Microsoft Build Recap – Ep.535

Microsoft Build 2026 gave Mike and Tommy plenty to unpack, but their biggest takeaway was simple: Microsoft is getting much more serious about agent-driven creation on top of Fabric. This episode connects the keynote themes, the Fabric announcements, and the practical implications for Power BI professionals who care about semantic models, app development, and where AI is actually becoming useful.

News & Announcements

• Microsoft Build 2026: Building agentic apps with Microsoft Fabric and Microsoft Databases — Microsoft’s Build hero post frames Fabric and Microsoft Databases as a unified data and AI platform for agentic applications, with shared business context becoming the key constraint instead of raw model capability. The article highlights Rayfin as a managed backend for app development on Fabric, Replit integration for faster app creation, broader IQ investments, OneLake expansion, and GPU-accelerated analytics in Fabric Data Warehouse. That directly matches the episode’s main theme: AI becomes more valuable when it helps create governed artifacts and applications, not just answer isolated questions.

• Orchestration in Fabric Data Factory: Build 2026 recap — Fabric Data Factory is pushing pipelines beyond ETL into orchestration of people and process, including a new Approval activity in preview that pauses execution for human sign-off before continuing, branching, or failing. The recap also emphasizes easier promotion across environments with variable libraries and a simpler Azure Data Factory migration path through a workspace-level migrate experience and assessment report. It’s a strong example of the larger Build message Mike and Tommy keep returning to: the future is not just data movement, but coordinated systems where agents, people, and governed workflows work together.

Main Discussion

Mike opens with a first-hand conference recap from Microsoft Build, where the format felt more developer-focused, faster-paced, and more oriented around live demos, lightning talks, and practical creation. From there, the conversation locks onto the announcements that matter most to the Power BI and Fabric community.

• Build felt like a turning point for Microsoft’s agent story. Mike argues that Fabric and Build are finally showing a clearer vision for how agentic development fits into Microsoft’s ecosystem, especially on Windows and in day-to-day developer workflows.
• Rayfin was the headline announcement for both hosts. They describe it as a major shift because it connects app experiences directly to Fabric-managed backends, allowing developers to build full applications on top of semantic models, SQL databases, lakehouses, and other Fabric assets.
• Replit integration makes the story more concrete. Rather than only talking about AI in theory, the pairing of Rayfin with Replit suggests a practical route from prompt to working app while keeping data, services, and governance inside the Fabric tenant.
• Semantic models remain the heart of the Power BI story. Tommy emphasizes that even if agents are generating more of the interface or app layer, strong measures, relationships, and model design are still what give AI reliable business context.
• The hosts prefer creator and action agents over “chat with your data.” Mike argues that AI is more compelling when it helps build reports, apps, and workflows than when it simply spends tokens answering questions a dashboard could answer faster.
• New report authoring and design skills matter almost as much as Rayfin. Tommy describes testing the newer Power BI report skills with agents to clean up layouts, add pages, align visuals, and polish PBIP/PBIR-based report assets in ways that were previously much harder to automate.
• Fabric IQ still feels underexplained to them. While the broader IQ direction may make sense strategically, both hosts say they still do not see a clear, compelling usage story for Fabric IQ itself in the same way they can already imagine practical value from Rayfin, report skills, or Work IQ.
• OneLake keeps showing up as the foundational layer. They connect shortcuts, catalog improvements, SharePoint and OneDrive access, and broader storage strategy back to one central idea: AI and agent workflows are only as useful as the quality and accessibility of the data foundation underneath them.
• Local and lower-cost agent workflows are becoming more important. Mike sees a strong trend toward running more models and harnesses closer to the machine, both because of rising token costs and because teams need more efficient ways to operationalize agent-based work.

The result is a Build recap that is less about memorizing every product name and more about understanding the direction of travel. Microsoft is moving from isolated copilots toward systems where developers, analysts, and teams can build governed applications and workflows on top of shared data context.

Looking Forward

If this episode is right, the next wave for Power BI professionals is not just better prompting but better architecture. Expect more attention on semantic model quality, OneLake data strategy, Fabric-native backends, and the skills needed to guide agents toward building useful, governed outcomes instead of disposable demos.

Episode Transcript

0:02 Setting 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. Has kings feel the crowd. Explicit measures. Good morning everyone and welcome back to the explicit measures podcast with Tommy and Mike. Hello Tommy. Good morning. morning. Good morning Mike. It is Good see your

0:32 Good morning Mike. It is Good see your face. Oh my goodness. Your audio is a little cutting in and out there a little bit. So you may want to check something on your side. Okay. why excited for Microsoft Build? Build? Oh bu. So today’s topic will be all around Microsoft Build the conference that was just happening out in San Francisco. should be really exciting. There’s a lot of announcements that were being made. a lot of buzz around the community. I was able to have the opportunity to go speak at the conference.

1:02 conference. yeah, yeah, fortunately I had well I initially had offered up a couple sessions to go to build and I got three sessions picked and while I was close to them the conference they actually wanted me to redo one of my sessions because there was so much interest in building things with aentic stuff and talking about MCP servers and how does this fit with PowerBI. So I actually expanded my talks to four sessions while at Microsoft Build. Wow. Wow. The community was super fun. It was a lot of It was very interesting. It was very different than a normal conference, Tommy., you’ve been to

1:32 Tommy., you’ve been to conferences in the past and you you go like into a room and there’s like an hourong session. Everything was more like like was much smaller. They had a couple sessions that were longer that like a 45minut session. They had a number of demo areas where they were doing demos about specific software and code and watching things together in like these demo theaters. but more than that, there was a lot of these things called lightning talks, and they had just a couple rooms around the campus where you would sit down and you would just give like a 15-minute demo or talk, convey a specific thought or idea,

2:04 talk, convey a specific thought or idea, and yeah, it was it was very interesting. It was a lot less typical conferenceesque and it felt like a lot more faster paced quicker information, trying to create connections and and communication and talking conversations, which I thought was really really fun. Was that for everything or just for the fabric stuff? Because I know these builds huge. Yes, it was. But the the conference actually wasn’t that many people there, I don’t think. I think in total they were like at 2, 500 or 3, 000 people

2:35 were like at 2, 500 or 3, 000 people total, but they did a lot more at the AV recording sessions and hosting them online. So more more than that, there was actually a number of people that were having sessions like the more demo focused areas. Those are all recorded and broadcast live. So, I think they had like 150, 000 people registered online and only a couple people in person. So, it was just it was a very different feel., I saw some people on social media talk about the Chain Smokers showing up at the very end,

