Fabric Ideas That Stick, and What's Next - Ep. 509

Mike and Tommy tackle a mailbag question about submitting Fabric ideas, what makes a good idea stick, and where data engineering is headed next. Plus, Mike reveals he has 147 versions of Power BI Desktop downloaded.

Main Discussion

Power BI Desktop — 147 Versions and Counting

Mike kicks off the episode by revealing he has 147 versions of Power BI Desktop downloaded on his machine. The latest February 2026 build clocks in at 859 MB — nearly double the 553 MB from January 2025. The Windows Store version? 3.1 GB. Tommy and Mike reminisce about the early days when the Store version and the .exe had different bugs, like the infamous drag-and-drop issue in Power Query that persisted for years.

Pre-Conference Jitters

With FabCon Atlanta right around the corner, Mike warns listeners to expect some turbulence in the Power BI service. Microsoft typically holds back announcements before big conferences, and the last-minute service changes that support conference demos can sometimes break things temporarily.

The Evolution of Fabric Ideas

The heart of the episode comes from a mailbag question by Javon, who asks about submitting Fabric ideas and foreseeing the future of data engineering. Mike and Tommy trace the history — from the early days of ideas.powerbi.com when the product team actively built features based on community votes, to the current state at community.fabric.com where the sheer volume of ideas makes it harder to get traction. The early ideas portal genuinely shaped Power BI’s direction, but as the product matured, the focus shifted toward enterprise-grade features, CI/CD, and Fabric licensing migration paths.

What Makes an Idea Stick

Tommy and Mike discuss what separates ideas that get implemented from ones that languish. The key factors: clear articulation of the problem, broad applicability across customer segments, and alignment with Microsoft’s strategic direction. Ideas that solve enterprise pain points or reduce friction in existing workflows tend to get prioritized. One-off niche requests, no matter how many votes, often get deprioritized.

The Future of Data Engineering

Looking ahead, Mike and Tommy share their predictions beyond AI hype. They see new data types, better integration patterns, and more domain-specific applications of data engineering emerging. The conversation touches on how ODBC drivers for Fabric data engineering signal that Microsoft is filling connectivity gaps, while GraphQL represents a lighter, API-first approach that pairs well with agentic AI workflows.

Looking Forward

FabCon Atlanta is next week — the guys remind everyone to check it out. With Microsoft holding back announcements for the conference, expect a wave of new features and capabilities to drop soon. Keep submitting ideas to the Fabric Ideas forum, but be strategic about how you frame them.

Episode Transcript

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

0:02 Tommy and Mike lighting up the sky. Dance to the day to laugh in the mix. Fabric and A. I get your feels. Explicit measures. Drop the beat now. Pumpkins can’t steal the crowd. Explicit measures. Good morning and welcome back to the explicit measures podcast with Tommy and Mike. Sorry, there were some technical difficulties happening this morning, which why we’re a little late today. Anyways, we’re here. For those of you

0:33 Anyways, we’re here. For those of you who are listening recording, you’re like, there are no technical difficulties. It’s just working. Great. Thanks for listening after the fact. All right., main topic for today, let’s talk about the ideas area. So, ideas. powerbay. com or ideas. fabric. com. What ideas are out there? How do you submit an idea? What does that look like? And maybe what do Tommy and I see around what’s coming up next?, what happens next with data engineering? What changes do we see occurring in the

1:05 What changes do we see occurring in the in the in the space that you need to be aware of as we start predicting what’s going to happen moving forward? I will say this, Tommy, and this is going to lead into my little news item here this morning. Go ahead. Go ahead. This morning, I was talking with Tommy about getting things going and you we’re going up here next to Microsoft Fabric Conference. This is down in Atlanta. And so because of the fabric conference, it’s a big event. And when Tommy and I go on the podcast, we always check like the the blogs for PowerBI,

1:36 check like the the blogs for PowerBI, the blog for for Microsoft. And typically typically leading up to these big conferences, Microsoft announces nothing. There’s no news coming out. There’s almost no blogging because they’re holding back all the announcements and blogs and really large efforts. They want to make a splash. Totally understand. So, all this to say, I was talking to Tommy this morning. I said, ‘Oh, what what version of desktop are you on? And I start, well, why didn’t any she said,, what version are you on? I said and I started rattling off some numbers for the desktop version that I’m

2:07 numbers for the desktop version that I’m on, the February 2026 version. And Tommy was like, what? What are you talking about? And I’m like, oh, we had it. We have it here. This is this is the I’m we are we are seeing the death of desktop now. It’s becoming is that what it is? More evident now than ever because Tommy was not able to like pick up on what version desktop I was on and and catch again. It’s early in the morning. I’ll give you some slack, Tommy. It’s early

2:37 give you some slack, Tommy. It’s early in the morning,. Go ahead. And first off, Mike, you are also the person that has every version of PowerBI desktop. desktop. I do. Weird. Yeah. No, it’s actually very weird. Yes. So, of course, you have all the version numbers. I bet the folder names are not month, they’re the version number. Oh, I don’t even do folder. I just put them all into the same folder and then I just keep making it. It finds that there’s already a file there and then I just say, well, keep and keep both copies basically. And now I’m up to I

3:10 copies basically. And now I’m up to I I’ll tell you, Tommy, how many versions of desktop I have. And I’ll also admit the more recent versions of desktop are very expensive to hold on to because they’re much much larger. Thankfully, I’ve got a fairly sizable SharePoint drive I can go grab that from. So, we go here to my downloads. I’ll go down to the PowerBI folder. They’re each now like 900. It’s like a gigabyte download for each file. So, yeah. yeah., I just downloaded desktop version 147 on my machine. And on that version we are pushing

3:42 And on that version we are pushing megabytes of file for that. It’s just becoming becoming how much? how much? 859 859 megabytes is what the size of that file. It’s massive. And just a year ago, January of 2025, we were at 553 megabytes. So in only a year, we’ve almost doubled the size of the file. Just a little over a year. the problem like is the Do you ever download the Windows Store version,

4:13 ever download the Windows Store version, the Microsoft Store version or I also have that one installed as well? Yes. Yes. But you don’t have the version history, right? right? That I don’t check that one, honestly. That one I don’t really look at. I just let the store update itself. Honestly, Tommy, I I think I typically use the one I download the. exe file. That’s actually the one I use all the time. I don’t actually use the store version very much, but I do have it because it updates automatically and other fun things. So, speaking of things you’ve been doing for a long time, do you remember because

4:43 for a long time, do you remember because I’ve always had the store and the. exe file. Yes. Yes. Back in the day, they were not in sync in terms of the like some little bugs. For example, the Microsoft Store. Yes. Yes. You cannot drag and drop your power your queries in Power Query. I remember it was like that was probably for like the first four years and it was the most frustrating really. really. Yeah. Yeah. I think they’ve I think they’ve since modernized that now. So, I’m pretty sure the builds they’re putting out now are more consistent across both the store.

