it a data dump of a schema from a database right are all those things that would be taken into account right now or is that still something that would have be not right now but if start doing it. I’m happy to start building it. Yeah. Yeah. I’ll tell you my my break point was I got done shipping this product and then I had to run to FabCon SQL. [laughter] So it was basically like get this thing done, run there. so now that I like it’s been a little bit of time, some people have started picking up on it from doing talks like these. It’s like hey, do we want to take this thing further or do you just want it to be

33:26 further or do you just want it to be architectural? If we want to build a full end to end, like I’m happy to go that route. Yeah. Yeah. I just need to know the community is there. And then Tommy like back to my concept of a central registry. So registry is what I call it. If you look through like the folders that JSON file, that spine that I talked about, that’s the thing I want the community to band around. Cool. New item types were introduced in fabric. Let’s add it to the JSON. All our projects that are out here we’re building. This is how we get to deterministic

33:56 This is how we get to deterministic values. values. So within that, I say this supports an API. Here’s the endpoint type. here’s ABFSS, here’s HTTPS, all those things as central repository and then it’s just scripts that we build. That’s how we get there fast is we go together and the more that we can say here’s the community things that we all want to band around. I think that’s where would be very amazing moving forward especially in the fabric space., of course tied in with any other product but like for me right now

34:28 but like for me right now that’s the thing. It’s not necessarily this one bot because what I would like this to be Tommy I want this to be up in the service. This is Yeah, continue. Let me hear more about that. So the PMs that I’ve talked to, especially,, now that they’ve watched the awesome trailer, they’re like, “Hey, this is interesting. We’re curious about what we [laughter] can can to sell them on the the idea as well. Absolutely. That’s why Yeah. production value, man. It’s not necessarily the bot like that’s amazing to have, but it’s a creation engine.

34:58 engine. Mhm. Mhm. At the very end, I want the task to be creating all the items for you. Hey, you chatted with the bot in the side pane. Here’s what I think we should build. It starts building it out for you. And then your next inline agents have context which is the user built a Python notebook. The language I’m going to focus on is Python. The window it keeps getting smaller and smaller, right? Because now it knows the full architecture and knows the problems you want to solve. It can then start moving on. I need a data engineering skill. I need a data ingestion skill. that’s

35:29 need a data ingestion skill. that’s where like the complete end to end development here. I think it’s just going to get wild. And why couldn’t you sit with a customer, here’s the design, and then two or three days later, you’re up in production. It’s filling in tables. It’s filling in schemas. And [snorts] the patterns, this is also things where I’m like, again, Alex, this is something I’ve been harping on. The patterns are well known. We know what the patterns are. It’s either full load tables or incremental refreshing tables, right? It slightly changes maybe per source, but that doesn’t change.

35:59 per source, but that doesn’t change. Those are binary decisions right there. Yeah. Like per per table. Like, and this is where, to your point, this is where you throw your agent at something like this and go discover these things. Like, hey agent, here’s 150 tables that I’m going to pull from this other legacy enterprise data warehouse. Here’s here’s a readonly connection. Go read the table. Y, Y, go interrogate it. Make your best guess on can any one of these tables be incrementally loaded or full loaded. And so we can stop all this like grunt work around just like getting stuff started

36:32 around just like getting stuff started just to create the first layer of what’s going on here. A lot of that can be absolved away. And now it’s very when you step back on data engineering the patterns are really repeatable. The patterns are really simple. and sometimes I think we over complicate things too much. And now it’s just a matter of what to your point earlier, Alex, is where I think this is part of why you built this is what is let’s ma let’s marry up the tools that Microsoft provides with the skills on the team, right? We’re a heavy SQL shop. We’re a DBA team. That’s how we that’s how we

37:03 DBA team. That’s how we that’s how we built ourselves. Let’s heir on the side of those types of tools., are is our company or is our hiring going to allow us to move more towards notebooks? Is notebooks what we’re going to do? Okay. Well then then we can explore some other tooling. So these are the like larger decisions the central business intelligence team is probably making or designing with data engineering in mind and then from there a lot of stuff just shakes out. I feel like