3:05 Smokers showing up at the very end, which was very interesting. Well, the Chain Smokers, yeah, they’re they’re a band. like a a dance electronic band kind a dance electronic band popular with the younger crowd but of popular with the younger crowd but they have started their own venture capitalist firm to to invest in technology companies. So at the very end they were like we’re investing in this capital firm. were helping companies get started with building technology systems, software, companies. And then the best part was they came out to the stage after the first day and they gave

3:35 stage after the first day and they gave like a really good hour, hour and a half long concert which was absolutely fun. Loved it. I was in the front row yelling my head off, screaming along with the lyrics. It was super fun. So I thought you were going to say that they had started their own fabric app, their own data app. They were not doing fabric apps at at the time. So anyways, it was super fun to watch them. overall the the the I think the the mix of announcements. Satia was on stage. I was a third row back from the stage. Got to an opportunity to watch him. I’ve never

4:05 opportunity to watch him. I’ve never seen Satia that close and in person at a conference like that. So that was kind conference like that. So that was cool. but the announcements were of cool. but the announcements were I thought they were actually pretty substantial. And I think throughout the most of the conference I was like oh you most of the conference I was like oh I’m waiting for the the PowerBI or know I’m waiting for the the PowerBI or or fabric data related things Rayfin was announced. They had a really good demo around Rayfin and how that works. and they combined the Rayfin and the new co-pilot app experience together as part of a demo in the main the main

4:35 part of a demo in the main the main session the main keynote which I thought was actually really interesting. So overall it was I thought it was I’ve only been to two build conferences. I I can’t speak to any more of them, but I thought it was by far of the two that I went to, it was by far the most fun event to go to. Nice. Nice. And, how did it compare to Ignite? Really interesting., Ignite when we Tommy, you and I went to Ignite in Chicago. This is like what, two years ago? We didn’t go to the Ignite event this last year, so I can’t speak to that

5:06 this last year, so I can’t speak to that one, but I can speak to the one two years ago we went to in Chicago., Ignite felt a lot more just all about co-pilot and it didn’t really feel like it had energy as much energy around the experience as it did as this conference did. This this conference seemed to have a lot more energy around it it and for me it was I was extremely excited around the announcement of Ray apps apps and fabric that was very exciting. There is a lot of excitement

5:36 exciting. There is a lot of excitement around Microsoft integrating with agents and things. So couple announcements around openclaw now natively host. I saw that that is yeah is that available today for people right now right now and I think it yeah but they made the announcement that stuff is coming and they have this thing called scout which was their they call it internally I think they called it claw pilot., so it’s basically allowing you to have an agent that lives inside your environment, gives you more access to more things. To be honest, Tommy, like up until this

6:07 be honest, Tommy, like up until this moment,, Microsoft Fabric in Fabcon Atlanta really started changing the momentum, I think, for agentic stuff that you can build with PowerBI. That was to me that was a momentum shift. Before that conference, there wasn’t a lot of talk about building things with agents and doing things and having AI help you build and create content. So, for one, that was a newer conversation which got me excited because I really think this AI wave is really going to change things. And I started this in back in January. So, I was like, where’s

6:37 back in January. So, I was like, where’s Microsoft’s story in this? What are they doing? What are they making? How is this going to get better? So at build it sound like they were doubling down on this agent and agents framework and open claw and like hey you need an agent that sits on your machine and can like do things actually like Windows computer too. Yeah for Windows like it was a lot of this has been like all Mac related kind this has been like all Mac related stuff of stuff or Linux. Yeah. Yeah. Or Linux. So I actually have a an old Windows computer that I turned into Linux and that’s what I run my openclaw on. So now I’m seriously debating like

7:09 on. So now I’m seriously debating like okay let’s just ditch that. Let’s go put it in something else. But this this ability to be able to have some combination like so I’m going to say some words here that are just making my head spin here. the harness of OpenClaw or the harness of VS Code or the harness of GitHub Copilot, those harnesses are able to run on your machine and then you can now step away and talk to those

7:39 can now step away and talk to those harnesses with your phone or your app and do things more remotely. So more and more of your workload is shifting away from like being actually physically at the computer and you can start saying words to it. You can start talking to it. Tommy, you were saying the death of the keyboard is in your horizon. You’re doing more and more talking to your computer. We’re at high noon with that right now horizon. horizon. Okay. So, it’s it’s like it’s like this is already becoming part of your workflow. So, I think there’s a lot of push into that new how do we rethink what computers are

8:10 how do we rethink what computers are doing? and there was a a pretty impressive announcement around AMD, sorry, Nvidia and the ARM processors also came out as well. So in general like the the excitement I saw felt a lot about like Microsoft was adopting what the community has built. I adopting what the community has built. Open Claw’s only been out since mean Open Claw’s only been out since November of 2025. Not even a year old yet and already you’re starting to see Windows computers show up with like a sandboxing environment for OpenClaw or Hermes or these agents on your computer

8:41 Hermes or these agents on your computer like like they’re moving quick it feels like. Well, the Nidia announcement was big because I’m reading a book about the founding of Enthropic and the founder of it in all his journey he was at Google Brain Open AI and he realized which is crazy like if you just scale the GPUs the computer will want to learn and he’s like and that’s just the question here is like how much can we scale it and it’s going to learn. So yeah, that announcement is large. But Mike, I I

9:11 announcement is large. But Mike, I I know there’s a ton of information. You mentioned Ray Finn. I know we’re we’re actually going to get into an entire episode on that. I want to talk about it today. today. I thought it would be a good place to start from Aun’s hero article. talking just a little about the theme. And every conference if you have not caught on yet, Aun always puts out what’s called a hero article. the main theme, the purpose of where Microsoft Fabric is heading. It’s like a state of the union for Microsoft Fabric. It’s like the it’s like the summary of

9:41 It’s like the it’s like the summary of like the biggest announcements that

9:43 like the biggest announcements that Microsoft cares the most about right now. Yeah. Where we’re at, what we’re doing, where we want to go,, is probably the best way to put it. So, I would like to read from this because I think it’s a good way to think of, okay, what is Microsoft focusing on for fabric? So AI is driving a fundamental shift in how work gets done and how applications are built. As the 2026 Microsoft work trend index report highlights, a growing share of workers are moving beyond asking questions to handle to handing off entire tasks and orchestrating multi-

10:14 entire tasks and orchestrating multi- aent systems. This shift introduces a new constraint. The challenge is no longer model capability but consistent shared data context across the business. That’s the the challenge we’re solving with Microsoft Pavoc. It provides a unified data and AI platform that empowers you to bring together data and move from isolated AI experiences to production ready agent systems in which new in which each new agent builds on shared organizational context. This vision is already driving strong

