The Intelligence Developer – Ep. 487

In the final live episode of 2025, Mike and Tommy unpack the concept of the “Intelligence Developer”—a role emerging at the intersection of AI, Fabric IQ, and traditional BI that prioritizes dimensions, data quality, and organizational intelligence over just building reports. The episode doubles as a year-end recap, with deep dives into 2025’s biggest SQL Server, Microsoft Fabric, and Power BI milestones.

News & Announcements

Tommy highlights three major year-end recap articles from Microsoft, each reflecting on the progress made across the data platform in 2025:

• SQL Server & Fabric SQL Year in Review — SQL databases in Fabric matured significantly with enterprise-ready features: ALM improvements, backup customization and retention, Copilot integration for SQL analytics, and mirroring for SQL Server and Azure SQL Managed Instance. The focus was clearly on making Fabric SQL viable for large organizations already running SQL infrastructure. Mike and Tommy note the strategic move of combining the SQL conference with the Microsoft Fabric conference in Atlanta.

• Microsoft Fabric Data & AI Recap — A broad look at Fabric’s evolution across notebooks, Spark, pipelines, and the introduction of Fabric IQ with ontology. The platform continued its push toward unified data + AI experiences.

• Power BI Year in Review — Reflecting on the steady stream of monthly updates, Copilot enhancements, and the growing emphasis on semantic models as the keystone of the platform.

Main Discussion: The Intelligence Developer

The Shift from Metrics to Dimensions

Mike introduces the concept of the “Intelligence Developer” as a natural evolution driven by Fabric IQ and AI. The traditional BI developer focused heavily on metrics—building measures, KPIs, and dashboards. The intelligence developer shifts focus toward dimensions: the descriptive, categorical data that gives context to everything.

Why? Because AI and ontology don’t just need numbers—they need well-defined entities, relationships, and concepts. A measure is meaningless without clean dimensions explaining what it represents. This is the shift from “what’s the number?” to “what does the number mean?”

Why Dimensions Matter More Than Ever

Tommy and Mike argue that the rise of Fabric IQ makes data governance around dimensions critical:

• Dimensions describe your business — Customers, products, regions, processes—these are the building blocks of ontology
• AI needs context, not just calculations — An agent asking about “customers” needs a clear, shared definition
• Quality at the source — If dimensional data is inconsistent (different departments defining “customer” differently), everything downstream breaks

A New Role or an Evolved One?

The debate: is the intelligence developer a new role, or just the natural evolution of what BI developers should have been doing all along? Mike leans toward evolution—the tools are finally catching up to what was always needed. Tommy sees it as more distinct, requiring different skills: data governance facilitation, cross-departmental communication, and understanding of AI systems alongside traditional BI.

Building Sound Data First

Both emphasize that organizations rushing to adopt AI and Fabric IQ without sound dimensional data are setting themselves up for failure. The intelligence developer’s first job isn’t building ontology models—it’s ensuring the underlying data is clean, consistent, and well-governed. This connects directly to their Ep. 489 discussion on friction: the ontology will expose your data maturity whether you’re ready or not.

Looking Forward

As the final live episode of 2025, Mike and Tommy reflect on an incredible year of change in the data space. The intelligence developer concept sets the stage for 2026 conversations about how organizations restructure their data teams to account for AI, ontology, and the shift from report-building to intelligence-building. They’ll be back in January with fresh episodes and deeper dives into these evolving roles.

Episode Transcript

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

0:00 Good morning and welcome back everyone to the explicit measures podcast. We are back again with another fun episode. Oh yeah.

0:32 Wrapping up the end of the year. This is going to be our final, I guess, episode live before we get to the new year. So, this will be like a wrap up end of the year scenario here. I I thought you weren’t right at first and I’m looking at all our recorded episodes. That’s right. I will see you after this in 2026. Yeah. , so it’ll be a see you later after this one. So this will be our fun last episode and then we’ll see you next year. That being said, our main topic

1:05 For today is just unpacking this new concept and I think AI is slightly pushing this as well is there’s this idea or concept around the intelligence developer. I think that’s something that’s going to be happening here. , and that’s going to be our main topic unpacking this. I think AI is pushing this direction with orchestrating and talking to things. You don’t need to know as much of the technical details about things. You can let AI handle some more of this. And I think this is also bleeding into BI and business intelligence. So, we’re going

1:36 To probably unpack this, see what this looks like and how the intelligence developer is now maybe a new role. Maybe this is a new position inside the fabric ecosystem. Okay, that being said, Tommy, you got some news for us? I do. And I think a good way to end the year is looking at PowerBI and Microsoft. They have some great articles on a holiday and year recap. So Mike, we can start with PowerBI. We can start with fabric and data and AI or we can end start with

2:13 SQL Server, Azure and SQL database. Each of these are different articles that the Microsoft team put out as wanting to look back at the entire year. I think a lot of the Yeah, I think a lot of the PMs are reflecting back on the year of like what they’ve done, what they’ve built. A lot of things are happening across many different teams. I actually just have the first one in the list opened up here. I believe I’m looking at the SQL recap from Anna Hoffman. Yes, Tommy, you’re a pretty heavy user of SQL

2:45 Databases. Yeah, I I would say I’m not necessarily a DVA, but obviously I’ve touted the importance of SQL servers. I was one of the most excited people when it came out in fabric and it just it’s honestly it’s a it’s something near and dear to me. It’s something that I’ve always used in terms of especially with PowerBI and it’s one of the better ways. Well, it was one of the better ways to connect to PowerBI semantic models before we had direct link. But that being said,