37:33 like Tommy’s thinking yeah can I ask Tommy’s cooking an idea here. Yeah. So, let me actually I’m very curious, Mike, and I don’t know if you want to take it this way, but your own building experience here because I would love to highlight for people who listen., , just to context, Alex, we’ve been finally getting questions on the mailbag from people around AI and I think we’ve pushed it so much, but and they’re really the questions have been very much focused on to be very clear, it has not been us pushing it. It’s people like

38:03 pushing it. It’s people like we just we just rode this wave when it was really small and didn’t even look like a wave. We were like just we were just just Tommy and I were just paddling like crazy on the ocean by ourselves like there’s something out here. We got to go find it. And like the wave is starting to pick up and I think what’s happening now is the wave is now catching a lot more people and the the AI topic is now becoming much more front and center. So like I I we to be very clear Tommy we’re not pushing this. We were just maybe a bit early to this early to the the indirectly. Indirectly and I think people don’t too if they want to get mentioned on the podcast. All I have to

38:33 mentioned on the podcast. All I have to do is put an AI question out. We’re like we’re gonna put on the episode [laughter] very yeah it’s not very we’re not very difficult there. But a lot of questions have been around like hey building it what’s a good practice where’s your workflow. So, just quickly here, I don’t want to have bird, but walk me through your own building experience building this tool because I think that’d be kind this tool because I think that’d be neat. Yeah, of neat. Yeah, I want to explore that too as well. Yeah, I would love to hear what

39:03 Yeah, I would love to hear what that process was, where you started, what the refinement was like. because I think that will be good for a lot of listeners and I’m also very intrigued. Cool. Well, it starts with taking screenshots and throwing them into a file folder. Ah, okay. I don’t want people to overthink getting started like that’s the crux of the idea. when we actually get into like hey like what’s the agent system and all the setup I prefer a lot of mission control type of setup where it’s one agent that runs through the entire delegation process

39:33 through the entire delegation process because my thing is very linear. Hey, we go through discovery we then advance to the next stage. everything I try and do and build is through deterministic scripting. if I was to like hey this LLM itself it’s probably like 5% LM everything else is Python and running through the registry. so the very first thing is asking you questions based on the business process. That’s what LM is amazing at. Mh Mh is taking text making sense of it.

40:04 is taking text making sense of it. here’s the guardrails I have like the four V’s. I need to map those four V’s. Did I understand the volume? Is it small data? Is medium data? Is it big data? did I cross like velocity do I have an understanding is it batch is it real time variety do I understand the different systems is it a database CRM is it spreadsheets is it whatever else it may be and then like what’s the versatility so like once the LM is at that confidence level it then advances the next stage next stage is going to be the architect where it’s

40:35 going to be the architect where it’s going to take all that information look through that JSON file make sure everything checks out and then it’s also going to be the QA a tester that says based on this architect you that says based on this architect did they come up with a random know did they come up with a random crazy concoction or is this something we can feasibly build? Mhm. Mhm. All that is advancing through scripting. Did are those agents that you built yourself to do that design or are you using already what’s in the tooling? Oh no, these are all mine. Yeah. All right. All right. Yeah. So one agent the rest are skills.

41:06 Yeah. So one agent the rest are skills. Okay. Okay. So the rest is just literally running through advanced. So it’s a Python. and it has like d- advance and then it’ll pass through like the different copilot prompts., so this is like an advanced script where it’s just moving on its own. That’s where especially if you sit down and start testing this, it’s very fast. All the time is spent on discovery. So if you open So again, just what Alex is saying, I’m going to try to like materialize back into like what’s in what’s in the repo, where are these things, what you’re physically pointing

41:37 things, what you’re physically pointing to, right? If you go to the repo, the repo has agitub folder and and depends on what language you’re using. I use cloud folders just because that’s that’s the same. You can use GitHub. cloud. They all they both work. this one is more GitHub specific on some of them. And again to be very clear if you if you can adapt it, you can literally pull down the repo and say, “Hey, I want to adapt this GitHub to be more cloud after the call just to appease.” [laughter] Yeah. Exactly. But in here there’s a there’s an agent

42:08 But in here there’s a there’s an agent folder and in the agent folder is here’s the fabric advisor agent right so there’s an agent. mmd file and that’s what’s that’s what’s picking up that custom agent around all the things you’re describing here Alex that’s the orchestrator that’s like narrating grabbing moving things around and then underneath there you have like a bunch of skills so the skills there’s like six different skills deploy

42:32 like six different skills deploy design document heal and test So those areas are the skills that you’re employing by the central agent that’s just like walking through this progression together. That that’s how this tool works. One of the things I I want to just maybe pick on here too like I want to go a little bit further like love it love the tool that’s here. How did you build this tool? I’m sure you built this thing with agents. Like what are you doing? Like what’s what’s Alex creations process like? So you you came with this idea task flows.