5:13 more consistent across both the store. But when you go to the store version, Tommy, I’m look at the store version here. The store version says it’s approximate size for the store version is 3. 1 gigabytes for the store version. Geez, that’s insane, dude. That is quite a bit also to help me understand like which version I’m on. I have like the store version in dark mode and I have the. exe version in light mode so I know which version of desktop I’m in. If I’m in dark mode, I’m in the one from the store. If I’m in light mode, I’m in the one from downloaded from the the

5:44 one from downloaded from the the. exe downloads area as well. So, I believe when I look at them now, I think they’re both on the same version. The one that I just downloaded, I’m just double checking here. 1182 and then checking my other one here. See what the version numbers are. Yeah, they’re the same. I’m on the same versions for both. Yeah, I will say this though, Tommy. Right now is the time everyone’s waiting because if anyone’s doing demos or getting to do publishing things

6:14 or getting to do publishing things directly inside or do demos inside the fabric conference that’s coming up next week. This is always very nerve-wracking because right now I see all the changes in this in in the service. The documentation that you were doing is slightly different. And what happens is some stuff typically breaks. little things get out the door that are just not quite right or maybe a bit rushed and it doesn’t quite work right. So this week, expect some turmoil in your PowerBI application either on the service or locally cuz usually little

6:46 service or locally cuz usually little things just go off the rails during this week. Oh yeah. Oh yeah. Anyways, all that to say, I’m happy that Tommy didn’t even know what version of desktop was on anymore. My memory and usually you tell me to close all my applications. Yeah. Before we do our thing, so we don’t have technical difficulties. So, 100%. So, now you’re blaming on me. You have PowerBI desktop open. Don’t you? you? Not anymore. I closed it because I don’t want anything I don’t want anything else

7:16 anything I don’t want anything else extra open to like take resources from my computer right now. I even closed Teams. I’m not even in Teams right now because I can’t I don’t want to Well, I always close Teams, too. I don’t want to risk,, having something freak out because Teams is a memory hog and and things like that as well. Awesome. Good things about news. All right, let’s jump into our main topic today. What do you say, Tommy? Does that sound good? Let’s do it. All right, main topic for today is another mailbag. We’ve been doing a lot of mailbag March things. This is an interesting topic. Let’s get into it. Tommy, read us off here. All right. Hi, Mike and Tommy. I

7:46 All right. Hi, Mike and Tommy. I absolutely loved your episode on living in a direct lake world, episode 504. Could you please talk about how to submit fabric ideas? What ideas are the best ones? How to foresee the future of data engineering? Maybe not focus just on AI but things like new data types. Interesting. New applications, new use cases and domains where data engineering can be used in I will submit more ideas ideas every week going forward. Cheers, Javon. And I

8:19 going forward. Cheers, Javon. And I believe that’s the right way to say it. Good. Good enough. I’ll take it, Tommy. So, let’s talk about this talk. I think this has changed for me over the years a little bit., do you remember those early days of like ideas. p powerbait. com? Everyone was like, “Oh, this feature would be great if we had this. This feature would be great if we had this.” And we went back and said, “No, you have to go back to ideas. p powerbi. com and go get all the data there.” there.” I don’t know if I actually recommend that very much anymore. Have you been like when you talk to customers or you talk to people? I don’t

8:50 customers or you talk to people? I don’t I in recent years I have not been suggesting run back to that area and go submit your ideas. What do you think? What what are your thoughts with back in the day? Yeah. I So give a bit of back of back back in the day to right now. So back in the day there was ideas or it wasn’t ideas PowerBI tips but it was PowerBI. PowerBI. Do you remember the URL? You remember it’s ideas. powerbi. com. That’s what I thought. So that was basic. basic. You can still go there. You can still go to ideas. parvey. com.

9:21 to ideas. parvey. com. It just redirects you now to community. fabric. com. Yes. Yes. One of the one of the forums called fabric ideas. And so now it’s just fabric ideas for like everything all in one place. So back in the day, this ideas, it wasn’t a forum. It actually didn’t look anything like it does now. And what it was is simply anyone’s ability to submit an idea or vote an idea. Pretty simple. But the thing was really I I want to say for the first four, five years of PowerBI’s existence,

9:51 PowerBI’s existence, the Microsoft team basically went through like their releases what they were going to work on very much off the ideas. ideas. they would mention it on the blog like hey these are all coming from the ideas, these are the most popular. the status of the ideas would change to be able to see even ones like the status was hey we’re gonna actually start working on this or this is going to be completed in the next cycle and it was a really everything that was submitted in PowerBI ideas was from the

10:22 submitted in PowerBI ideas was from the community was from really non-Microsoft people people and that really shaped the product I think the initial days it really felt like the ideas. power. com pyber. com was shaping the product and there was a lot of emphasis to go there submit your ideas. I think also at some point Tommy when things were small it was a lot easier for people to manage update and groom the ideas down into like real product features. I feel like some point in the middle here before we got fabric but then more when Microsoft fabric or

10:54 but then more when Microsoft fabric or PowerBI was like a lot more mature. I think we had a point where the number of ideas that were getting added to the ideas backlog was just proliferating too much. much. Yeah. And so I remember actually I I remember going to a point where I was trying to add ideas to the ideas area and it was there were so many ideas that I couldn’t even get my idea into the queue because I didn’t know if there was another idea that was similar. It just feel it just felt like to me there need

11:24 feel it just felt like to me there need to be like a lot of like really a better tool to like group ideas similar ideas together and start really consolidating the number of votes. And I don’t really it feels like to me some of the the focus has moved away from ideas and has moved more towards enterprisegrade features. Yes. Yes. I feel like for me like the product team has listened to the ideas area but the ideas area is for like everyday users.

11:54 ideas area is for like everyday users. Could be a small company, could be a big company. Everyone can vote for the things. Great. Love that. But I feel like right now a lot of the features we’re seeing come out are just focused on larger enterprise and it’s less about polishing or finishing things inside desktop or PowerBI desktop particularly right it feels like a lot more around like enterprise features CI/CD the service and I feel like there’s a lot more of a movement towards let’s get all these features added onto fabric so customers have a migration path to go from just PowerBI licensing into fabric

12:25 from just PowerBI licensing into fabric licensing as well. Is that your perception too, Tommy? Is it filled with you? you? Honestly, when did PowerBI premium even 12:33 Honestly, when did PowerBI premium even come out or pre per user? It wasn’t until maybe three or four year or years after PowerBI desktop and yeah, in the very beginning, the people who were submitting ideas were the developers, the individuals, the creators., and Microsoft was still working on an enterprise platform or an energized tool. They had the business tool, but they did not have nowhere near the enterprise tool., move

13:03 enterprise tool., move forward to where premium per user really became the main standard. And that’s when I noticed the shift too, Mike, where a lot of the features that were coming out which were needed and are needed especially in an organization level settings from security to not just your admin settings but to your point as we see now with CI/CD as we see with notebooks having this more enterprise tool the ideas that are being submitted on PowerBI

13:33 submitted on PowerBI are very much still about desktop. They’re about the little things that people do. Just looking at the hot ideas right now. Copy values from PowerBI visuals conditional formatting and then like test as an audience improve data slicer. So again that’s a very much on the builders which is really what power Microsoft needed a shape in the beginning anyways. Yeah, I would agree with these I I would agree with these comments too, Tommy.

14:03 agree with these comments too, Tommy. And I’m also,, the ideas website has substantially changed from what it was, right? It was it was like a another tool and it got moved into this new area. And so I I I feel like for a while there, I complained a lot about ideas were like unarchable. I couldn’t find them.,, looking at the different ideas, they’ve done a much better job improving the idea site. It’s much more like a forum now where you can actually read the first two sentences of every idea and actually understand what the idea before it was like literally just a

14:33 before it was like literally just a sentence. It was like here’s one sentence sentence of the title of the of the idea and you’re like well I don’t I don’t know what that means. Like do I do I even want to click into it? It was just so much extra effort here. so I’ll I’ll argue with you there on that one, Tommy. Like it’s it’s definitely improved from where we’ve been. But, okay, I’m going to go do a quick trivia with you, Tommy, here. here. All right, let’s let’s go back., hold on. I got to get the the sound effect here.

15:05 You asked the question. You asked the question. When did PowerBI premium per user come out? Give me a guess, Tommy, without googling it. I know you probably Googled it. Take a second here. I haven’t done it yet. When do you think PowerBI premium per user went was announced? Now, I’m going to be a little bit vague here because there is a blog post around when it went GA. Okay, so let’s just let’s just talk about when did PowerBI premium go GA.