3:20 Applications run on SQL. There are so many types of SQL server databases, but more importantly, it’s not just for our business intelligence. It’s very much for also again grabbing data, extracting data. a lot of people that even if they don’t have SQL or or they want to connect to something that’s more universal, that’s where SQL comes in. And there’s a ton, Mike, that came out in the year of 2025 obviously because I believe Mike, if I’m not mistaken, the SQL databases in

3:53 Fabric was the end of 2024 was when it was in preview. I think it’s been out for over a year now. Yeah. Again, I think a lot of the features for, , I feel that the sac SQL database or SQL in fabric, again, there’s a lot of other SQL things that are out there. One of the I think complaints from the data community was SQL might have been feeling like they were getting a little bit left behind when Microsoft built fabric and PowerBI or more, , PowerBI SQL wasn’t didn’t care, right? SQL just

4:26 Did its own thing. And once Microsoft moved a lot of its data platform technology into fabric pipelines, notebooks, Spark, all the other technologies, SQL didn’t really have a home. There was a data warehouse, but it was TSQL and you didn’t have all the bells and whistles that a normal SQL server had. And so I think there was a for a period of time here, the SQL community was like where do we go? How do we fit into this new ecosystem? Are we just going to be staying inside Azure and then be a data source or or connecting to fabric things? And I think by Microsoft

4:58 Bringing in SQL directly, the fabric SQL environment into fabric that feels to me like it’s the SQL serverless. Yeah. , manage database experience that you’re getting and you’re getting that directly in fabric. I think this is also strategic, Tommy, when you look at what they’re doing at conferences. The conference down in Atlanta will now be a fabric and SQL conference. So, this is a I don’t I don’t know if Microsoft is again I’m not as close to

5:31 The SQL community as other people are maybe listening to the podcast. So, , awesome. Glad you’re here. But the SQL database world, right, now that they’re bringing this all together into a proper data platform, I don’t think Microsoft had a actual SQL conference for a number of years. Someone can correct me in the chat if I’m wrong. I’m probably wrong on this one, but it feels like what was the other company? SQL Saturdays. SQL Saturdays. And SQL bits, I guess, would count. Well, but that’s But I’m thinking like what’s there’s like an organization that would do like

6:03 A really big SQL conference. Is it PASS? Yeah. Pass. That’s what Yeah. Yeah. Yeah. Yeah. Yeah. Okay. So, PASS felt like that was like the conference to be part of when you talked about SQL things. And so now that that pass got I think it got rebought out by Redgate and some other things happened that are interesting there. SQL Saturdays are still happening but like in different forms. I think Microsoft potentially is taking an interesting stance here of trying to combine the conferences down to one thing. So, we’ll see. This will be the first year we’ve had a SQL conference and a Microsoft fabric conference together. Seems like the right approach

6:35 To me. How about you, Tony? What do you think about that? No. And and I think you see with the updates too, Mike, you see that the major enhancements in the what Microsoft’s work on with SQL and fabric is being enterprise ready. So you have the application life cycle management improvements, backup customization and retention and copilot integration for SQL analytics. There’s mirroring for SQL server and Azure SQL manage instance. and you can see there’s really just this focus on because that’s what SQL is not necessarily going to be essential

7:09 For a team of five but larger organizations there are many SQL databases many ways that integrate. So that’s where a lot of the the focus is is hey if you’re using SQL and you’re creating a database odds are you probably have done this before and I think that’s what a lot of these features are especially from the backup customization when it comes to especially the ability for like the workspace level private link support the customer managed keys these are all

7:42 Things that you’d normally find in an enterprise SQL administration anyways and it just makes sense that they integrated that into fabric. What do you think the the threshold is or what do you think business’s thirsts Tommy are right now to say okay I want to build a SQL database I’m going to build it in Azure versus I’m going to just start building it directly inside Microsoft fabric. If you’re building an application that is needing some database, , you’ll you’ll probably think about SQL

8:15 Server at some level. Are companies building that? Do you see them building directly in fabric right now on top of the fabric database or do you think they’re going to continue to build in fabric in Azure for now and the SQL database experience inside fabric is more for like I’m a business team and we need to do some like manual manipulation. We want our own server. we won’t be able to query things like so is this more of like a like a oneoff thing where where and I’ve seen this before right Tommy you’re a business team and you’ve got a lot of data coming

8:46 In from a source of something and you just want you have someone you’ve hired that that is a SQL expert or a SQL DBA and you just want to manage your own database independently of all the other data sources are happening. So I’ve seen this occur. What do you think the fabric database is going to turn into? Is it going to be the backbone for apps or is it going to be more one-offs for for building data systems like in a department? Does that make sense? I think yeah. No, I think it’s going to be one-offs because usually SQL databases are so heavily managed by it

9:18 And DBAs and that customization. For example, I’m I will spare you going through all the different types of SQL databases you can spin up in Azure. depending on the scale or if it run depending run. Yeah, there’s a ton. And generally speaking, those enterprise teams, they want that customization, different databases for different needs in terms of when is it going, how is it going to pull data, etc. And fabric right now, you can only really create