10:44 vision is already driving strong momentum across millions of developers building on Microsoft fabric in Microsoft databases. We’re extending this foundation with a bunch of new capabilities from de developers moving from prototypes to productions faster. And with that, I just want just touch on this. this. Notice that none of that really did not say co-pilot. Co-pilot’s obviously a big part of this, but I thought it was very intriguing on when you looked at when AI was first becoming a large component of

11:16 was first becoming a large component of our world. I think there was still a lot of direction on what does AI mean for workers, what does AI mean for developers and especially in the context of PowerBI. fabric and AI have always been interwoven really from the beginning because they both jumped up at the same time around 2023 and we’re really seeing now I think a really strong vision on what does it mean to work with AI and Microsoft

11:47 mean to work with AI and Microsoft fabric it’s changing very rapidly there’s a lot of new movements here the two main things that I think you agents or agentic building. So, one of the stories here I’m continually repeating this over and over again and I think Microsoft is finally taking note and changing how they build with agents. So a lot of times it up until this moment Tommy it feels like a lot of what Microsoft has been doing which is let’s

12:17 Microsoft has been doing which is let’s throw agents and AI at the data itself and have a system around you ask questions it writes SQL or DAX and then it pops out an answer to you and I look at this going the beginning this is a highly inefficient way of doing this. I don’t want to spend tokens every single time I went to ask what was my sales this year? What were sales last year versus this year? That’s not helpful. And honestly, those are things that are can be answered with regular regular reports or regular visuals. So,

12:47 regular reports or regular visuals. So, you may have that question, but the goal of that question should be not to have the agent respond to you with an answer. It the goal of that question should be helping the agent explore with you what visual should be built to resolve that answer. And so, I want to introduce this concept of the creator agent, right? Agents that help you create. I really feel like that’s what OpenClaw and Hermes and all these other agents are doing now, right? They’re they’re taking input from you and helping you generate, build files or

13:19 helping you generate, build files or structure or think or reason about something. And the output of that reasoning is a a physical artifact, right? I find the most valuable part around agents right now is I need to build some software. I want to build an app. I want to make modify something. I’m going to spend the time and the money to tell the agent what the requirements are and have the agent create something, right? right? So, I think that’s to me that’s the really exciting part of this is the creator agent. The creator side of things is what I want to see. I want to touch on what you said because

13:51 I want to touch on what you said because I think this is really important and it’s one of the things I’m very happy to see with Microsoft. Mike, logically AI and asking questions about your data makes a lot of sense, doesn’t it? If you were to say when AI came out, oh, you can ask any question, get any analytical answer, answer, it makes a lot of sense and you see why Microsoft pushed that initially, but then you start using you’re going I have a dashboard that’s a little faster that’s going to provide the answer and you realize that there’s for whatever reason for the human mind or the human behavior just there’s a disconnect.

14:23 behavior just there’s a disconnect. I am getting more and more from organizations how can we make AI work for us and no one’s really looking for a chatbot around their data you it is the action you talk creator agent and I’m thinking about this also from a large organization too it’s the action agent is the other side of that coin I I would imagine to me those are two different types of agentic solutions there are going to be actions that people want to take off their existing data and that’s not necessarily just asking to your point what were my sales

14:54 asking to your point what were my sales Q2 it’s going to be providing the task in the organization not necessarily building apps is going to be part of that but it’s allowing the hard data the databases the tables interacting with the soft data the files the shareepoint to be able to actually provide actions to find where we can automate to find things where human tedious effort can be but also to really enhance just the workflow of people and I think we’reing finally seeing that to your point from a

15:25 finally seeing that to your point from a developer point of view we’re finding now a different capability where where I may not need a single data point from AI but I need a custom I need a different layout based on the situations that we have and really what we’re just opening up is this open playground both from the action and the creator point of view of what is the solution we are trying to do what are the Oh, let me back up. What are the problems that we have? What’s a perfect

15:56 problems that we have? What’s a perfect solution for us? And we’re actually finding ways that we can really accomplish that quicker. And I I think the gap between that our work is not changing, Tommy. We still have these problem these gaps between we have a bunch of data and we need answers about that information. What we’ve been given now is more tooling at our disposal. So, Rayen is made up of like two parts. Yeah. Yeah, part one of Rayfin is you Yeah. Yeah, part one of Rayfin is Rayfin is basically a backend for

16:26 know Rayfin is basically a backend for fabric right it’llo it’ll host secure connection between an app side and it will host the the backend side which is semantic model SQL server or lakehouse like that that’s the story there so let me be perfectly clear I’ve been building apps like full-blown production like apps on fabric where the front end is the application that business users or the company uses and the back end is SQL database or something in fabric or lakehouse, right? I like this pattern

16:56 lakehouse, right? I like this pattern because then the data lives exactly inside inside my fabric environment and I can go use it. So Tommy, the amount of friction that companies produce around like like just say Salesforce, Oracle, like whatever data system that you have, there’s a lot of friction behind that that makes that system difficult to get data out of, right? It’s expensive to move data out. And And I really like this idea of keeping the data inside fabric because then it’s

17:26 data inside fabric because then it’s right next to all my analytical reporting. It has all the things that I need to move the data around. I can get I can modify databases with notebooks and schedules and things like that. So I I think this is a really neat solution. The downside is if you build just pure apps on top of fabric, you need app registrations and they go out of you registrations and they go out of they eventually time out and know they eventually time out and yes how to connect to the database the all the things. Yes. because I’m,, that the SQL database is doing something, but then I want reporting, so I need the SQL database to mirror down to a a lakehouse and get a semantic model. Like there’s a lot of things that coordinate and keep together. When you have all

17:58 and keep together. When you have all that just built into fabric, I think it simplifies a lot of the experience. And so,, so,, the part of Rafen that is interesting is the whole managed backend piece, which is really slick. One note that was made at the at the conference that I want to just maybe mention here while we’re in here as well. the note from Aroon here was a very it’s a oneliner but I think this is going to be extremely impactful. They said, “We’re excited to partner with Replet, a leading AI coding

18:28 partner with Replet, a leading AI coding platform, to help customers build enterprise enter enterprisegrade apps and interfaces they know and love while keeping the data and app and services managed in their own fabric tenant.” Tommy, this was not highlighted as much at the conference. I went over and talked to the Replet team at their booth in the conference and just said, “Dude, you’re great. Congratulations on the announcement. I absolutely love what you’re doing. Give me a bit more insight. What’s going on here? How many? They showed me a couple prompts they were using to go build dashboards or reports and things right