43:02 you you came with this idea task flows. I want to do this. Want to want to absolve make some things simpler here. I want to leverage agents to help me get the business requirements out. Maybe we can talk a little bit around like okay how did you get to here? What what is your what does your VS Code look like? What agents did you build separately to get this completed? Oh my gosh. so especially like this is so meta. This is so meta right now. now. That’s what I’m This is what I’m intrigued too. Yeah. So early on the project it was a lot of guess work. you could think of like all the docs and like learn and like all

43:32 all the docs and like learn and like all these different things where I’m like I love learn I love web searches but my goodness at a certain point you’re just like crawling in your project. [snorts] so once I decided like hey like all these things you’ve been doing back and forth like why don’t we just create a JSON file JSON file and then that’s where you store all of your knowledge. So especially as we kind your knowledge. So especially as we think about like the item types like of think about like the item types like that’s just a complete registry now every item type that’s possible every made metadata attribute about that and then I moved from web searches to just

44:03 then I moved from web searches to just Python file execution hey Python file go read this file look for items that are general availability cool now my agents and that skill they have tooling so especially anyone that I talk to I’m like if you were to sit down with an employee like what what is the thing you’re going to put on their desk? You’re going to put on a laptop. You may put on a a notepad. You may put on a pen. Like what are the tools that they need to get their jobs done? so especially as I go through these, I’m thinking more and more about that.

44:35 thinking more and more about that. for a document, like their knowledge is very limited. Like they just care about filling in a scaffold template. Here’s a ADR template, architecture design review template. It needs to be filled out every time the exact same way. Here are the fields. the sections especially for like the architect like it’s limited tooling is reading that JSON file and making sure that it’s all concise. the QA tester like again it has a template.

45:05 QA tester like again it has a template. It has to do like I think AC mapping where it says yes this is architecturally complete. I can go ahead and do all these things. It’s pass my quality benchmark. everything I’m thinking about now is just this is my coworker. what tools do I need to put on their desk? And the mission control agent, this goes back to like the fun part where it’s just it’s it’s NASA like you’re communicating with a central place and then everything else that’s from there is just delegating. So because my processes are very linear. I

45:35 because my processes are very linear. I don’t have creativity here. What creativity do I need? Discovery, architect, put it in front of the customer, deploy. That’s a linear process., this is where I try and impart on people like a lot of this thing is automation. The 5% is LM. What do you need to build? What’s your problem? Cool. Now that I have all that, let’s go through automation. So, you spent and to be clear what I’m hearing you materialize here. You’re saying you spent a lot of a good amount

46:05 saying you spent a lot of a good amount of time a lot of time on the obs, you of time a lot of time on the obs,, going through the documentation. know, going through the documentation. So when I look at the item type registry JSON file. So so a lot of your a lot of your fumbling around or starting to this process like yeah I got to go to learn to go find a lot of this stuff here. I had to go scrape all this stuff down. And then I think so let me just unpack some of these things here too, right? When I when I go look at a specific item, right? You got like a display name. You have other things in here which are very useful for the business discovery thing like which say here’s an event house,

46:37 like which say here’s an event house, right? People may describe something in multiple different ways. They may call it event house. They may call it an eh. let me go down here. KQL database is another one. Right? you have these aliases that could be like you could say this could be a KQL database, it could be a KQLDB, it could be a custo database, it could be a custod like we’re all saying these are all language words that all mean the same thing. But that’s where like you’ve implied business knowledge into this definition of this particular item. So

47:10 definition of this particular item. So now that it could be just absorbed into this item type registry. And then there’s a whole bunch of other things in here like is this thing availability? What’s is it GA? And you have acronyms for that as well. So you say okay what are the different availability denotions GA private preview or public preview. Y Y so so looking at this but so if you go through this this makes a lot of sense. All you’re doing is you’re saying, “Here’s the thing, and here’s all the metadata that the thing needs in order to be knowledgeable to an agent to