9:49 One type of fabric database. You once there’s the only thing you get to do is choose the name of it. So I and I I forget what Azure actual SQL instance is spun up, but that’s a deal breaker that I’ve seen with a lot of companies saying, , this is just a standard. I need X, Y, and Z. I need this additional customization. Now that being said, Mike, what you said about the business is where I really agree with you where too often there are teams that have they

10:24 Work with vendors, other systems, and like hey, give us access to your database or we’ll push the data to you. And that’s always a big to-do. But now with Fabric Mike, well, we can spin up that database. we can provide those connection strings or e either to the vendor or get that data in without even it or the DBAs having to get involved make sure everything’s set up properly. So I think to your point this is not going to be something I think the organization is going to be like as a

10:56 Whole is going to migrate over right now but I I see a lot of businesses being very very dependent on this very keyed into this cool well I’ll just point out the link from Anna Hoffman great link great article all the things are changing if you don’t know all things about SQL this is the article to go to this is a link tree of links of all things. Down at the bottom, there’s tons of GA generally available items across fabric, manage instances, mirroring, all this

11:29 Really crazy cool features. So, there’s a ton of stuff here as well. I definitely would highly recommend looking into all the features that came out for for SQL and particularly inside Fabric. All right, jumping over to your next article, Tommy, we talk about Microsoft Fabric 2025 holiday recap. This one’s by Alex Powers. Alex. Oh, it’s good to see him anytime he has a article out. What is Alex talking about? So, Alex is talking about the holiday recap unified data and AI innovation. And Mike, what we got here is really

12:03 Just looking through some breakthrough advancements especially around the AI space platform security and trying to mod modernize basically what they’re calling the data states. basically going through the different capabilities for the fabric warehouse. Mike, you and I ourselves this year got to talk with a ton of people from Microsoft which is really cool from Brad about the data warehouse Chris about real time and it’s just trying to make it a very seamless effort

12:35 To get started and get really up and running. Yeah. And obviously Yeah. Go ahead. Then the other parts copilot go ahead. Yeah, the other part is copilot because co-pilot is made acvable across all the SKs. It’s accessible for really every user and it’s across all the different products data engineering, data science, data factory, SQL, real time in PowerBI, you can use C-pilot. So there’s a ton of stuff and then

13:06 Finally professional developers get a ton of toys for them as well. Yeah, I like the integration of co-pilot across most of the tools here. Anywhere where there’s code being written, there needs to be some co-pilot period. I just think it’s anymore, it has to be like part of the development experience where you’re leveraging an agent, an AI, something that’s helping you get through that as well. So, this is a really solid article, lots of details. , and again, Alex calls at the very end here. Look, if you want to learn more about this stuff, you should go check us out. Go mark your calendar for March 16th through 20th, 2026. Fabcon and SQL Con

13:39 Will be together in Atlanta, Georgia. So, , also a great great article, great wrap-up here of all the innovation things. I I like what he was his article or the picture found around the security where workspace managed security pieces was described. I think that’s actually a really good diagram that he’s using there. So, the article is in the chat window in case you want to check it out. , another solid article there by Alex giving a good wrapup here around all these things. One thing I’ll also note is the fabric data days he calls out in the very beginning was huge. Did you did you attend any of

14:12 The fabric data days, Tommy? I did not. Did you? Yes, I spoke at one of them and they there was like 500 to a thousand people watching live on some of these events. there was a huge initiative. It was just about learning. It was about educating, sharing people, sharing information about what . , there’s a there’s a world championship of people explaining how they got through the world championships of visual building and reports they were building. It was really neat. It was it was very nice to see like the community sharing what they know and how they’re using the tool to to build interesting

14:45 Things. So, I thought it was really interesting. , really excited about that. That So, that was a lot of fun. There was a a data viz challenge for professionals. They had a student data viz challenge. I got to call this out because it was a very fun activity. I hope we see more fabric data days coming out. I think this is a exciting event. Seems like we should do more of these like regular stints around just getting people together to learn. Anyways, I thought it was really cool. Good call out there as well. All right, Tommy, what do you have for our last article here? And the last one, Mike. , we’re all about PowerBI at the end of the day.

15:18 And this is also from Alex Powers, the PowerBI 2025 holiday recap. and they’re calling it a decade of innovation and it’s true 2025 was the year that PowerBI turned 10 years old and July 20 25 Mike do you ever say when you’re talking to clients or when you’re doing training that you have been doing it for 10 years and what’s changed in that first moment because it does sound like we’re very old now. Yeah, doing something for 10 years.

15:51 Yeah, I do. I do remember a moment when I decided to move away from staying in a corporate world and go out on my own and build my own business around fabric or Microsoft PowerBI. And I remember my family talking to me and going, “Are you sure you want to do that? Are you sure? , what if Microsoft screws up the product and it goes south and like what will you do?” And I made a decision really early on in my career, Tommy, when I was moving around different business units inside at the time I worked for Johnson

16:23 Controls. One of my towards the end of my career there at corporate America, I started I think around like year six or seven, I really decided that getting into data is the way to go. And I started learning more about it. And the customers we were working with had large amounts of data. We were working with customers that had point of sale systems that were doing 200,000 records a week of information and data and sales and like this this is big like there’s a lot of data going on. Yeah.