15:36 I remember even submitting the special Microsoft form on what they would want premium per user to be way before it was even out. Yep. Yep. I’m gonna say if data flows were 2017 and I was still a pro feature. Mike, I’m gonna go ahead and say 2019. Great, great comment, Tommy. The announcement for PowerBI premium generally available. This is PowerBI Premium generally available. I’m not sure if this is premium per user. So I

16:06 sure if this is premium per user. So I PowerBI premium went general availability in 2017. So that’s premium., okay. Now I’m trying to figure out Yeah. where premium per user came out, which is probably You’re right. Sometime after that. So, I’m checking that one right now., premium per user general availability. Oh, wow. This is way later. Okay. Good guess, Tommy. March 2nd, 2021. 2021. Yeah. So, premium came out in

16:38 2021. Yeah. So, premium came out in 2017. So, desktop released in 15. Premium came out in 17 and then you roll up to 2021 where now you have premium per user showing up and that’s an announcement made by Arun announcing PowerBI premium per user generally available. So that’s So that’s 2021. 2021. Yeah, that’s about the time we started the podcast. Just a little bit thereafter. We started the podcast I believe in May of 2021 as well. Yeah.

17:08 Yeah. Huh. Yeah. So So that’s crazy. And then two years later, fabric. fabric. Then two years later, all fabric shows up. Yes. Correct. And now we’re back on to like let’s go back to premium,, just straight premium. We got to get that everywhere. Like let’s let’s everything now. Then premium just becomes now fabric, right? So fabric just basically migrates into like, hey, this is now premium. The reason I I bring this up, Tommy, is you’re talking about diff different ideas. How do you get things to stick, right? Where’s where’s the investment from Microsoft coming from? And I think

17:39 from Microsoft coming from? And I think there’s a lot of when you go to the new ideas website there’s a lot of areas that Microsoft is trying to build into fabric platform admin capacity CI/CD get data hub one lake there’s now an IQ space IQ ontology powerbi general product real time results data factory data science data warehousing I would argue Tommy the areas that where Microsoft has seen the most investment are the ones that have the categories

18:10 are the ones that have the categories potentially inside the ideas area. So like there’s a lot of investment at the data factory level. There’s a lot of ideas. There’s a lot of breaking down of categories, right? That data factory data integration space. Databases are a little bit there. There’s some like some Cosmos DB. There’s some SQL database stuff. Okay. The fabric platform admin capacities data hub governance one lake realtime security support workspaces. That’s a lot of space right there. the the the platform is quite large with its ideas and there’s only one little tick

18:40 ideas and there’s only one little tick item for PowerBI. Wow, Wow, that’s sad. I and okay so this is where we’re at now and I think it’s hard to navigate especially fabric because Mike fabric is predominantly an enterprise tool even if you are a small organization like the major updates that people would submit on the ideas and you had the report didn’t you build the report when it was built in community jam jam I did I had the report on community jam I think I took it down eventually because Microsoft moved the site from

19:12 because Microsoft moved the site from the old ideas site to the new idea site one of the things that got me was I looked at how many items were like in the needs votes area. I’m like how many votes do we need for something to actually get like worked on, right? When something says needs votes and this is where I think part of the ideas area is interesting for me right now is how let me let’s just conceptually step back. Let’s just step back and say look Tommy, you’re Microsoft Tommy. How do you manage your product teams to get them to build what people

19:43 teams to get them to build what people want when you have, let’s call it, thousands of ideas fracturing across many different people’s vision of like what the software should be doing, what the software should be making. Like there’s there’s I can’t even imagine and again I’m trying to relate to this Tommy because I’ve got I do software building. I build software for stuff and you can build a lot of different things and it’s difficult to rein in all the ideas and centralize around here’s the

20:14 ideas and centralize around here’s the thing that we should build that is going to drive the most monthly active users. Here’s the thing we’re going to do to drive growth of the platform. A lot of these ideas I think are good polishing ideas but they don’t necessarily grow the platform all the time. time. Yeah, 100%. And here’s the thing though too and to me this makes a big difference. I know you’re probably going to disagree with what I’m about to say, but the PowerB ideas worked because it was an application, right? There were ideas obviously for the service and you

20:45 ideas obviously for the service and you ideas obviously for the service and things that were happening on know things that were happening on the web, but what made PowerBI ideas so impactful to me was very much the fact that you were dealing with a it was just it was an application development. Microsoft was building a software that was PowerBI and those are the people using that application that anyone can download. Well, your deal with fabric now is you can’t really just like open up power fabric without an internet connection, right? Like

21:16 internet connection, right? Like PowerBI desktop, I can use any CSV file, get started in two minutes. Usually, if you’re doing anything in fabric, and this is to me, I think a shift in just behavior. you’re not just I don’t want to say messing around in fabric, but usually when you create an artifact,, that’s you’re creating something in the workspace and there’s always an intention behind it. I know you and I know a lot and you’re like me where there was just a lot of let me see what I can do in PowerBI and a lot of things I didn’t

21:47 in PowerBI and a lot of things I didn’t have to save or just you’re testing around. So it really had that I want to say more friendly experience. Now if you’re going to update something in fabric or it has to integrate with everything. It has to integrate with all the different services. So ideas don’t make as much sense now too. It’s the individual versus the the wider group now that’s interacting with fabric. you’re bringing up an interesting point and maybe it’s maybe it’s the idea of

22:17 and maybe it’s maybe it’s the idea of are what are what is the purpose of the ideas right at some level there’s some very big architectural decisions right it’s hard to throw ideas from the community at these large architectural decisions right I’m going to build a lakehouse what does that mean someone has to have vision for what the lakehouse means and how it integrates with all the different tools how does each tool integrate with the lakehouse like that that’s something you don’t just like on a whim pull up an idea from an ideas page and then just implement, right? That’s more strategic. That’s a lot of work. That’s a lot of investment. I understand that part. It

22:49 investment. I understand that part. It feels like the initial days of the of the ideas page was really around like refining the rough edges of desktop and just trying to getting a consensus from the community what people in the community really wanted to feel for like shifting a feature like refining a feature. Right? Microsoft’s already going to build new visuals, right? Okay, great. What new visual should they start with? Because there’s a limited amount of time and money to build all of them. What should they build? Right? I look at a lot of these ideas here on

23:19 I look at a lot of these ideas here on the ideas page and just it just feels like a lot of these are stagnant to me. Now, I know things are happening. I know things are moving along, but Tommy, have we have we hit a world where an ideas page for your your product is no longer a thing? a thing? Like let me let me give you let me give you an example. I’m going give you a thought experiment. Okay, I’m I’m listening. I had someone send me a link yesterday and the link was someone talking about

23:49 and the link was someone talking about their application, their web application. The web application had a instead of an ideas page, it had a submit your prompt here. So it was basically a window that you would click and you would type in your prompt. I wish the app would do XYZ thing. I wish the app would build this and include screenshots, include images. It had like this whole whole area.

24:19 It had like this whole whole area. When you turn into a world, Tommy, where anyone can when when the creation of the code becomes developed by an agent, the agents can build the branches, the agents can do the testing. I think I just read an article this morning that said that Anthropic is using its current AI to build its next AI. 70% of the cloud code that’s written right now for their next version of cloud code was written by the prior agent.

24:49 written by the prior agent. Software developers are getting aware of this and saying if you could just prompt your way into the next feature. Does this change Tommy how ideas in general work with software? Do you build software and just say submit your 25:03 software and just say submit your prompted feature ideas here, describe the feature in detail, and then you just let the AI go chew through like you have some checks. You have to do some checks. You can’t have like,, bad things coming through prompting things. But do you just go out and build a version of the software with this feature in it? Mike. Mike, and just have it built you as an individual. Sure. or if you’re not supporting,, I don’t know,, over 1 million companies, then maybe

25:33 , over 1 million companies, then maybe not because it’s very different when you’re as big as Microsoft. Like, I don’t think Google’s doing that anytime soon. soon. But, but imagine this though, Tommy. Like, what if your product is what if your product is really truly developed by the users who use it, right? Let me let me just pull back something a little bit here. Let let me let me re tease the idea again a little bit, Tommy. If your if your software can be replicated by anyone, right? If your whatever software is a service, right?