19:00 dashboards or reports and things right from fabric items. It was as simple as, hey, I want to connect to this semantic model. Here’s the link, the URL. URL. And it built stuff that in iterated with Replet. And so Replet has been around for a while and you can build these great coding platforms. you can have access to all the files and things, but you can also not use all that and instead just allow you you can just

19:26 instead just allow you you can just talk to your agent and then have it handshake between the replet side and the PowerBI side. I said, well, what does deployment look like? And they said, we’re working on that. There’s still work coming. They they don’t they’re not ready to really show like how the publishing part is, but they’re they’re basically to the point of it it sounds as simple as once you have the replet application running, they will let you publish right from there directly into fabric. So I’m very excited for this integration to get out

19:56 excited for this integration to get out into the world. I think this is going to be really impactful of how Replet’s going to really change how we build things. And already, Tommy, the community is a buzz about these things. Yeah, actually looking at the buzz, Reddit’s on Reddit called it the best announcement since the release of PowerBI desktop. That’s my statement. That was your statement. That’s my statement. Like I didn’t found you. you. Well, I I’ve been saying this since like I said that with Reza. I said this other people. That’s my statement. Like I’m saying like I think this is you’re not the only one who

20:26 you’re not the only one who this is the well I’m I’m trying to see this into people’s mind because I think this is like this is the largest announcement that we’ve had since desktop period 100 and I I honestly completely agree and the thing I love and what you mentioned with the replet integration is the semantic model thing and I need think we need to really hone we’re a PowerBI at heart podcast and this just holds That’s true. This is another seinal moment for the semantic

20:56 another seinal moment for the semantic model. How many times have we heard in the last year or two that the semantic model is the best thing for AI? Which we may agree with, but I don’t think we’ve ever seen Mike yet a good practice of that, right? The actual application of that. This goes back to our water cooler. In theory, it makes sense, but then when you say, “Show me,” we’re like, “I don’t really have anything.” this the fact that we can now with Rafe and connect to a semantic model not necessarily a database it holds true you still need the skills of

21:28 holds true you still need the skills of a good semantic model measures relationships all those things are the heart of what AI is going to be building on whether you’re doing data apps you’re using co-pilot or you’re still building a dashboard the heart of fabric is the semantic model model you may call it one lake you may call it the lakehouse but at the end of the day the heartbeat is going to be the semantic model that’s what runs everything that’s going to run also your actions here that’s what is going to run it’s your source of truth and it’s the key so I think this is really essential

21:59 key so I think this is really essential that we are moving to a place now Mike where I do not have to design a dashboard again there are pros and cons for this and I think it’s important to say when is you’re just relying on prompting you have less control over your own customization if you’re willing to do that that’s fine but at the end of the day you still need a solid semantic model. So I think this is really essential to even when we’re moving into this new place, the semantic model still is going to be the the key

22:30 model still is going to be the the key the the heart of what we’re doing. But no, I think Rafe and like I said, we when we were messaging about this offline, we’re like, how much do we want to talk about it today? How much do we want to do? I I think we’re gonna have a lot of I think we may have a series on this. I I really just from the announcement itself, you’re seeing what this goes on the t-shirt, right? This goes on the game changer t-shirt. I think we could both say that. and I don’t think it’s too early to say that. But Mike with Rathan, what are your immediate I guess

23:01 what are your immediate I guess impressions? Not just the developer review, but I would like to hear a different angle. How do you think this is going to impact businesses? How do you think this is going to be impact organizations? any issue you had with PowerBI desktop any issue like visual can’t have multiple data sets visual can’t do XYZ thing visual doesn’t have I I built a table has limitations on it whatever whatever the whatever the thing is whatever the hesitation you had was

23:31 is whatever the hesitation you had was you now have an an immediate action between an agent that has ultimate knowledge about everything that is code related like D3. js JS building tables tan stack tables like any any thing that you want to build if you can dream it now it can now be part of the application. So like this is full-blown apps on top of fabric. So you full-blown apps on top of fabric. So why is this so revolutionary? know why is this so revolutionary? If you can dream up the experience that you want to see inside

24:02 you want to see inside data with your application you can now do it. Already I’m seeing things on LinkedIn and Facebook about Rayfen where they’re saying look I made this Zork game and I made this other game and I’m building like full you you could do that right now. you could you I think you can physically run there’s there is definitely some limitations right now, right? So it’s going to it feels like there’s needs to be a couple more features added to it to help it become better. But I I see the the heavy potential of this. And I also argue Tommy we talk about developers

24:35 Tommy we talk about developers people that are writing a lot of code right right when you add the flavor of replet in here here I really think we move away from developers we move away from create to towards creators if you can envision it right everyone has worked and used a website page before you have visions for how things should work and look that’s all you now need to get it to do what you want the vision vision what is what is required. So if anything this just opens up the ability for us to

25:06 this just opens up the ability for us to solve more problems with more unique solutions moving forward. so that so there’s I was earlier I was talking about there’s two parts of Rafin. One was like the back end of infrastructure. The second part that I actually didn’t touch on was the second part of Rayan is there’s a whole bunch of skills around report building now and you can now go see them., if you go Microsoft or is it Microsoft fabric skills fabric skills for fabric? Yeah, skills for fabric. Microsoft skills- for yeah skills for fabric. So now if you look at

25:36 skills for fabric. So now if you look at skills for fabric on GitHub, it’s a collection of skills and MCP servers and things that are there. If you go look specifically at the skills area, you’ll notice there’s now PowerBI report authoring, PowerBI report design, PowerBI report management, PowerBI report planning. These are four skills that exist. Now the report authoring is actually really interesting because it’s actually talking about you’re going to create a PowerBI report that modifies the PBI and PBIP format. So these are skills that are wrapped around the context of what you need

26:07 around the context of what you need to build and inside the report building skill. You have all these other things like authoring, card, cartitionian stuff, images, maps, visuals, visuals, page formatting, like all the information you would need to describe how to build a report that is now included inside the skill. Now, this is quite a lot of information in here. So, yeah, let’s touch on the the new ones here because they’ve already had the semantic. Yeah. Go ahead. Go ahead. But when you pair that and what’s happening with Rayfin, like that’s the combination of the two things, right?