16:55 And the more I got dug into it, the more every company was generating more data. We’re adding more devices. There was more information coming to us, not less. And I think I made this decision said I’m going to tie my career to something that’s data related. I don’t think at the time I knew it was PowerBI, but I knew it had to be something related to data or data engineering. And it’s done me really well. Like it’s it’s been very solid. So, I didn’t know I’d be running this for 10 years, but there was a moment where I was like,

17:27 I don’t know if this is going to go anywhere. I’m going to just step all in both feet. We’re just going to make it make it a thing. [snorts] So, luckily for me, it turned out really well. Mike, at this point it’s been a long time or you’re old just with your company, , right? Just jumping ship. Shoot, I’m almost at a halfway mark, too. But how long have you been, , just from the consulting side? So, I I did consulting for a bigger company starting in 2017 and so then I started my own company in 2019. So, been consulting on data and

18:02 PowerBI pretty much since 2017. whatever whatever that is. So, , not super long, but like almost a decade of just doing it on my own. I haven’t quite hit the mark yet. Tommy, when you maybe do the same thing, too, Tommy, like you start your career in college, you work in like the corporate space for a period of time. You leave and then you have this period of time where the amount of time you spend not in corporate America is almost the same time as the time you spent in corporate. So, I’m still I’m still on the smaller side of that. I haven’t quite spent enough years. But but in another

18:35 Three years though, I will have spent more time outside of the corporate America world and more time in the private consulting world of like running your own business. And like that’s you just do it. You just take one step one year at a time. You just keep clipping along and it just figures itself out. And all the more crazy too that this to Microsoft released like a new feature PowerBI Mike just 10 years ago. It it changed our careers and we’re not the only ones. it. I know a lot of people and I know especially you and I our

19:10 Entire career path the trajectory that whatever we were going on completely changed in terms of what we wanted to do and more importantly there are two businesses two LLC’s that have started because of it and Mike it’s amazing because if you told me 10 years ago or maybe I guess 11 years ago at this point that I would be working in data and business intelligence I would not know what that is at the time. And it’s amazing what we’ve seen at user groups. It’s amazing what we see at

19:44 Conferences where it’s not just that we’re working in a product. It’s we just at the perfect time took on a new career. This is interesting you said. So this is I’m reflecting on this article here from Alex as well as what you’re saying here Tommy. So one thing I’ll just say is the PowerBI viz championships. Awesome job. Having a world title belt for viz championships available to you as you’re doing the champion is awesome because it makes for really great pictures to people holding on this belt, this visualization belt. which is super

20:16 Fun. very neat idea there. But that being said, , looking at what where we’ve come with all the data and space to things, what what was been the igniting fire? What what has been the igniting idea here that makes this data thing so much useful? so much more useful to us. Tommy, I think it’s been this whole idea of like graphical, , visually based systems that are moving your data through. And the reason I bring this up, I I feel it’s like around a lot of power query. Like if we didn’t have power query in those

20:48 Early days, we wouldn’t have been as interested in manipulating and transforming and building data. We would have just stayed in Excel and what we knew. So I think Microsoft making this pivot between taking a lot of complex data things and making a a simple UI and then basically simplifying the tooling away from pure code. I think that was a trend that started to happen and you were seeing this I think appear in other places like Talend and there’s other data transformation tools that were out there that were getting in the hands of the

21:21 Business teams right Tableau was another one of these like inspirational early versions of this getting visuals into the hands of people that could just build really impressive things with a very easy to use interface and I think that continued to grow and I think now we are where we have an entire data platform that is all running like It’s all trying to be easier to use, simpler to get your hands on, put it in the hands of the business, and let them just build what they want to build. I think that has been the momentum that I’ve seen over these last 10 years. And yeah, I’m just excited about all these

21:56 Extra things that Alex is putting in here like AI co-pilot, right? You’ve got desktop and web modding modeling parody now going on. Like there’s a lot of big changes that have been happening. So, who knows what this is going to look like in the next 10 years. Tommy, I think this is going to lend very well to our topic today, which is we’re going to stop doing a lot of the manual work and we’re going to start focusing a lot more on building out what does intelligence look like for our business. How does that work for us? I I am looking forward to having this conversation with you. And obviously I know this initially came out of the announcement of fabric IQ but as I’ve

22:31 Been thinking about it more you realize too with PowerBI and just that segue to for me PowerBI it wasn’t just that there was a nice graphical interface there were other tools that did reporting like that it was the fact that PowerBI gave us more or less a pretty simple intuitive way to do really what is an enterprise best practice around data we are dealing with, , the semantic model which is the table relationship that’s been around for a long time and really under allowing us rather than just doing the

23:03 Reporting side where I know you brought up Tableau but that that was very that was report by report basis but with PowerBI you started really having to think about the business in terms of managing that data that source of truth and There were so many other things that came from that. But yeah, Mike, we’re shifting and no matter what you want to do, we are shifting to I think a new space where AI is just a part of our

23:36 Day-to-day. It’s going to be part of organizations. Whether it’s going to be to the capacity that some people think it is, it’s I think it’s going to be something that’s just gonna just seep through little things as well and especially automation. And as we’ve said before, AI runs on data and it runs on good data. So who better at an organization who’s already dealing with the business and already dealing with the data and being that unicorn or that liaison to I think take a stab at the

24:10 Role that I think is going to come up where we’ve been the business intelligence analyst, we’ve been a developer, we’ve been a data engineer, but is there something new Mike that we can call the intelligence developer? I think this is I think this is a lot a lot of these things I think when we talk about intelligence developers and this is our main topic here. Let’s let’s talk about what this looks like. I think a lot of our focus has been modeling DAXs and building like let’s call them like little libraries of