26:03 whatever software is a service, right? If the if someone can take if someone can take screenshots of your product and go send it to an agent and say, “Write me or build me or do whatever make this thing thing and it looks very similar.” I’m I’m seeing right now this happening for Figma to some degree, Tommy. Figma is a great design program, but it’s a it’s now owned by Adobe. They have this kind now owned by Adobe. They have this,, clench on the market of,, clench on the market around like graphics and graphic design stuff. Okay. Interesting. Why can’t you I’m seeing people grab entire Figma screenshots and just

26:33 entire Figma screenshots and just rebuild the entire app and now it’s like a an open- source or a low code version of the same software. What barrier do you have in owning that software now? So I’m under the impression here that if this is if the new world we’re walking into is everyone is a software developer, everyone can prompt their way into the next feature, the next idea, the next design, the only way for you to

27:03 the next design, the only way for you to remain competitive against your competitors is allow your community to help you build not just create the ideas page because this creates a lot of extra friction right now, Tommy. Right? go to the ideas page, submit an idea. Someone on the product team has to go review the idea, see if any other ideas are similar in nature. And why not just shortcut all that? Why not just figure out how to make that thing become a new feature branch and then someone just reviews the feature branch and say, “Does this work or does this not work?” Or refine the

27:34 or does this not work?” Or refine the feature slightly and make it fit the product goals. Do you really think Microsoft I’m I would love to know their workflow building office like you’re missing I feel some incredible ex excuse me components when it comes to enterprise software and that that’s what we live in now that’s what fabric is this is not just a anyone can download security security can I argue can I argue your your point

28:04 can I argue can I argue your your point real quick around enterprise software Okay. Okay. The barrier to make enterprise software, I think, is dropping. Why Why it’s dropping? The barrier to do that is getting less and less every day that we have this going on. So, keep going. I’m just going to throw that there. Well, first I’m going to vehemently disagree with that because you may be able to build something. However, you better dang well support something from networking to also security. and who’s on the line if from anyways I’m not

28:35 on the line if from anyways I’m not going to get into from the IT point of view view how how is that any different than like would you argue claude coowork is not enterprise grade no not yet you don’t think claude coowork is enterprise grade where you you literally have an addin for powerpoint word and excel and it’s competing directly and so much so that Microsoft came out with co-pilot coowork which a good announcement there. Satia just tweeted recently co-pilot coowork is now a

29:07 recently co-pilot coowork is now a branded thing. It’s a thing that’s inside co-pilot and licensing some changing now but that happened because of claude coowork that that’s the whole reason. So I would argue anthropic and co-work is an enterprise piece of application which now 70% of that is now being developed directly by the agent itself. So, I’m gonna I’m going to disagree with you on that point that enterprise software is not arriving that isn’t almost entirely built by agents. Now,

29:39 almost entirely built by agents. Now, keep going Tanya with your point. I have another point I’ll follow up with. Yes. So, again, individual individual editing a single file or files. It’s me, the individual going in saying update my PowerPoint, update these Excel files. The application’s already built. Enterprise is a little different, Mike, because it has to want to obviously work across systems, cross collaborate., we’re talking about something like Microsoft Office or Gmail. Yeah, I’m

30:11 Microsoft Office or Gmail. Yeah, I’m sure you could build your own Gmail if you wanted to, but the other side of it is for yourself. Again, this is what we’ve we’ve gone back to back and forth when it comes to the AI conversation. AI right now is amazing for me the individual, for me the few. But when it comes to us, the collective, the 50, 000 people who need to use the same software or application, well, that’s where we’re again finding the limitations. I know

30:42 again finding the limitations. I know what you’re saying. I could probably build something that looks like Outlook that maybe has a domain, but I cannot support that across,, 100 million thousand emails. Anyways, let’s go back to Fabric here when this is why I’m calling fabric an enterprise software because you have to be in the system. PowerBI desktop free to download. I didn’t even have to submit an email and I could download PowerBI desktop and use the entire tool. Obviously there was the sharing and that was part of that enterprise feature but

31:12 was part of that enterprise feature but anyone can get started. Okay. So hold on hold on. I understand your comment around enterprise software here Tommy Tommy and this is maybe a bit forward thinking more than most people. I think software companies and especially larger ones are going to have to have the ability to I think they’re going to need the confidence to be able to burn down their entire software stack and rebuild it and start over. I I think I think the level

31:45 start over. I I think I think the level like let me give you some perspective why I say this. Someone was doing an experiment with cloud code. They were like hey we want to build a compiler for Linux or something like that. They’re trying to compile an operating system which is a very complicated solution to to go do they this is where the agent teams idea aroused from right so anthropics doing testing they build these things called agent teams and now agents can then talk to each other and and make bugs and make branches and fix code and make code new code all these things. They basically threw a

32:16 these things. They basically threw a team of agents against a compiler and had it work over a weekend and it rebuilt a new compiler all by itself. It it debug things. It handled merge conflicts. all the normal software things that you were doing previously was happening faster with a handful like and again I think they spent like 10 or $20, 000 in tokens to do it because it was it’s an expensive project for like from a token perspective but they were able to execute this and have it build all the

32:46 execute this and have it build all the software software I don’t think traditional software companies are going to be able to compete Tommy in this space unless they are re-imagining their software ware at a speed in which my ideas can go from idea into product in like minutes or hours as opposed to weeks or months. The tolerance is not there. And if you don’t do this, your competitors will do this. And I think I think to me this is the

33:16 And I think I think to me this is the point where I’m like looking at this going going I don’t know if I want to be a software developer as much as I want to be a person who enables a community of people problem solver like pro so problem solver is is a better I think Alex P put out a little thing on Twitter or or LinkedIn or something like that that I thought I resonated with a lot. Instead of being a program manager, a PM, you need to be a problem manager. Figure out where problems are in your

33:46 Figure out where problems are in your workflows, those are the things you can automate out. And so what I think I’m seeing right now is I’m seeing a lot of companies need to focus on what are the workflow problems that we’re having and how do we use AI to build tools that help that become smoother, easier, faster. And so me me internally as our company, we’re tearing down all of our software and rebuilding with agentic like experiences to help speed up the development cycle of software. Any new customer we’re doing is going to have a

34:17 customer we’re doing is going to have a level of agentic development software patterning with how we build things with our engineers. The engineers are still involved. They’re still doing the reviews. We’re still doing all the things we used to do before, but we’re adding now with AI, we’re able to add more agentic spaces on this thing. So I think ideas in its traditional sense is now going to have to shift more towards how can you get an idea from the community into your product as fast as possible by letting the AI noodle on

34:48 possible by letting the AI noodle on that idea for you. Let me just pause there. What do you think, Tommy? This is where I think we differ because I can see a lot of hesitation there. Yeah, I I think right now I I can see that being a possibility, but what you’re basically saying is I Mike in a world of 2. 5 or how many billion people are here, everyone is basically going to be on the run for creating their own application, their,, their own closet. It’s like when we

35:19 closet. It’s like when we transitioned from everyone had the same television shows when you went to school and you’re like, “Hey, you guys watch the same show because we only have four channels.” And now no one watches the same thing because you have Netflix and everything’s super catered which is fine with entertainment. But I don’t think you can do that with applications and especially in a work environment. This is where we differ because because can I can I give you another story? Sure. Sure. Okay. PC Mag just talked about I read this article the other day and I will put this in the chat window for anyone