47:42 to be knowledgeable to an agent to figure out how or what they would deploy or what they would need to know about deploying this one.” So, I’m going to go back to something that you guys were talking about earlier. earlier. Yeah. Yeah. I don’t think that Tommy and Mike’s value is in knowing when an item hit GA or in knowing,, a acronym. I think your expertise is knowing when to use the right thing or what question to you ask at the right time. I don’t think any of these other metadata attributes are important. So that’s where I’m like

48:12 So that’s where I’m like really really who cares what Tommy’s going to say, oh this item went GA on June 6. It was a a it was a summer day. it was 73 degrees out. Nobody cares about that. Like let’s get that out of your brain. There’s way more important things that we can store. that’s where like for me like the item registry is something I’m like oh my gosh this is amazing. it has like the endpoints. It has like is it deployable or not through REST API otherwise it’s you and me fumbling through the learn docs. Yep. Awesome. I know I you made it scan this right

48:43 I know I you made it scan this right beforehand. So yeah. yeah. Yeah. So as I was going through and I was hitting like the deployment issues I was like why is this thing such a bear? And I was trying to recall these rest API information. And I was like, can I just use fabric cicd? It’s like, yeah, LET’S DO THAT. WOW. WOW. AND THEN,, YEAH. From there, it’s very structured YAML file file and within the different REST APIs, there’s certain bodies., so like have those templates out there, Mike, like, hey, this thing requires a B 64 string. Here’s the different,

49:15 B 64 string. Here’s the different, attributes you need to hit for this type., some items do require that, some don’t., but now that I have it, again, we’re at that point in the project where I’m like, I know what it takes to fill those out. Do you want me to do a full end to end project deployment from sitting down to cool, I filled in that notebook?, one of my favorite scenarios is I work at a bakery. I want to visualize my square data in a column chart. What would you do for that?

49:47 What would you do for that? Square data, it’s an API, right? Yeah. Oh, you’re talking Square the the application like the the credit card company. Yeah. API bakery. I want to visualize my square data in a column chart. Right away I’m like right now right now everything everything I’m doing on that one is pipelines lakehouse and then figure out the rest of the downstream stuff. Right. So as quickly as I can get to a semantic model I’m if I’m again thinking a single bakery versus an a suite of bakeries. If I’m just the one shop that’s what I’m doing. I’m I’m trying to land the the data in the

50:17 trying to land the the data in the lakehouse, maybe clean it up a little bit, and then right from there, drop it over direct lake where we’re down into models, and we got charts showing up. So, like that the fastest path there is how I would build it., what would you do, Tommy, Mike?, I’m comfortable using Claude. Yeah, just [clears throat] ask it. No, how does that change your recommendation? Oh. Oh, I see what you’re saying. hm would you still use a pipeline or would you be comfortable going to a

50:47 or would you be comfortable going to a notebook and writing raw code? Yeah. Well, I I was even thinking like,, if I’m thinking about the square side of things, right, immediately saying Cloud’s available to me, not that I wouldn’t go this way anyways, but that might actually put me push me towards like, well, maybe I should have some like SQL database in this pipeline mix, right? is or something else that I’ve,,,, not that I’m going to use cloud to write stuff in in SQL, but again, I’m not as afraid to go to more complicated systems, pulling down APIs, right?

51:17 right? I I really do still feel like,, API pulls of things., I don’t really love APIs going through notebooks directly, just like for me like a stability standpoint. I’ve gotten burned by that a couple times. [laughter] Like you can do it like it will work. You can go straight notebook and call the API right from the notebook, but I actually do that in some other situations. I do a lot of like RSS feed grabbing and things like that via notebooks and scheduling things. a little harder to debug, but to your point, point, you could skip a whole pipeline and just get rid of it all together. Just have

51:48 get rid of it all together. Just have notebooks and just Lakehouse. Yeah, that’s but these are the that’s to your point that’s where Tommy and I would have the discussions with clients or how any contractor would or any even internal business users your your central BI team you’re still need to have those conversations with the team who and my question usually is how do you want to maintain it and this yeah go ahead and this is where I was leaning towards Mike you you’re is firing me up here and is really the next part of this too so again a user can go and maybe