24:42 Information or data across the business. I think when you start having tools like ontology and you’re talking about fabric IQ, I think the addition of agents to this whole data space is changing what we need to be thinking about on the day-to-day basis. And what what do by that? This is where intelligence and things make sense. An agent has the ability in the same way, let me step back here. Let me just let me let me reframe things for for chat GBT in the early days, right? Yeah. Yeah. We remember this is interesting. Tommy, we’re in a a interesting position where

25:15 We are so tied to technology. We saw the evolution or the beginning of like early versions of chat GPT. We heard about the wave. We jumped in and we had these initial moments of like it was super impressive. Like it gives you lots of really interesting answers and it was to me this moment of well it stopped being a novelty and started adding real value. Mhm. Right. that that wave of wow this is impressive. It can scan and search and

25:48 And comb through a lot more things than I can individually. It can look at 30 articles, summarize each of them, take all those summaries, pull them together, and then provide you an output of like multiple websites summarized together to get you a better picture of data or picture of information. That I think is the catalyst now for the intelligence developer. Right? I can stop I can I need to stop thinking about a single semantic model as the only source of information. It may be a semantic model. It may be some PDFs. It may be some data from SharePoint. It

26:20 Potentially could be other semantic models or other data tables that we’re going to provide to this. Maybe there’s something I’m doing predictive and I’m leave I’m landing tables down in my lakehouse and a combination of semantic model with some predictive data is now what I want to inform people about. Right? When when do we get to the point where we can talk to an agent and say, “I need you to predict XYZ things and have it do a deep think.” And so the agent sits down and says, “Okay, I found these three semantic models. Do these things

26:53 Make sense to you? What do you want to understand this part of the relationship of data?” And you say yes. And it says, “Well, how about I build you a notebook and we’ll make a new table for you and we’ll explore this together.” and it just starts building things on like it’s going to be more of this interactive thing that we can jump back and forth with to have the agent help us out with building net new tables, building net new things. I don’t think we’re there yet. I don’t see agents and fabric building things for you automatically yet, but I don’t think we’re far away from that space. Does that make sense, Tommy? What I’m

27:24 Describing? Yeah, it so I’m intrigued that you went that route. So, and if I’m not mistaken, you’re basically you’re talking in the mindset of the individual developer working with AI like an already someone that exists in PowerBI or fabric that’s just going to be enhanced with AI. To me, this is a little different, Mike, because I’m thinking about this in the terms of when an organization’s when an organization is going to foresee or expect from the developer

27:59 Role at an organization. So, let’s rewind a bit. Mike, for the last 10 years, if you were working FTE and you were working PowerBI, what did people just assume that you created? What what did you do at the end of the day? And obviously that’s a heavy loaded question, but how do they know that you were working? What was your output? Well, I think a lot of what I was forming in the early days here of PowerBI, which was like the center of excellence, talking about policies,

28:33 Turning on the right features, supporting the business to build their own things, right? These were all very focused around the semantic model and reports. That’s the place we started from. But where I’m going with this whole new world of the intelligence developer and we need to encompass more than just the models, right? There’s a lot more our business A lot of my effort was taking all the company knowledge and narrowing it down to a model, right? Get it into the model because then I could serve the model back to the business in other ways, right? What I think is happening though is the

29:05 Agent space or being agentic in things that are cross data is going to allow me to unblock to some degree away from only having to focus solely on developing just semantic models. So that’s where now knowledge and information will span multiple models in the same way that this is why I bring up the chat GPT question right in the same way chat GPT in the early days was summarizing multiple pages bringing me back summaries providing value that way I think we’re going to start seeing the same shift in the data space where it’s not going to

29:36 Just be one single enhancement and Tommy to your point we’re already seeing agents become part of this you need to provide it instructions I need to have a data agent I need to provide at multiple data sources. So, we’re giving already the agent more context. We’re basically helping it like post train, right? You have the training of the models, but then you have like this supplemental data that you’re providing to it that’s then helping the agent become smarter, better, give you more accurate answers at the end of the day. honestly, Mike, you could say that for I

30:13 I think any role, not just for us in PowerBI or Fabric. I think marketing is going to see that, operations going to see that where AI is going to be an incredible integrated part of their workflow. Let me stop doing I’m going to stop writing DAX like the thing is I’m going to stop doing the actual work like I’m going to be okay the idea is the intelligence is this higher level thought right this higher level concept like what what data should we be collecting to to get the the answers that we need for our business.

30:44 Sure. I think the agents and this is what I see in development already like with app development with with code. So code and agents already have shifted this mindset. You’re you’re going away from this Tommy in the in the AI things that you build. You’re not building the actual app anymore. You’re you’re instructing an application an agent. I’m going to build this app. Here’s the features I want. And you describe it to it and it figures out where the button should go, what the UI should look like, how do you do the swiping. , there’s all kinds of really rich things where you’re not trying to

31:15 Tell it. You’re just guiding more now and you’re letting the agent figure out the code. Dude, like I said, this is happening in advertising. Have you noticed how many commercials there’s a little little tagline at the bottom that says, “These images of this video was created by AI.” I think Okay, we haven’t seen that yet. Is that is it Coca-Cola is not they’re putting disclaimers on ads that are saying that some some companies are some Coca-Cola is not but I can guarantee you what they created was AI but there was a car

31:48 Commercial that put on it’s like hey everything here was generated with using AI so as a as a disclaimer. So this is going to take to your point mo I think a lot of different roles are going to go to that same flow. So I I I don’t think that’s too distinct for us in our space. So a heck [clears throat] [laughter] we can create a PDF a nice graphic design right now with claude.