35:49 put this in the chat window for anyone who wants to read it. PC Mag just put out another article here. Hopefully I got this right., there’s actually a whole bunch of text things here. Let me just read let me rec copy this. I’ll put this here in the chat window, Tommy, so you can see it. This article talks about Mozilla Firefox,, right? Enterprise software. People use it. It’s fairly secure. According to Mozilla researchers, anthropics AI code claude opus 4. 6 discovered 14 high security bugs and issued 22 CEVs over the course of two

36:22 issued 22 CEVs over the course of two weeks. weeks. Almost 1ifth of all high severity Firefox vulnerabilities that were remediated in 2025 of the security vulnerabilities resolved in an entire year has been compressed down to two weeks. This this is what I’m talking about. It’s stuff like this. It’s it’s the idea that I understand Tommy you’re talking about like code has to be secure. It’s got to be good. all these when you have agents that can go through and find

36:52 agents that can go through and find security vulnerabilities at this rate, at this speed, and not only just identify them, but then also come up with fixes for them and push them back into your software. This is this is what’s happening to me. I look at this going when I see articles like this on large software companies that are using these code things to help accelerate their security, the new features, the things they’re building. This just wants me to push more towards this direction. And so I think your hesitation here is just like a very traditional software

37:23 like a very traditional software development pattern. And I think companies that don’t adopt this are still going to have are going to have a really hard time keeping up with the new guys who act like if my software has 14 less security bugs than the next guy, 37:37 14 less security bugs than the next guy, that just makes my story all that much stronger. If my software is coming out with features every couple days as opposed to every couple weeks, that just makes my software stronger. And I think I think to me, Tommy, that the the the aha moment here for me was in January, February of this year, the the agents that are coding things aren’t just making new code that is better, but then falls off and it needs a little bit more work to make it

38:07 a little bit more work to make it improved. I think we’re starting to see the ability for agents to not only create and look at code that you already have and actually improve it over time, not incur additional tech debt every time it builds something. Okay, I think we’re there’s a lot to remain to be seen. Mike, what about fabric in this? Are you then just suggest suggesting that Microsoft has an AI to say what you want updated and they’re going to use that? Yeah, it’s called Copilot. Just just

38:38 Yeah, it’s called Copilot. Just just tell Copilot what you want. It’ll just work. work. No, with fabric. So, yeah. So, okay, all jokes aside, but I do think Right. I I don’t know how you get there, Tommy. Right. It It’s a lot more of this. I can say a lot of things because I’m the bystander. I’m the one on the edge looking at Microsoft and then I can poke ideas at things and and say how it should or should not work, right? I should or should not work, right?, I would I would love to walk into

39:09 mean, I would I would love to walk into the Excel team and say, “Hey, let’s throw a whole bunch of cloud code and rebuild all of Excel. Let’s just re like just make it all look the same but rebuild all the back end of it. So all the bugs and issues and performance challenges that we’ve had in the past just go away. I don’t know if that’s possible, Tommy. To your point, there’s been so much work on top of Excel, how it’s been built, the legacy of things that have been there, the way it’s been written. I also think there’s a challenge, Tommy, around legacy software. So anytime you’re dealing with a lot of software that is like really old and has old tech stacks

39:40 like really old and has old tech stacks in it, I find it more difficult for AI to figure out patterns and innovate and build on top of it. So you really have to like again as a software developer I have to really step back and say how can I re how can I rethink my software solution in a way that agentically will allow it to be developed. Even your development cycle your process has to change. change. Right? If I if I have a really complicated build process that the agent can’t see what code it’s making or what

40:11 can’t see what code it’s making or what changes it made to the code or the codebase or whatever that is, it’s going to be very difficult for the agent to just make a change, see the change and then fix the change. So that’s why a lot of what I see software right now is I think this new world of ideas and how people build software is pushing every company to build the browser because browsers are easy to build compile code for the browser is a very clean simple interface

40:41 browser is a very clean simple interface and now you can get a lot of automation and scraping and playright and MCP servers that help you interact directly with the web page. So I think this agentic space is just going to accelerate more web development because the the reward of building things agentically is easier when you build for a website. You can get all the CI/CD stuff done. You can have it compile. You can have it build locally as opposed to building like full-on desktop applications where you have to like build the app, compile it down, get an

41:12 build the app, compile it down, get an executable, install said executable, and there’s there’s just more friction. I think it can be done. I’m not saying it can’t be done. It can be done. It’s just more friction and you have to change how you develop in your development cycle. I don’t know if we’re going to get away from that in desktop or building things with software on computers. But I will argue most of what I use now is 100% web browser brace web browserbased which is also but again that’s all right. So for Javon

41:44 but again that’s all right. So for Javon who submitted the question where if he’s listening going where this go are they off this is a rabbit trail. So, I got to ask you though, though, how do you get into get your ideas if you have a great idea for Microsoft Fabric?, get it to the team, get get some traction around it because before it wasn’t just submitting the idea. You and I both know there was also a lot of I don’t want to say politics, but, we had to like then post on Twitter. I just submitted this idea.

42:14 on Twitter. I just submitted this idea. Everyone should take a look at this. So, and it would get traction that way. So for someone, if I want to get people to know about what’s going on in with my idea with fabric, what should people do? That’s a great question, Tommy. Well, a couple things. Let’s let’s let’s answer the questions in order, right? Can you please talk about the how to submit fabric ideas? Well, one is go to ideas. par. com. It will automatically redirect you to the existing fabric ideas page. page and then on the fabric ideas page you can you can sort

42:45 fabric ideas page you can you can sort and filter down to like an idea that you care about or or a part of the product right you go there you can filter down to PowerBI so there is a dedicated website page for ideas you can go there I’ll put the link to that in the description as well I’m just going to confirm that this link works correctly it does not hold on let me go back I’ll hopefully get the right link here let me just go back here to the home I don’t know why fabric or PowerBI

43:15 I don’t know why fabric or PowerBI ideas and it did redirect me. It did. It did. It did the redirect. Okay. Yeah. So, it is PowerBI ideas. Okay. so, going there, it’ll Google you right to it. So, okay. So, just go to ideas. powerbby. com. It will redirect you to the right site. I’ll put that in there. And then it’ll bring you into all the fabric ideas. And then you can select different items there directly., and then it’ll it’ll work, I think, a little bit better there as well. So, you can put that down. Here’s the ideas site. That’s how you

43:45 Here’s the ideas site. That’s how you get there. Once you’re there, you can search for your ideas. Search is okay there. It’s a little bit hard to get navigate around. You can look for top ideas or new ideas. Those are the things I usually throw my votes towards. If I’m voting for something, I’m signing in. I’m looking for some of the,, the hotter ideas. Oh, yeah. Oh, yeah. Filter my area. Pick on some of the hot ideas. Throw my votes towards some of those things. I will say Microsoft is looking at ideas. That’s anything above I think they said like above 150. It’s at least making it to someone’s radar where they’re looking at maybe it’s a bit higher now. Maybe 500 votes.

44:16 bit higher now. Maybe 500 votes. Wasn’t that just PowerBI? Was that this is fabric too? I don’t know how they do it. I don’t know how all Well, so there’s many teams now working on this Tommy. Like it’s not just one team. It’s like many teams. So I don’t know how every team does every team manage it the same way. Is every team getting pushes to information the same way? I not sure yet. But that being said though, Tommy, you you can go in there and look for ideas and find ideas that are there. You can also submit an idea and during the submission of ideas, it

44:47 and during the submission of ideas, it will try to find similar ideas based on your titles., my preference around submitting ideas, even though I don’t do it very often, is to find search for ideas that are similar and instead of making a brand new idea with my idea in it, I try to find something similar enough and add to the idea in the comments or the thread, hey, here’s what I think would be added needed in addition to whatever the the main description is. That way, you’re again putting your effort towards things that are already in motion. I think that’s a good idea.