52:19 too so again a user can go and maybe they have some experience maybe they don’t but it builds this pip pipeline it builds the architecture okay well again the maintaining this and revisions are the point where I think hopefully the expansion of this tool but I think very much to Mike’s point here is yeah this may be a working situation but without some experience or scale you without some experience or scale again what have you been burned on know again what have you been burned on what are your best practices or what are your preferred tooling because you may

52:49 your preferred tooling because you may have something deployed that no one at the company knows how to use, right? If it is notebooks, maybe everything is dead simple. I don’t know. I don’t think I saw any dataf flow gen 2 type resources in that GitHub repo, by the way. So, way. So, oh, there is. Okay. Okay. Just want to make sure. but I think the biggest thing here is

53:10 but I think the biggest thing here is okay, I’ve created this architecture, right? Now, what’s the next step from here? How do I maintain it? And then more importantly, what are the skills that the whoever is going to be managing this need and how is it managed? And I think this is such a big part here where when we really talk about agentic tools now for me, Alex, I’ve really come to this conclusion that I don’t think AI does not work successfully alone and when it’s

53:40 successfully alone and when it’s isolated. And you’re showing part of that here, but my big aha moment lately, I know we were talking about this offline, was part other tools right now that have really great connectors to other tooling that they can talk to., not just an API, but other a, agent to agent. So like for example, I have notion which is my brain for consulting and,, personal and everything else in between. and Claude copilot can actually read notion,

54:11 Claude copilot can actually read notion, read those instructions right back to it and it has all this other extra content context not just feeding in information then moving forward but this very much the sympathy or ah that’s a word I’ve always avoided symphony the orchestration let’s say orchestration so of things working together and I think when you’re looking at something like this which as amazing as I cannot wait to start trying it is how does that work in the context of a business that like with their own what are their resources and their skills what happens next and I think that’s

54:42 happens next and I think that’s where I’m going to start asking you in terms of when you look at this from a team point of view or more importantly when you look at what’s coming next what do you see or what do you foresee for people to really dive in [snorts] so I think right now it’s still a lot of just discovery amongst people like how are we going to use this stuff I think the part for me especially like looking at community skills like Kirby Buer is doing some wild stuff with like the PBR CLI.

55:12 everyone is running so fast that I think for some it’s just you need to take the long weekend and just go for it. it. And you’re going to stay up till 1 or 2 a. m. or maybe even later. I won’t admit. but like you’ll have that unlock moment. From there it’s also like coming back from that brink of like everything is in an agent., I I I think getting the low value things off your plate is highly important. So like, hey, I love talking with customers. I

55:42 hey, I love talking with customers. I love talking about their business problems. I hate searching for docs. I don’t think that’s a valuable use of my time. my time. So a project like this is amazing to be like like fabric is hard. We’ve gotten past that. Eliminating every fabric acronym, term, etc., etc. Here’s your business problem. Here’s what I think we should start building. building. Mhm. Wow, I can visualize it. I see it now out in a workspace. I understand the the flow of information because I have IoT and it ends up in the event house. With the event house, I have KKY set. I have sepic models, all

56:12 have KKY set. I have sepic models, all this other fun stuff. getting people past that initial hurdle. It’s less about fabric then. It’s more back to their business problem. Especially on a project like this, I would want it to be sync with GitHub, sync with ADO. All of your items are now version controlled. And then from there, like start filling in the items. Either do it through click ops,,, where you’re doing 200 clicks to get something done or just have an agent start filling in the different values. But from like an agentic development side for anyone

56:44 an agentic development side for anyone that’s out there, my goodness, it’s moving so fast. It’s really figuring out which parts you do or don’t want to take advantage of. I’ve got full notebooks that I use for FabCon SQL to deploy all of our different kind to deploy all of our different workshop environments. I didn’t write of workshop environments. I didn’t write a single lick of code. That’s this is the hook though. Like like Like like I’ve given up like yep yep my value is much higher in the sense of like hey I can describe the problems I want to solve. I can do that in greater detail through like planning modes