32:20 They have that new feature called claude skills. it actually can spit out slides and can actually make it look pretty nice with if you feed it the right information. I I want to argue with you though because I think the idea here that we’re calling this intelligence developer is not so much to me about the tools I’m using in my workflow like using AI or how much how many agents are part of the workflow for people working in data as

32:53 Much as really what my final value or final output’s going to be in an organization. The question I asked you, I asked you like how like what was our end goal for the first five six years with PowerBI when you were developing something when you were a developer. It was a semantic model and a report and that was it. That was the end of the road. That’s what the holy grail was, , for most the majority of projects. I think this is changing with this with AI because if data is going to be

33:28 What AI runs on especially with copilot studio and custommade tools and organizations I think we’re finding now that the report is like I said it’s been downgraded to me in terms of I think what the you’re going to see for roles people who are working in data and I think just like people in PowerBI these are not just special developers are already at the forefront that we have a now a final or like a side destination. We have reports and

34:00 Semantic models. Those are great. I think semantic models are going to still be our beall endall. However, we have to also design now not just for how well it’s going to work in the report, but I think these intelligence developers, which is probably going to be us, is going to be designing databases and designing data. So, it does work well with anything that’s going to be a AI agent or organization. So, now I’m designing things differently. again 10 years you’ve designed more or less if

34:34 You’re working in PowerBI or opened up PowerBI desktop or Power Query or whatever it was the idea that this should work with the semantic model as efficiently as possible and this should work for reporting as efficiently as possible that was a beall endall but Mike we’re now seeing the gap here where rather than all the KPIs are the essential part well how do we categorize information do we have the right input output for what is an AI going to see? What is an agent going to see with this data that’s going

35:07 To make it run as efficiently as possible? And I think this is now we’re shifting there. I don’t think we’re there yet, but to me, I think organizations going to just look to the people in business intelligence. Heck, we have intelligence in our name. Say, well, you already know the data inside and out. I’m assuming you’re smart and you see all of our data. You already know the business and what their problems are. Well, let’s start designing data in a way that’s going to work better with agents.

35:41 , that’s been my point this whole time is it’s the agent space of this that’s going to continue to change like what we’re doing. But even even what I’m observing right now, how agents are working, the agents all still need to be pushing down direct information to something that is tangible underneath, right? If you look at co-pilot home, if you don’t have a good graph or chart inside the model that already answers the question the user is asking, it’s more difficult to get them the right answer or have the co-pilot

36:14 Display a chart that me is meaningful. Right? So this is the challenge that we’ve had since Q&A visuals inside the report, right? It the the the model is going to think about what it it feels like it needs to have informationally wise, but to be able to prompt directly to say I need a table of this this and this information unless you describe in detail and this is what a lot of I think our time is being spent on right now which is writing more descriptions, adding more information into the model. where this data come from that’s becoming

36:48 Incredibly more important and that’s where I think the intelligent developer is coming from is yes we have the data we can we can get these tables made but under the underpinning of all the semantic models or under under all the AI is a bunch of semantic models and reports describing why those reports are there what information they’re providing what intelligence are these reports that we’re building gives back to the business that’s where the agents I think are going to shine because then the agents actually are able to resolve real questions that you’ve already built for. Where I think

37:21 Things are weak right now is when people are asking questions about your semantic models or reports that you haven’t built yet, right? You don’t have a sales by category visual on the page. The agent struggles to figure out how to build that stuff yet. And it’s going to get better. It’s already starting to do better of like querying the tables directly, going to a lakehouse and seeing that there’s tables and there’s relationships in them. But you still have to describe all that stuff. , I’ve seen a lot of articles. Marco Russo a couple months ago put out an article around SQLBI that said semantic models are the best thing you could ever throw

37:54 At an agent, an AI, because it describes so many additional properties, relationships, descriptions on columns, descriptions on measures, identifies calculations, it has dimensions, it has KPIs or calculations you’re going to want to run. All that stuff is required for the agent to be smart about what you’re doing the model. So having that stuff at its disposal makes sense. Yeah, but I feel like you’re still thinking of the report too much of co-pilot with PowerBI because those have been I know that’s not what you’re saying, but those

38:26 Have been your examples so far. I want to challenge you because I’m seeing this as a responsibility shift for what we do and I I’m willing to predict that. take five years from now, Mike. Like reports will be a byproduct of what we do, but it’s not going to be the the the final destination of most of the data that we’re building. I think a lot of the things that we’re going to be creating and I more better way to say that are the things that the organization will be using that we

38:59 Create is not going to be reports as much. it is going to be utilizing AI throughout their throughout their services or their operations. Whether or not they know that we created that agent or not or helped work on that agent’s flow of data or not is not the question because again yes we can develop a report so it works well with co-pilot but to me I I’m expanding this a little more where I think we’re going to be doing a lot more with

39:33 Helping the entire organization even if there are AI specialists in an organization or whatever the case may be. Well, we’re going to be the ones responsible for making sure that the data sets that are for reporting, there’s data sets also for those agents because that is I think to me a slightly different format also helping develop that. I I really think that take five years from now, reports are going to be a byproduct, but it’s not going to be the goal of what we do. It’s going to be something that we