45:19 that’s a good idea. It generally works better because here’s the thing. You either want the recognition or you want things done. I would rather have things just getting changed and done. I don’t need my name on this is an idea from XYZ per like I don’t need my name on it and that’s fine. fine. Yeah. I would rather just see the feature done. Correct. Now I’m going to also offer another option here too, Tommy. There There there are some things of fabric that I think are very core to what you want to build. There are also some things in

45:49 build. There are also some things in fabric now that you’re not as thrilled about. So there’s some challenges around rightback. Rightback seems to be expensive. Rightback doesn’t seem to be very easy to do. to do. So how do we leverage things like that? Like how do we get those features improved? And what I would argue here is this is where you can take it upon yourself to integrate with things that are of adding value to you through

46:19 are of adding value to you through fabric. So for me right now I’m building applications using semantic models, PowerBI embedded and SQL servers like I I’m the fabric is my backend now to my applications. I’ve done three applications now and all of them are residing in a web application but the back end of it connects directly to fabric. This is proving to be a really interesting pattern for me. So I’m picking out some of the things of fabric

46:50 picking out some of the things of fabric that are like really important to me. Semantic models, lakehouses, storage of data, SQL servers, these things are difficult to just vibe code into space, right? So, I’m using those experiences and I’m building new front-end UIs on how I want to see things built, right? So, that’s where I’m spending some of my time. And I I like this idea of like build in public. Share what ideas you’re building.

47:20 ideas you’re building. tweet about them, put them on Reddit. I know a lot of the product team reads Reddit right now. So, if you have a really good idea, go vibe code something. Go build a web app on top of what you think the feature is or the idea is. Go put that idea on Reddit. Let people see what you’re building. I think the product team is really looking for these things and and they will pick up some of these ideas if they see like working prototypes of these ideas and they can see some emotion or feedback or or directional from the

47:50 feedback or or directional from the community like, “Hey, this is a neat feature. I built this. Let’s go check this out.”, I think that’s really interesting and I think that’s another way of informing Microsoft around here’s something that the community wants. We built it, the community builds it on top of some of the foundational pieces of fabric. And I think that’s really going to be the way you’re going to start moving Microsoft to start to adopt and and develop net new things that’s going to be easier for you to build. Yeah. And honestly though too it’s tough because right now

48:21 because right now the last two years where was Microsoft number one focus it was integrating all these enterprise tools because not I don’t think any of the fabric features were individuals from notebooks spark notebooks data warehouse real time intelligence machine learning all these were enterprise tools in an existing place. So, am I missing anything? That was more just a from an individual point of view

48:53 just a from an individual point of view that they’ve integrated into fabric. So, they’ve been focusing on just making sure that all works together. Mike, I think we and they’re going to keep listening to those enterprise customers for a bit. It’s going to be harder to get your voice out. For me, I actually do believe though now, not that things are settling down, but we’re beginning to see, you down, but we’re beginning to see,, less kinks know, less kinks that the fabric ideas will come through. And also, Microsoft loves examples in

49:24 And also, Microsoft loves examples in the open source community., many times they’ve taken that and run with it if something gains traction. I don’t think just a simple ideas board is enough now to say like what you want to see get done get heard but create an example GitHub repo with either I’ve seen examples of people of their proof of concept Yep. that would have all the infrastructure built out or what they’ve tried to do. And if that gains traction,

49:54 tried to do. And if that gains traction, I know and I’ve seen Microsoft listen. I believe some of the scripts that they had for the audit log came from that or something like that with the PowerShell module and they took it over. So you I think there’s more work that you have to do for what features you 50:08 you have to do for what features you want. want. I I wish that Javon put in some of the ideas that he was thinking about to some ideas that he wanted to submit. rather because Mike, if you could ask the team right now and you could submit five ideas that the my the fabric team would see, what would like what would be some of those ideas? You don’t have to think of five. That would be crazy. be crazy. You mean like if I had if I needed some features that were like must-have

50:38 features that were like must-have features? features? Yeah. And you wanted to get some quick recognition of those features like like it’s going to make it in the product. Like if I had a m So this is like Michael’s magic wand, right? If Michael had a magic wand and I could just say feature existed, make it happen. Like it would just materialize and appear. Okay. I need to ask a question. Are we talking about only features for PowerBI? Are you talking about any features in fabric anywhere? Well, what are you what are you asking about? about? It already gets complicated, but no, it is fabric because for Javon, he said, “How do we submit fabric ideas?”

51:09 “How do we submit fabric ideas?” Okay, so I’m going to put a couple ideas toward like I’m I’m just going to throw out some ideas I think have been interesting to me. where I think would be useful. I think one of the areas I would put my magic wand towards would be there is I think a really big need for a table slicer. I think I think adding a table inside PowerBI reports that functions just like a pivot table inside Excel was would be very useful. Right here is a table. These are

51:39 Right here is a table. These are columns. You select the dropdown on the column. The column lets you pick or filter items in the column just like you would in a pivot table. And then when that table filters, the entire page filters. So it’s like a slicer and table all mixed together. The column headers become a series of slicers that filters the data of the table which also applies those filter context to the rest of the page. I don’t think anything like that exists., and to me personally, I see a lot of gaps currently inside table and matrixes that just don’t meet my

52:11 and matrixes that just don’t meet my needs as a,, a traditional Excel developer coming into PowerBI. It just seems like it’s a a half effort thing. So, that would be one thing I would solve right away would be that part of it., the other idea I would probably throw at some things here, Tommy, would be around it would be a feature more for like the fabric side of things. I would like to see a little bit more emphasis on API and API access for everything in the platform. I I think I really do

52:42 the platform. I I think I really do think think all software is going to need to be is going to need to shift its design work. So let me let me describe what here. Tommy, how many how many AI agents do you have running for you right now across just different computers? Like how many would you say you have like just rough just a number? Just rough numbers. Yeah. say no more or no more less than seven seven seven different that they do very specific things like there’s one agent that’s doing this work if I I may have you have another agent here

53:12 may have you have another agent here that’s like helping build my statement of works I may have another agent that is is managing my website just little Tommy you’re one person you’re managing seven employees that are helping you run your business okay now let’s just let’s just say this ratio is let’s just be fair Let’s say there’s a 1: 4 ratio. Microsoft is now lifting out co-pilot co-work. Enthropic has co-pilot co-pilot has claude co-work right

53:42 has claude co-work right all these things. Let’s just say the average employee now gets four agents for whatever they’re doing. You’ve just multiplied a 4x need for other tools, other individuals, other agents to access your tools. So pull that pull that thought forward forward. Six months from now, the number of automated computer agents needing to access your software, it’s not just one person. It’s now four times that in

54:13 person. It’s now four times that in agent mode. Agents don’t like UIs. Agents don’t like web pages. Agents like APIs. Agents like forms. Agents like things that are very programmatic and code-based. So, I think the new world is going to be designed for agents. It’s going to be designed for a CLI. Hey, Tommy, I’m going to build a new feature or something on my website. The first thing you should be building is an API, API layers, and all this code, agent level access directly into that

54:44 agent level access directly into that code because for every one person, you’re going to have four to seven additional agents trying to access the same information. that that needs to happen. It’s going to happen. It’s already happening right now. And so, because you’re seeing the proliferation now of agents becoming so effective and so useful, you’re going to see agents popping up everywhere. And the only thing that’s going to make agents even more effective, Tommy, is going to be when you bring the price to run these models lower. Apple is making a very big

55:17 models lower. Apple is making a very big stance on we’re going to build the the hardware to let you run really big models, right? When you can run models on your phone that can do things like if you could have an agent bundled into your phone, Tommy, Tommy, it’s going to go insane. So for me, I’m looking So my my one feature would be like there’s some things in desktop I would fix. I think the other side of things I would fix is make everything very agent centric, make everything very