26:38 combination of the two things, right? How do you create with an agent? and they’re they’re trying to formulate what are the set of skills, what can we build, what what does this look like in this new agentic world. I I want to make sure I touch that too. You don’t have to use a rafin to use those skills, especially the report design. Like so actually just this week I I’ve been testing out those new report design skills and and this is just continued enhancement to the workflow. Yeah. Yeah. So I’ll give you an example in cloud. I have the MCP. I’m like let’s let’s go to the measures. let’s verify a few

27:09 to the measures. let’s verify a few things. And I said, and now I want to use these skills. So we’re like, all right, I’ll tell you when I close the report. Close the report. It goes in my pip. I tell it where the folder is. And I was like, I want to I just want to test something like let’s polish it up. Let’s do alignment. Let’s add a new page with these measures that were now created with in the same format as this page. Let’s add some titles. Let’s clean it up. And that is completely new. We’ve we’ve talked about this I think since AI was a

27:39 talked about this I think since AI was a thing about the lack of design in the report side and now well here we are Mike. it there are still things I would like to see improved but this is awesome. I I think this is such a critical moment too. I don’t I don’t know how much farther down the road it is than Ray Finn in terms of what kind is than Ray Finn in terms of what impact this is going to have. If Ray of impact this is going to have. If Ray Finn’s here from an impact point of view, how much like how many levels down is the skills? And I don’t think it’s

28:09 is the skills? And I don’t think it’s much. much. It’s a close game. Yeah. I wouldn’t even put it on the same level., I I Yeah, Yeah, I wouldn’t I would put Rafe and the Fabric report authoring skills on the same level in my opinion. Like that’s what I agree because I think they’re they’re so intertwined. Like yes, the skills that you’re talking about here and the ones that are on the GitHub page are a little bit more focused on the PBIR and PBIP format, but there is things that you could extract from that that would help agents build things. And, the other thing too is these skills are like fluid,

28:39 these skills are like fluid, right? It’s very easy to use these skills and then go back to the agent and say, “Hey, I need you to build extra stuff or more things or change stuff.” You can always adapt and modify the skills for how you work, what you like to build, what looks good for your eyes. So I think there’s there’s a lot of opportunity there because the skills are highly flexible when working with agents and agents understand how to read them and as well as know how to use them. So I think for that respect I think a lot

29:05 I think for that respect I think a lot of that is makes a ton of sense. One thing I would I maybe would echo here is I’ve observed this on the Microsoft skills library. So the skills for fabric I’m not sure why they’re doing it this way. There’s a whole bunch of skills that they’re producing, but a lot of them are only markdown files. The skill itself is marked down and then all the references is in there as well. And I would argue there may be some usefulness about

29:35 there may be some usefulness about building some scripts that make regular output happen directly inside here. So, I’m not sure why everything in this skills area is only markdown. Maybe they’re trying to make it really repeatable., but I would really like to see additional files other than just pure pure assets and references. So there there should be like some base JSON files like so in the power in the PowerB report design you have a base asset called base JSON and it’s basically like the the basic

30:05 and it’s basically like the the basic JSON that they’re using to be able to modify and build skills or still from there. So there. So I don’t know it just it feels a bit heavy-handed and I feel like it’s going to be a bit abusive on your token usage. So, a combination of skills and scripts I think is a bit more efficient, but I’m not sure how well that applies in reporting. So, I don’t know. I think needs to be flushed out a bit more. One thing I wanted to touch on that too, I I think Kurt mentioned this in one of his many amazing posts that the fact

30:36 his many amazing posts that the fact that when you get a skill, you should not try to avoid using it just out of the box. There’s always some customization you can do. And I’m trying to right now I’m actually testing combining that. Someone else mentioned that HTML actually works better when it comes to for skills. So they’ll provide scripts that are HTML because that can actually be something visual the the computer can render or the AI can render. So is we’re still early on but I render. So is we’re still early on but this is I think right now Mike the mean this is I think right now Mike the big the big point I want to make is the

31:06 big the big point I want to make is the fact that we can now the fact that this is now possible. So it’s going to open up a lot. Mike, before we get to the database, I need I need people to know, as much as we love all the announcements, you and I keep it real. We keep it 100. And there’s something here I’m a bit disappointed about when it came to the fabric announcements. Fabric IQ is now GA. What is Fabric IQ? What is Fabric IQ? Right. Like, yeah, exactly. Like, so what? Okay. So,

31:36 exactly. Like, so what? Okay. So, ontologies, the operation agent. I I’m a little I’m a little concerned about this because I still don’t know what the end product is of a fabric IQ or an ontology, right? You can create an ontology. Is it just for the operations? Is it just for planning? I have not seen buzz about this. I have not seen people using this. I’ve not seen a lot of interest. I’ve barely seen a lot of articles on Microsoft. They still have six articles on their website on the docs about how to set up the entire

32:07 docs about how to set up the entire fabric IQ, not all the things you can create with this. So great, it’s it doesn’t have bugs in it, but it doesn’t feel fleshed out. Like I still don’t know like Yeah, to your point, what’s GA? What is GA? So yeah, Tommy, this is one area that I’ve really struggled to understand. What is the value proposition for this tooling? So, Fabric IQ is just missing the mark for me. I’m not really I don’t really understand why it’s there. I do understand why data agents exist. Data agents do make sense. I think that

32:38 agents do make sense. I think that one makes a lot of sense to me, right? operations agents that they’re focusing on these other things real time. Yeah, I again maybe that’s okay. Hey, it it feels like a mix of like, you it it feels like a mix of like,, hey, why do we need an agent know, hey, why do we need an agent involved to reason about stuff when you can just program it in to your real-time pipeline using just code? So, I don’t know, Tommy. The fabric IQ feels to me I’m not that that one announcement is

33:08 I’m not that that one announcement is like, okay, fine, it’s G. I’m not using it. I’m not building ontologies. I’m not using the graph element. It doesn’t seem to do what I want to do. it seems expensive to run. In general, the whole fabric IQ to me is just a miss. You’re right, Tom. That was a very underwhelming announcement for me., Fabric IQ is not something that I’m looking to go really jump into. I’m going to wait and see., I think that one may go by way of the metric sets a little bit because it’s I don’t You’re right. No one’s talking about it and there’s no clear story to it.