40:06 Do, but not nearly as much. There’s going to be other things we’re going to be creating in this space that’s going to take up just as much or if not more time than building the nice visuals and building the reports. I may be a hot take, but I I would argue that’s probably a hot take because I’m going to disagree with you on that one. So, Tommy, I think the model the reports is everything that people’s going to be continue to use, right? You can’t you can’t upsert I don’t know what 30 40 years of table based information and table or reporting information that people have been using

40:39 For years. It started with SSRS and pageionate reports that’s been around for a long time. Now you’ve got PowerBI reports for 10 years but that’s also been like add into this Excel and people being able to manipulate and shape and push data around. That’s also so what are we seeing? I think agents are going to help us push this data around faster. I think agents are going to help us build things quicker. I just saw an article talking about how well agents are able to manipulate Excel now. Like you can now have agents

41:10 Manipulating Excel better than experts in your company. So the data is in there. You can ask it to build things. It will create the formulas. It’ll make new tables. All these things that you could just do. And so instead of actually having to go in and know how to write all the formulas, you can describe what you want to do with the data and it just builds it for you. So I think we’re going to think the agents are going to be much more of an advantage to ourselves in the creation and curation of data, right? And that’s where where I’m seeing myself doing a lot more the intelligent developer. But at the end of the day, if you don’t have a report, and

41:43 This is where I think maybe Microsoft is missing the mark here a little bit around the agents and like using agents to get answers out of data. I think agents should be used to generate things more so than agents being used to resolve the answers. When I go to talk to co-pilot or co-pilot at home, okay, I can throw a model at it. I can throw some things at this agent, but what do I really want to get to? I’m still trying to land on a visual, an insight, a collection of visuals that

42:18 Get me my answer. Whether or not those have been pre-built into the report or not, that’s debatable. And I think that’s where AI is going to help us create these experiences. But I think at the end of the day, you’re always going to land in some form of a report, whether that’s generated dynamically by the agent or that’s been pre-built and you’re just telling the agent where to go look for things. I think it’s going to be a hybrid mix of this piece moving forward. But I really feel like at the end of the day, we’re always going to have to land in some visual or report or collection of visuals that are going to help us get to

42:51 The answer for the answers. Will we save that result as a report page somewhere else? Maybe, maybe not. We may just throw it away and start again somewhere else. But I really think there’s a need here for once once you get through some data and you start finding insights about it, you’re going to need to capture that insight or capture what was produced there and you’re going to want to like revisit it, share it, have it updated with new data. So I I don’t think I’m going to I think you’re a bit too optimistic on how much the agent can do on the report side. I think I think I’m a little bit

43:24 More pessimistic there where I think agents are going to be better in the develop the information or the intelligence developer side of things is we’re going to be using agents to help us quickly create reports, quickly create pageentated things. That’s where I think the value is going to really come from because we’re already used to interacting with reports and tables and things as well. [sighs] Yeah, this just goes back to my long-standing policy that no one actually wants a report like the report is just the best medium right now for people to consume that their information and right now pretty visual

43:57 What do they want? They want their they want their information what they need to do and how they need to act. And right now the best but how do you trust that? How do you So that’s an open-ended question. Sure. Well, this is my point, right? The report supports that decision-m. Mhm. Right. So I agree with you Tommy in the fact that people want to make decisions or make smart decisions on based of the information in order for you to get to a place to like people want the what action should I take place to provide

44:31 The outcome that I want. Totally agree with you there. What you what you need to have behind that is how do we support that with the tools that we’re providing? Like how how do I trust what information is being given to me? Do I just blindly trust that cop? Hey co-pilot, what should I do today? And it l it goes through my calendar. It goes all the things and tells me here’s what I need to do. Unless I built that system myself, I’m going to be skeptical and not trust it. And this is the reason I think why Excel is so popular right now is because most systems if you

45:07 Most systems at least dump out the data to Excel and the first feature and any project I’m ever on if you ever show someone new the very first question is where’s the export to Excel button on every app and if you don’t have it and if it’s not easy that’s when people get really bent out of shape. So, I think there’s a trust issue here of like, do I blindly trust the AI agents to aggregate and produce the right information? Is it lying to me? How do I know that it got there? What’s the line of thinking that helped me get from point A to point B? That’s the stuff that I want to

45:41 See in order to trust what the agents are spitting out to me at this point. Does that make sense? Yeah. So, let me ask you a question though. In five years from now, take a semantic model. Will more queries be coming from humans querying the report, going to the report or from agents through via an API or connecting to it in in some automated fashion or through some chatbot? I think you’re going to enhance the models to allow more agent level things. Will if you

46:15 Look at the query number of queries coming to the model, will it come from agents or will it come from users? I think the majority will still come from users. I think agents will will supply some of this information. there might be some interesting things, but I think what will happen, let me say it this way. Agents are expensive to run, right? So I want an agent to go find the information, the insight, the the stuff that’s important to me, produce the visual or produce the report and then

46:48 Fall back to just using that report to connect to the model. So I think the agent is good for discovery and creation of the first artifact or first couple artifacts and once you have it, I want to lock it down. I want to reuse it, right? and the reusing of that part doesn’t come back to the agent every single time. So I think agents make a lot of sense to help you like do data discovery or build your first visual or come up with an interesting page design. Like those things I think are where agents excel to help users create things. But once that’s been produced, I