57:15 detail through like planning modes but like me writing my hundth thousandth loop in Python is that giving me any value? No. yeah, No. yeah, talking with the people who are running the workshops helped them improve it more because I’ve gotten things off my plate. Yeah. Oh my gosh. Like I know I was helping you guys in the back end too of just like what can I do? How can I help? A lot of that was because I wasn’t writing writing code code notebook code. Yeah. Yeah. Yeah. And and I think this is the we got to end here. Tommy, you got to ditch here. So we’ll have a

57:46 to ditch here. So we’ll have a new outro here. But this has been a great topic. I I re so to be honest everyone like for those of you in chat I hope you’re checking out this project for the first time. lots of good things here. I was really pushing hard to get Alex on here for two episodes. So we may have to get you back Alex going to keep unpacking this a little bit more because I you said one episode I’m like oh man I’m disappointed because I want I want to do how I build it. I want to do a deeper dive and be with you. We got to do that one as well. So with that being said I just want to say Alex, thank you so much for the

58:16 say Alex, thank you so much for the time. This tool is awesome. Really incredible. I really want to highlight this a bit more of like the the point of this is it’s a community development project. Alex has got it started. There’s a really good place to begin here. There’s already some great questions. John Kursky, I see your question there around like does it handle capacity management? What are considerations for capacity and volume? How much? How often? Like there’s some stuff in there. There’s a lot of words about capacity. Literally did a search on the repo. So there’s parts of this that are addressing capacity. Even pointing to learn documents. Here’s some guardrails for capacity requirements for

58:47 guardrails for capacity requirements for the direct lake overview, right? So, there are some capacity related data points in here. But this is where we’re pushing back to the community like or your custom like inject your knowledge as the consultant. Fork this sucker. Yeah. Yeah. Go build your here’s a whole section around capacity management and figure out what you want to do with the story run because it it’s in your head. It just needs to get programmed out into something that the the the AI can reason about on top of this as well. The registry, central registry. Oh, yeah.

59:17 Oh, yeah. Central registry. I want community to band around., with that being said, Tommy, where else can you find the podcast, dude? Well, thanks again, Alex. You can find us on Apple, Spotify, wherever your podcast, make sure to subscribe and leave a rating. It helps us out a ton. You have a question idea or a topic that you want us to talk about. Well, head over to powerbi. tmpodcast. Leave your name and a great question. And finally, join us live every Tuesday and Thursday, a. m. Central on all of PowerBI tips social media channels. All right, and with this ending, we’re going to do a little bit of a different

59:47 going to do a little bit of a different ending here as well. So,, I leave you with this, everyone. I’m going to give you a a very unique ending. It’s not our normal ending., but I give this in honor of Alex and his tool. We have custom written his own song for episode 252. [laughter] So, enjoy this song as our outro. It’s not our typical outro. Alex, thank you so much. We hope you all have had a great time today enjoying this conversation. Please let us know in the comments. Do you want to know more about the tool? Let’s do you want to see a real demo of it? Let us know in the comments like what you think about this. Please give us a thumbs up or if you

60:18 Please give us a thumbs up or if you like the video as well if you found this content valuable. we’d love to get Alex back on. And the more thumbs ups you give me, the more I can convince Alex to be here more often. So, thank you all so much. Appreciate it, guys. Here’s your outro.

61:00 In shadowed code where pipelines twist and stall and stall guessing configs watching servers fall. GitHub copilot steps into the fight. Fabric task flows rising in the night. [music] Less guessing, more building. Co-pilot forging. No more stalling. Fabric ready. CICD. Watch it flow. Less guessing, more building. Let’s go.

61:56 prompt the AI. It hammers out the scripts. Task flows for fabric. Seamless. Seamless shift. CI/CD pipeline steel and cinematic glow. Deploy with power. power. Let the data roll. No trial and error. No wasted midnight grind. Co-pilot builds it, leaves the doubt behind. Industrial [music] thunder, cinematic rise. Fabric task flows underneath.

62:38 [music] More building, co-pilot forging, no more stalling. Fabric ready CIC. Watch it flow. flow. Lessing, Lessing, more building. Let’s go.

63:30 Explicit measures pump it up [music] high. Tommy and Mike lighting up the sky. Dance to the day to laughs in the mix. [music] Fabric and A. I get your feels. Explicit measures. Drop the beat now. H feel the crowd. Explicit measures. [music] Explicit measures.

Thank You

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

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

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