33:39 there’s no clear story to it. Microsoft may have a vision for what it looks like. A great way to put it. Yeah. But what’s what’s the story? What are the action items? Hey, hey, if you use Fabric IQ, you can get a 10% or 30% reduction on your tokens. Great. That’s a story I have that tells me use this, get value from it, right? use your ontology to do XYZ things. Great. I’m not seeing those stories coming from Microsoft. I’m not seeing those come out the out the door. So I just don’t feel like that project or that part of the

34:09 like that project or that part of the team is just really getting getting the right functionality features of it. It’s just not very I don’t love it. I also think now Tommy if there are missing gaps in the product Rayfin feels a lot more useful here. Yes 100%. So if I needed something like an ontology or if I needed to go build relationships between things, I feel like the Rafen app would be a much better place to go vibe code what I want to go build. Like that’s that’s where I would go spend the time and effort to go

34:40 would go spend the time and effort to go build it and then I get what our company needs and not some of this like canned out the box thing that Microsoft provides. It it’s funny, Mike, because I feel like the Fabric IQ story is like directions. It’s like take a left and a right and when you take that first left, you’re in a culde-sac. You’re like, is this the right way? Yeah. You just like I guess run into a dead end street. Right. Exactly. So I agree and I to your point like for a developer first you got to ask the question how is this valuable to the team or the business? What’s the value ad? Y the second question is for especially

35:11 the second question is for especially new things is why is it worth it for me to learn this? Why is it worth it to me to spend the time to put this in? And I think both of those questions I’ve not been able to answer or better yet have a better a good answer to that right the configuration for it’s like okay once I’ve spent the time to build an ontology what’s the value out of here so I just wanted to just make that note hopefully we see more hopefully Microsoft emails us angrily and says how could you say that about it and then we’ll talk about it with them just to have an idea

35:42 it with them just to have an idea so that’s all I really wanted to say about fabric hubies I think Mike that’s all there is to because they didn’t they didn’t announce anything about Fabric IQ outside that it was G. They didn’t say by the way we’re adding feature X Y and Z. It was simply guess what? No bugs. So So I I think more than that, Tommy, so there’s there’s this general work IQ element. element. Yeah. Right. And I think that is actually more impactful to me than the fabric IQ piece

36:12 impactful to me than the fabric IQ piece of this from what I see right now. So when I look at what IQ things Microsoft is producing a lot of my data for meetings or calendar or teams meetings and transcripts that’s the valuable stuff if I think about my workflow dayto day Tommy we’re on teams a lot we’re doing communication we’re doing teaching and training lessons and things like that so we’re spending a lot of effort talking to people to people a lot of that information needs to be

36:42 a lot of that information needs to be distilled down and put into a place where my agent can go use that information. So the work IQ opens up more of that Microsoft ecosystem giving direct access to VS Code, my agents, that to me feels a lot more useful. And so the work IQ announcements or just the idea that that’s now a bit more front and center. I find a lot more value from that work IQ than I

37:12 lot more value from that work IQ than I do the fabric IQ. I I completely agree. I think the work IQ is much more fleshed out. You see the story there. So don’t want to completely poop say everything’s wrong with it, but I think to your point what’s the story? And I don’t think the story is completely fleshed out. So, and and that’s why I’m like like leaning on like I want I don’t want to say all the IQ stuff is like not useful. I want to focus on like the Microsoft work IQ stuff seems to be fairly useful. Integrating that with my workflows using

37:42 Integrating that with my workflows using that that how to apply it. Yes. using that with my harnesses and agents makes a lot more sense because I can I can draw together disparate things and I can actually build a pattern I’m seeing evolve Tommy I think you are too which is how do we use Microsoft Microsoft products agents to build like a workflow right right right hey I just had a meeting with a customer customer I there is a series of actions I need to do after that meeting summarize some notes make some action

38:12 summarize some notes make some action items go do some things, make some tasks from those tasks go accomplish something. So from those sequence of items, what is the system that you put in place between you the agent and then the Microsoft data or products or pieces like what does that look like? How does how does that get help get work done? And I think that’s where this open claw integration with Microsoft is coming from. I think that’s where individual like that’s not that’s that’s

38:43 individual like that’s not that’s that’s not a team of people that’s more for an individual. I don’t know yet Tommy. Right. So

38:49 I don’t know yet Tommy. Right. So as of today if you scale out as you So I’m having this moment where I’m looking at like what agents and things that I’m building are doing. They’re just being added into my company as if they’re like another engineer, right? that that pairs up with my my principal engineers or my senior engineers to help them do work. So agents are becoming like part of the ecosystem of what we build. As I look across the landscape of what we’re

39:19 across the landscape of what we’re creating and how we’re building things, I have to get my head around where does all this stuff fit? How do I put it all together?, to your point, Tommy, agents right now feel very per person, like an assistant to the person. person. But how do you integrate the agent as part of a team member, a part of the team? What does that look like? How do you have multiple people working on the same bit of code or product or report or data set or how do you get all of

39:49 or data set or how do you get all of that to work together as a system? system? And I think that’s what we’re stepping into now is we’re starting to to use to use the harnesses are getting more creative. Work IQ is adding more access to data. And so it’s becoming part of you’re we’re starting to see the very beginning feeling of agents are becoming part of the system on how we produce work. Does that make sense? sense? 100%. And honestly, Mike, something that I’ve been really,, contemplating myself and I think there’s two different approaches here that I’m

40:19 two different approaches here that I’m finding that are really going to be successful. There is agent as automation and AI as partner here. So let me I I I’ll briefly go into them because they may be different top or or different episodes but I think this is important to how does an organization apply an agent right because successfully impactfully the first one I want to talk to is the automation side is most teams have a workflow and I’m finding more and more that agents are very very successful on operations

40:51 very very successful on operations departments. Let’s say you have POS that come in, purchase orders come in and you want to make sure all the information is filled out. It has a workflow like the like Power Automate agents as like basically the enhanced or the the steroid version of Power Automate is incredible as long as you can identify those task flows to go to different people to actually apply the information fill it out organize that information in the right place. The one I think is going to provide more value in the future is AI as partner.