55:48 very agent centric, make everything very APIdriven, make everything capable with service principles, make everything capable with automation of stuff because that’s that’s going to be the ability for people to create whatever they need and bolt on to the existing core functionality of what fabric is doing. semantic models, lakehouse storage, data warehousing, SQL databases, all the data. You can’t agents can’t make that stuff up. You’re going to have apps, you’re going to have data. You got to have a place, a platform to store it all. What it looks

56:20 platform to store it all. What it looks like at the front layer can can change wildly. wildly. You are going full only ideas for AI then. Yeah, I think I honestly Tommy I think that’s I believe we are moving to a world where we have to design agent first. So strategically, you’ll notice on our website, Tommy, we have PowerBI tips has now gone full bar on all the knowledge Tommy and I have been talking about on

56:51 Tommy and I have been talking about on the podcast. It’s all there. You can search it all. PowerBI tips has been updating itself. We now have 300 episodes, full text transcripts, full videos, everything. all landed on PowerBI tips. So, if you haven’t checked out the new site, go check it out. PowerBI tips. Go look at it a little bit. We now have a search window that will let you search three years of knowledge all the way back to 2013 of episodes, conversations, things that are happening, things we’re unpacking on the podcast. This is a huge knowledge

57:22 podcast. This is a huge knowledge library. I’m trying to work on now is how do I get a gentic search on top of the website? So, how do I take the corpus of text that we’ve now just published and make it agentic a little bit more? So, I think to me this is the new world we’re going to be living in. We have to design agent first and that’s going to really inform the website design, the backend design, and I think most companies should start thinking, if you’re not already thinking API first

57:52 you’re not already thinking API first anyways, you need to be getting API first. That’s really where things are going to go. I’m gonna take it a little little differently than you are here because help yourself, Tommy. I’m just gonna ask for a fabric desktop application. That’s all I want. It’s a great ass, Tommy. You’re never going to get it. That is a pipe dream. That’s a pipe dream and a half. That’s so much more likely, too. And it’s probably going to happen. So, as crazy as that sounds, honestly, what?

58:25 crazy as that sounds, honestly, what? Yeah, the Asian stuff is awesome. But I I’m not gonna argue any more with you with that because you can’t though. You can’t like I know Tommy, you’re you’re arguing only because you have a little bit of hesitation in your heart. You’re like, “Ah, I just I miss it. I don’t want it to go away.” I think there’s a little bit of resistance to this. And the sooner people a acknowledge like this is the new way things are going, it’s it’s going to move like it’s rapidly going to move away from building code now. And

58:56 move away from building code now. And and John Kursky was talking about here in the chat here. So John, thanks for jumping in and chatting with us today. We’re having a lot of similar thoughts here. John is saying VS Code is the new fabric desktop app. I’m gonna I’m gonna agree, John, to a point, but I think VS Code is also not the right solution for this one. I really do think I I really think it’s going to be something more like a lovable or a base 44 or a spark from GitHub copilot. It’s going to be an experience where you can say, “Hey, agent thing. I’m going to

59:27 can say, “Hey, agent thing. I’m going to talk to you as an agent.” Just a chat window. Let’s plan out what I want to build and oh by the way, your source of data comes from this part of fabric. The agent should already know how to connect to that data source and build what you want. Kurt Buer is building full beautiful dashboards by prompting it. And he says, “I’m enjoying the development process.” Tommy, we’ve been just we’ve just we’ve just been given a gift. We’ve been given a gift of being able to talk to a

59:57 given a gift of being able to talk to a program, a computer, and articulate it with pictures and diagrams and natural language words. That is the way we will be building things moving forward. We will be building everything by just talking to the computer and giving examples and images. People are going to enjoy this experience. And it takes the barrier away from me having to understand heavy DAX. I don’t need to learn SQLBI stuff anymore. I have an ultra knowledgeable model that

60:27 ultra knowledgeable model that understands all aspects of DAX. And I think people are a bit there’s a lot of development in the community particularly around PowerBI that is very hesitant in this space. like they’re like, “Ah, but we can’t. DAX is so good. We need to have all this.” Yes, I understand. But the the gap between what the agent can do and what DAX it can write is getting smaller and smaller and smaller every day. It’s getting more capable every day. And every new model that comes out, it’s building better and better DAX. And the aha moment for me is

60:58 better DAX. And the aha moment for me is I can now watch agents absorb thousands of pages of information, distill it down into reasonable information, and build something on top of that. I can’t even do that. It’s doing things I can’t even do. So, how do you get like here’s the thing. My hesitation is not that I’m going to miss out or anything like that. To be honest, Mike, it’s okay.

61:30 To be honest, Mike, it’s okay. Like a beating drum or a drum that keeps on drumming. Yep. It is one thing for you and I to talk about what I’m doing for myself on my own PC, but then once this when you try to deploy things again right now, how many companies are creating enterprise software that are universally used? You have your Gmails, you have your Outlooks like out like let’s take email as an example and

62:00 like let’s take email as an example and this is going to be no different I think when we’re looking at the in the world of data. of data. How many email providers are there really right? Are you going to build your own? Interesting, Tommy. I I this is where I think things shift a little bit for me, right? So, I understand the email provider, there’s a system, there’s machines, there’s something to say for having the infrastructure in place to do that work, right? It has to be there.

62:30 there. But who’s to say Gmail and Microsoft and the outlooks of the world, who’s to say that’s what it looks like in the future? Who’s to say email still functions? Like 62:41 Who’s to say email still functions? Like we’re already seeing like agentic developed email solution patterns of agents and things going through emails and highlighting what’s important and autoresponding for things that just aren’t as aren’t as needed. Like helping you prioritize these things. I already don’t like email to begin with, right? If I could throw an agent at it to make it even easier for me to ingest what is going on, I’d be happier with that. So Tommy, I pretty good. I agree with you. Who is building enterprise software? I think it’s not a

63:12 enterprise software? I think it’s not a matter of who’s building it right now. I think it’s the it’s the rate of which someone can come in and tackle the problem a different way, way, right? And what are people willing to pay for? Most of the I would argue most of these software email programs you out today, the reason you use them is because they’re free. It’s just free, right? You get a Hotmail because it’s free. You get a Gmail because it’s free. You’re the product of that information, right? They’re giving you that product

63:42 right? They’re giving you that product because you’re actually the product. They want to see all your emails on their servers and be able to use them for whatever they want to see what you’re purchasing on Amazon, right? let me give you another example. This one Square just built a payment system for aentic spaces. So if you have Claude or Anthropic or Microsoft GitHub Copilot, Square has just built a real time real time biller of code because all the prices for all these different AIS are changing

64:12 for all these different AIS are changing all the time. It’s it’s more expensive here, it’s more expensive there. You use this model, it cost this much. Use that much, it cost that much. It’s complicated. The pricing is so dynamic right now inside agent stuff. But Square came in and said, “Look, just integrate with us. It doesn’t matter what platform you use. We’ll take care of all the negotiation, the pricing, and you just bolt on AI to your products and whatever AI is being used, we’ll capture the cost, mark it up, and send that bill to the customer automatically built into their product. What this does is this

64:44 their product. What this does is this says says Square is now going to have the price for every single AI model for every single company out there and they are going to know what the markup on every single model is all over the all over the world. And so if you’re a founder and you’re trying to build a new product, you’re going to go to Square and use them because they’re going to have all the information around the best advantages of which model to use in which environment. It’s this information that can be built when you start centralizing some of these things together. I don’t even know where I was going with that, but that just seemed

65:14 going with that, but that just seemed like a relevant fact. I You’re basically telling me like, “Hey, man, email is just what we want it to be.” to be.” It is. And I’m Do you ever remember what was Google Inbox? Did you remember that program? program? I loved Google Inbox. Okay. Google Inbox. They took it away from me. I know, Tommy. I have seriously contemplated rebuilding Google inbox like taking screenshots of