41:21 value in the future is AI as partner. So, I know I talk about notion a lot, but I think it’s rightfully so because of the integration. I’m working with a team right now in notion on a few of my projects and anyone can comment or mention an agent. I have two a few different agents that they can mention a consulting one a one to organize all the information and meetings and they can mention it anywhere in the notion docs or in a page to say I’m finding that not just myself using it but each person be able to use the

41:53 it but each person be able to use the agent as an assistant on the same content is incredibly impactful. But again, these are very specialized agents to perform specific tasks. But the thing is we can all share and collaborate on that agent together. It’s not a chatbot, what? It’s not like just a chat text box where I ask a question. It’s on somewhere it’s already existing, but it’s custom for my team. It’s custom for that project. Yes. Yes. And these are the really the two avenues I think right now we’re seeing where agents are going to be successful. the automation side but also that partner

42:25 automation side but also that partner side for a team rather than it just existing only in the background or again people still see agents as ask a question get an answer that’s not where the impact is so I don’t know have have you when you think about a team do you see either of those two approaches or have you d do those resonate with you yeah the yeah the yes it does Tommy and what I’m hearing you say is I’m going to rearticulate this into more maybe general terms what right notes not notion is just a harness

42:56 right notes not notion is just a harness around your agents and it’s a great one at that yeah it’s a good one and and Matias and I were talking on the other podcast which is a gentic thinking which was a lot around like 90 95% of the effectiveness of your large language model is built on how good the harness is around what you’re doing doing applying it in the correct place now the thing that comes to mind Tommy when I when I think about your scenario that you’re you’re saying here like multiple people working in notion calling off which agent when you need it to help you do actions or specific

43:26 to help you do actions or specific the what you’re doing is you’re building there’s a a harness that has a framework where you can draw in the large language model give it specific information and it is able to reason about something or do some reasoning work on its own for you with inside in lie of your process. So that’s identifying where those events need to occur, identifying how that content should be

43:57 that content should be handled, given to the agent back, organiz like all these things. You’re building a system. Yeah. Yeah. And you’re just plugging in the team of people and the agents where they need to be. The my only hesitation is how do I know like if I’m a new person to your system, Tommy, like how do I know which agents can do what?, how do I know which custom agent I should be using where inside this notebook? How do I know what features are available? So, to some degree, there’s a little bit of like it’s the same thing as like when you would introduce a new employee to your company, company, right? right? Here’s what we do. Here’s how we work.

44:27 Here’s what we do. Here’s how we work. How do we onboard you into this process that you build? But I think at the end of the day is it’s really impactful that we can build these things. Notion is one of these items that I’m like, man, if Microsoft would just fix Loop a little better and like make Loop more like a Notion thing. There’s a lot of things that are loop that it’s trying to do that just doesn’t work for me. It doesn’t doesn’t feel like like there’s a there’s an idea of like the I don’t know if it’s a notion, but I know there’s other ones that are doing like your second brain, right? So, you can you can tag things

44:57 you can tag things first. Yeah. Yeah. You can have this basically visual graph of like information that’s linked together and you can go to this together. You can go to this graph and see what articles and and posts and things. things. I found an open- source version of this that is like Obsidian. So, Obsidian, I think, is the first company that made it that I’m aware of. There’s this other company called Log Sequence. L O G SQ open source version, but I really like it. It’s very similar in a lot of ways. It doesn’t have any agents attached to it. So,

45:27 agents attached to it. So, no, no, there’s there’s something here that needs to be refined a bit more. Like, how do you put all this data together? How do you make these notebooks and things work really well? I I think we have an episode, but someone asked the same question. So, I will save all that for another episode, but I know we got a a few more things to get through for at least the fabric updates. Mike, the other big announcement was around databases, and I found a few things interesting. We talked about the semantic model as being the

45:59 semantic model as being the core, the heartbeat, right? But we do also know when you’re trying to do more autonomous things with an agent, a semantic model is probably not the best case, especially when you’re dealing with loads loads of data, but again, not necessarily too action based when it’s more autonomous. Well, a big part here too and I thought is going to be pretty big because this goes back to the scaling idea here is fabric databases can now be GPU accelerated. Yeah, Yeah, you can actually have your data

46:29 you can actually have your data warehouse accelerated with GPUs. Why is that important? Well, first off, your AI AI runs on GPUs. Graphic process units created by Nivid Nvidia. Nvidia. Thank you, sir. So and the fact that now the data warehouse can be run off that along with what do they announce the Horizon DB Horizon DB which is a very mixed database with has integrated AI model management direct connectivity to

46:59 management direct connectivity to Microsoft fabric vector search and we’re finding now the enterprise versions of how an agent can run and really be train your own model Mhm. Probably no better way to say that. So, what did you hear from the conference? Did you go any sessions on this on how it’s being applied or why is this so important to you? Yeah, I didn’t actually spend a lot of time on this one, Tommy. This is one area I focused a lot of my effort again, I was very busy. I had four sessions myself, so it was pretty much training.

47:30 myself, so it was pretty much training. You had a talk. I was doing a lot of training stuff. So,, I didn’t get a lot of time to spend on the NVIDIA side of things, but I guess maybe from the keynote and the couple things that I was hearing was there’s a there’s this movement from Microsoft to move some of the large language model inferencing to your local machine. machine. Open source models are really interesting. Custom weights and models are really interesting., hosted agents are very interesting. So, I think what’s happening, Tommy, is there’s this need to have the ability to host

48:02 need to have the ability to host reasonably sized models on your machine, right? right? That way, you’re not spending tokens in the cloud. And so, I think, yeah, yeah, I’ve been saying this for a while now in the agent space, you agents can be used to loop through problems or issues. There’s there’s now a new language or or or thought that’s coming out on the internet, which is you should never prompt your agent anymore. You should just make loops that prompt your agents to do things that you want

48:29 your agents to do things that you want it to build. That just means you need a lot of time for it to reason through, build stuff, create things, find problems, resolve problems, change it, rebuild, like there’s a lot of things the agents need to do in order to make this possible, right? right? If you’re going to spend money to have that run inside Anthropic or Microsoft, that’s going to be very expensive. So it felt to me that Microsoft themselves was saying we want to get in to more of the hardware game like what Nvidia is doing. Nvidia is doing like we build hardware.

49:00 Nvidia is doing like we build hardware. We run inferencing stuff on our hardware. hardware. I think what’s happening now is people are trying to get to the place where I can buy some desktop hardware that can regularly run models over and over and over again and still get reasonable amount of value from this one. I personally feel like we’re in this phase now that okay, we’ve proven that agents are useful. We have large language models and harnesses that are working fairly well with us. The next wave of things is how do we optimize it? How do we reduce our cost? What does that look like? Do is it a mix of combination of

49:31 like? Do is it a mix of combination of hardware and like the price to run tokens is getting so expensive especially with GitHub raising their prices. prices. Yeah. Yeah. We need to think of other ways of using agents locally on our machines to make it make us more efficient. Yeah. So, I think this is the new the new game that we’re playing right now, which is how can we bring big models agents and do more of them locally on your computer. Listen, I I keep mentioning notion because I hope one day one of them’s listening to go, hey, we’re getting free advertising, free tokens because yeah,

50:02 advertising, free tokens because yeah, the to your point token management and I know you Matis have talked about a lot which I think it’s becoming more and more relevant is that’s going to be a skill is how do you manage all that?, one other thing I want to touch on, I think that was a major point here was not just the AI side, but was really Microsoft focus on one lake., both from a governance point of view, but they’re how they’re expanding it. I have seen this more and more I think the last two years on this concept