65:44 Google inbox like taking screenshots of it and and saying every email has a folder and you have these buckets and every email gets like I would love to vibe code Google inbox again again open source it and make it an email platform bolt on whatever you want Google inbox somewhere in the GitHub universe maybe someone’s got it somewhere I haven’t found it yet. So, someone has found Google Inbox. I really like that program. It was on my phone. It was very easy. Everything was categorized in a very simple way. It wasn’t super

66:14 very simple way. It wasn’t super advertising in any way. Like, I’m looking at this going, “That’s amazing, Tommy. I’m also going to argue with you.” Like, I go to Reddit now. There’s Reddit ads all over the place, dude. Reddit has so many ads on stuff now., and I don’t really like it. So, I built my own Reddit grabber. Basically, I grabbed Reddit threads from specific channels that I’m interested in, and I have a list of all of my Reddit threads that I care about,

66:45 threads that I care about, ordered by score, and I can see exactly what comes in. No fluff, formatted the way I want, easy to look at, condensed information. All this other fluff on Reddit is just not helpful. It’s just gotten busy with stuff I don’t care about. I I want to look at very specific things. So what I’m doing now, Tommy, is I’m taking open-source feeds of things and I’m consolidating them down into a condensed knowledge. I’m doing Google inbox inbox across most of my workflow now. And so

67:17 across most of my workflow now. And so now my I now I focus on the workflow. What helps me get from part A to part B as fast as I can in my workflow? How much of that can I automate? What parts of this are difficult and slow in in my process? and can I add any AI on top of that to make it more efficient? That’s where I think companies need to be thinking. And if you’re not building this way, if you’re not letting users directly directly So I’m going to say ideas are dead. Ideas pages, ideas sites are no longer a

67:48 Ideas pages, ideas sites are no longer a thing. thing. I think you’re going to have to I think you’re really going to have to come back and rethink that mentality and actually let how quickly can you get someone’s idea into a feature branch. I think that’s the new hurdle people need to figure out. And the sooner companies figure that out, the faster they’ll be able to produce code and good features and make a product. Tommy, who’s to say the product does I have a vision of of what my software program should look like. Is that the right idea? Is that the one that serves the most customers?

68:18 the one that serves the most customers? It may not be, right? It it that may not be the right way to think about it. Instead, you might want to think about something differently. You might want to think about let’s rethink this and actually focus on how can we have our community of users building the next feature. How can you shortcut that cycle? That’s where I think the next gold mine is going to come from. Mike, I await that moment in time. But

68:51 Mike, I await that moment in time. But here’s the thing. I think the whole time we’ve been wanting to say or not wanting to say that ideas are probably dead and you’re it’s just a the Lum is not worth the squeeze. Mhm. Mhm. But I think here we are because it it served its purpose. I do want to give some flowers to the ideas board for what it was because I was checking it all the time. time. Sure. Sure. and it lived it out and just like like to your point, Mike, so is Microsoft and Amazon. I’m not going to

69:21 Microsoft and Amazon. I’m not going to need Amazon anymore because I think I can build some AI to get my groceries. So, it’s just the way of the world. So for Javon or anyone else who wants to help contribute actually one thing I didn’t we did not mention too is if you really want to help contribute to what fabric looks like take a look at the people who write the blogs and start I don’t want to say start talking to them on Twitter but reach out to someone and because

69:51 but reach out to someone and because there’s always some preview or some testing that they’re doing that they want to get,, users feedback and if that’s really something you’re passionate about, they like, I would love to help out with,, the new thing you guys are doing and, you new thing you guys are doing and,, take a look. Generally, there’s know, take a look. Generally, there’s probably something going on., and that’s not just unique for MVPs. I know that other people as well who are not MVPs but have been in, you who are not MVPs but have been in,, certain testing programs depending

70:22 know, certain testing programs depending on the size of your company, etc. Sure. Yeah. to me if you because that’s what anyone is submitting idea is doing basically saying I know a better way at the end of the day that’s really what they’re trying to say no yes and no I would yes to some point Tommy but I also would argue that person submitting the idea is my workflow doesn’t quite fit with what you built and here’s what here’s the friction I see and what would I think would make things better some of these ideas are

70:52 things better some of these ideas are going to be genius and on point. Some of them are just going to be not well thought out and not really fit the the vibe or or you’re you’re actually adding a feature that’s going to actually add extra work to your process. You need to rethink,, reframe your mind a little bit about how it works. But like I think to your point, Tommy, yeah, I think I agree. Like it’s it’s this idea of I have an issue or a process I’m trying to do and your program doesn’t quite fit it. That’s what the idea site is saying. These are these are friction points for me that are just difficult to use. And so the faster you can take

71:24 use. And so the faster you can take those friction points and remove them into your product. Tommy’s I I’m trying to grapple with why would anyone want to use your software? Why would anyone want to use it if I can just go out and and recode it with something else in a couple days realistically? Right? I don’t know. If I can spend $20 in tokens to go get your software done, literally I’m looking at every piece of software in my business right now and I’m ripping them out one by one and rebuilding them and not paying for it.

71:54 rebuilding them and not paying for it. I’m paying for the to run my own application, which is way cheaper than having to pay for all these other software companies, storing my data that I don’t want to be stored in places that I don’t know where it is. Like I’m killing all of it. It’s all going away. this the barrier for me to get make this I’m having I’m tasking my team to learn agentic things by rebuilding core pieces of software for my business. So because I I think this is the way to go and now if you build everything with a code every time you want to make a feature change if there’s a process

72:25 feature change if there’s a process difference if you want to modify something it’s now just describe the feature to the AI and have it fix it. Yeah man Yeah man it’s crazy what’s going to happen. I I really I think this is the future of what’s going to be going on here. Anyways, all that said, really interesting discussion around ideas, product shaping, and where to go from here., let’s wrap here, Tommy, with some of these final ideas. So,, So,, for now, the best way to get your ideas to Microsoft is go to ideas. parbay. com,

72:57 to Microsoft is go to ideas. parbay. com, submit your ideas there, try to add your ideas to existing ideas today. Currently, how do we see the future of data engineering? Michael will 100% say it’s agentic and it’s talking to a computer with co with words. Tommy may have a different opinion on this one, but I think the future of data engineering is cooked. Everyone will be a data engineer in the near term here and you’re going to have an entirely different way of thinking about building things with agents. That’s that is the new world that we’re going to live in.

73:28 going to live in. yeah, submit more ideas. Let Microsoft know what’s going on. They do look at that today currently. Anything else Tommy you want to wrap with? Ideas are great. Ideas are fun. Build yourself. Use AI. There’s my ha ha coup with Yeah, I don’t Yeah, that’s the best I got. I We got to get out of here. I’m going to keep talking. Dude, it’s late, man. The the the world of things are is very interesting right now. I’m so excited

73:58 interesting right now. I’m so excited where things are going. Anyways, I’ve said too many words. Tommy, thank you so much for the conversation. Great topic today., Javin, I think is how we say the name. So, thank you very much for the question. Really good thoughtprovoking things. I think we’re in a in a world here where things are going to change very rapidly, very quickly, and I’m excited about it. So, as you can tell, that being said, Tommy, where else can you find the podcast? You can find us in 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

74:28 or a topic that you want us to talk about a future episode, head over to poweri. tips/ tipsodcast. 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. Thank you all so much. And we really appreciate listening. Thanks for hanging out this long. I know some people are still around. Hope you found this somewhat informative as well. We appreciate you and your listenership. We’ll see you next time.

74:58 It’ll be it high. Tommy and Mike lighting up the sky. Dance to the day to laugh in the mix. Fabric and A. I get your fix. Explicit measures. Drop the beat now. Pockets kings feel the crowd. Explicit measures.