50:32 think the last two years on this concept of hard data and soft data for your Microsoft estate where hard data is your tables your databases your semantic models but what’s also essential for an organization to work is that what called soft data your Excel files your SharePoint information the word docs that all contain valuable information it may not be structured as in a tabular format. format. Yeah. Yeah. But if you want an agent to really be effective in a team, it needs to be able to access those files and also work on

51:02 to access those files and also work on top of them. And we’ve seen co-pilot for SharePoint. Well, and we’ve seen,,, agents around fabric now, but there’s still been a disconnect for each. Well, we actually now have more shortcuts. We actually have a shortcut to SharePoint in one drive that is now GA. So, you can create shortcuts directly from fabric data warehouses. there’s also link for Azure private link link and now one lake catalog is available in GA in Microsoft foundry. So this is so

51:35 GA in Microsoft foundry. So this is so essential Mike because one lake we initially saw when it first got announced back in the day of being the central platform but now we’re seeing how this is going to be the key for making anything around agents work. I think this is going to be much more impactful than people believe. I I think this is going to be definitely the way people are going to want to move forward with lake houses. I I keep saying this and people I think I think it resonates. Storage is cheap.

52:05 think it resonates. Storage is cheap. You you can put a lot of things in storage that are extremely cheap and easy to use. I really like that capability of having that cheap storage in place for me to to do the work I need to do every day. So the storage layer having Microsoft integrate that better better you’re going to spend time getting data into fabric you’re going to want to put in the lake. There’s really good sticking power. I think once you move data to the lake and get data in the system you’re not going

52:36 get data in the system you’re not going to want to leave. So I think it’s very important that Microsoft make it as easy as possible to bring data to your lakehouse and get it in there into the system which makes it a lot simpler to move through going to be mission it’s going to be critical would you call actually Mike one lake right now if you’re consulting a company or guiding them on they want to do some they want to have agentic solutions they want to build AI and they don’t know what that means crit mission critical critical is building in one lake. Would you agree

53:08 is building in one lake. Would you agree with that or disagree as like the foundation at this point?, I think I think it’s the main part of your story. Honestly, I think it’s if you’re talking about the story and what what we’re going to use and and be make use out of things, that is the main part. Yeah. Yeah. Everything comes to the lake. and everything goes into fabric at some level. Now, there’s all these like,, who controls what, where does the data go, what does a workspace design look like, how many lake houses do you have, what lakeouses are there, all kinds of things. in the lake, but you got to get it to the lakehouse.

53:38 but you got to get it to the lakehouse. I think that’s that’s the main story here is just bringing your data to lake and now lakees come with like hot storage, cold storage, warm storage. Like you can actually like start archiving things and putting it in cheaper storage places so you can hang on to a lot more of your data for very cheap and easy to use things. I think that’s a huge story for Microsoft and the fact that it integrates so well with all their other products that’s really useful I think and so yeah I it’s it’s definitely the main theme of that story and then you figure out okay from there

54:08 and then you figure out okay from there what do you build with it right do you build apps on top of it do you build reports reports you can’t do that until it’s in one lake correct everything has to get there first yeah I would 100% agree with that all right so unfortunately I know I think we’re going to keep touching a lot of these things, but Mike, after going to after going to Build, after seeing what their announcements were and now having a little more time to digest and bake on it, what’s your big takeaway and like what are you going to take moving forward from this?

54:40 take moving forward from this? My main takeaways are probably Microsoft is really stepping into the agent world. personal agents like OpenClaw or Hermes or running them on Windows is is coming first and and foremost to the front and center, you’re going to want to learn how to do that stuff. It’s going to be the new way to work moving forward. There’s going to be agents here. I think in general trendwise, there’s a general trend of agents will be used more on your local machine. Using models

55:11 more on your local machine. Using models on your local machine is going to be a trend moving forward, right? that that’ll that’ll you’re going to buy a new wave of computers that are going to use agents on your machine. That will make it super having an agent that runs on your machine that cost you only electricity on your machine. That’s a huge sticking point. Yeah. Yeah. When you have that, Tommy, like holy smokes, that that is a massive win for you to use agents over and over again all the time, right? right? Period. Like if you can just use it as much as you need to. So I think that’s

55:42 much as you need to. So I think that’s going to become more effective., one part of this whole story that as I step back and look at this as like an architect, I think systems that help you optimize for usage, we’re in the phase of optimizing your agents. Now that’s going to be very much important moving forward. Optimizing, two companies can get the same result with two different ways of building agents. One can go to Opus 4. 7 or 4. 8 eight and run it all the time and spend x number of dollars to build that

56:13 spend x number of dollars to build that system. Another company can come in and do the exact same system build, but do it for a fraction of the cost because they’ve built custom agents and they they’re now understanding how to efficiently use the agents. So, I think we’re stepping into an era of of agent efficient use is going to become more important. Yeah, I I I think Yeah, I I think we’re still seeing the vision. But Mike, I cannot wait to dive into more of this with you because like again, we just have a whole new world to discover, a

56:43 have a whole new world to discover, a lot more to dive into. But Mike, that’s all I got today. Sounds great, Tommy. I appreciate all the listeners here. Thank you for in the chat talking about things. yes, we hear you. And there’s a lot of really good ideas here. Obsidian and open claw. People are talking about the claw now a little bit in the chat. So, I think we’re we’re interested in seeing how these open call or other agents are going to integrate into our fabric and PowerBI building experiences. this is definitely gamechanging level stuff. By far you want to pay attention to Rayfen. We already did a demo video

57:13 to Rayfen. We already did a demo video on building with Rayen on Friday. There’s going to be more Rayfin demos videos throughout the rest of this week. Stay tuned with PowerBI tips. You’re going to see more things coming out there as well., we are doubling down on the rayin and and agent building space for elements. All right, Tommy, where else can you find the podcast? Well, you can find us on Apple, Spotify, or wherever get your podcast, make sure to subscribe and leave a rating. Helps out a a ton. You have a question ID or topic that you want us to talk about. Maybe something from build you want us to dive into. Head over to powerbi. tipsodcast.

57:44 powerbi. tipsodcast. Leave your name and a great question. And finally, join us live every Tuesday and Thursday, a. m. Central on all of PowerBI. social media channels. Thank you all so much and we’ll see you next time. Pump it up high. Tommy and Mike lighting up the sky. Dance to the day to laugh in the mix. Fabric and I get your explicit measures. Drop the beat now. Feel the crowd. 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.