47:23 Don’t want the agent involved anymore, right? I want consistency in that output and I don’t want to have to give it another prompt and then Microsoft under the hood has shifted which agent’s doing what and now it’s giving me a different result. So I think people expect consistency and this is where I think the let me give you let me give you an example here. One of my developers and I were working on building some application stuff. We wanted an animated icon to be presented right. So, my developer went over to the

47:56 Figma AI and was building some things there. Here’s my icon. Here’s the SVG code. Can you make this thing animated with a little bit of coloring and blah blah blah. I myself took the same icon, put it in VS Code, and just started talking to the GitHub copilot and said, “Hey, I have this icon. I want it to look like this. I want it to do this , , color animation thing to make like a loading icon for some application.” We came up with two entirely different results, but the output of the agent doing the creative piece was a bit of code and I said, “Build me a component that does

48:30 This.” So the creative side, the creative portion of the agent was using it to build the code. Once I had the code, I deleted the agent away. I did not need the agent anymore. I had the output that I wanted that was consistent and I used it over and over again. So that’s where I think the opportunity of this lives and this is the same thing with all these other this is the same pattern I see Tommy with like all these coding agents that are building apps on demand lovable base 44 cursor code right all these applications are

49:04 Building Google’s what is it Google’s anthropic not the anthropic cloud code no or Google oh the the gems the new apps they just released Yeah, it starts with an A. I thought anthology or something like that. I don’t remember what it was. Anyways, Google’s got its own. So, every every computer program, every company has its own like build app code generated thing, right? You use the agent and the AI to build it the first time, but then you’re done. You have the code, you build the app, and I’m even seeing things now

49:38 Where you use the agent to build the code. The code gets now you could have a onebutton deploy. I saw this just recently where you can add a login screen, add a database, and then you add a payment system and then you publish the app directly to the app store like automatically. So that’s the agent helps you create, but once it’s done, you’ve built the thing and then I use that thing over and over again. So I think the that’s the same pattern I want to apply to data and AI. use the agent to help you create the first insight or

50:12 Help you discover the insights in the thing, but then all of those insights are going to be immediately pushed directly to anti-gravity. Thank you, chat. That’s what I was looking for. Google’s anti-gravity. That’s the program. It started with an A. I knew it started with an A. But so, so that’s where I think things are going to go, right? Agents are going to help you create the first artifact quickly. Once you have it, then you lock it down and you use it over and over again. That’s where the power of agents come from. I think an organization that’s going to be different where there’s going to be

50:44 The verb dip task and I see what you’re saying, but to me, dude, if if the pattern’s already You’re not going to just throw away something. No, but that you’re you’re not going to keep using the agent every time. I’m not gonna I don’t want to go back and prompt the expensive agent like the agents are using GPUs on all this stuff. So I don’t I don’t want to keep going back to an agent and having using a GPU, high CU usage, all this stuff to create like the agent should be there to help me create really good stuff and and that’s where the agent should sit. It should help me build repeatable

51:18 Efficient things. Now maybe the agent will recommend an entire semantic model build. Great. But what that’s going to do is it’s going to I throw an agent at a lakehouse. The agent will help me go through thinking out what should I be building, how many domains should we have, what are the table relationships. I describe what I think I what I want to give to the agent and it just figures things out and say, “Okay, Michael, here’s six models that I think you should have. Here’s the measures in each model. Here’s how these models should be broken apart. Here’s the audience for each model and let me build them for you.” Great. Go ahead, agent, do the

51:50 Thing. And and that’s where I think what’s going to get really effective. And I think that’s the area where Microsoft should be continually investing is working on the agents and not trying to pull insights directly from the data. I think yeah the generally people are trying to use an agent to pull the insights out. I don’t think that’s the most effective use. Completely agree with you completely agree with you on that because we’re going to see the most effective use case of AI in organization is not is going to be automation. It’s going to be whereas like the insights is going to be

52:23 Something where they’re going to be acting upon where it’s like again I think operations is been my example I’ve been using where an operations team is like hey co-pilot show me all my out customers with an outstanding PO and send them an email right and it can automatically go through that’s I think it’s going to be some data that we’ve created some of it’s going to be also coming from your shareepoints and Excel but it has to be structured the right way and that’s going to be something they’re going to do all the time. , I know we’re already getting near time, but I do want to ask you in the future if we have this

52:59 New role and I would even argue that this idea of the intelligence developer is not something new. It’s just I think it’s already existed. We’re just changing the name of the report developer here in the future for someone in our space for the report developer or the data developer. Are there skills that someone who was like I’m going to keep being in this space? Are there any new skills that you think people need to learn to

53:31 Transition? Is is it really just keep going as you learning what you already know or where would you focus your core skills in terms of looking at the future working with AI and working with fabric? This is a great question Tommy. I I think this is going to go back to a lot of what we see already inside the BI space. I think those soft skills are going to be most essential here, right? Being able to listen to the business, understand what they’re trying to ask

54:03 For, and then I think the skill here is I’m already seeing, and Tommy, I think you’ve already felt the shift here as well. Being able to write good prompts and understand what the agents can do and can’t do well is going to be essential. when we started learning how to Google this how do we how do we write a Google query how do I write a statement inside Google to get back to this the search results the way I wanted right it’s the same thing now I’m I’m reprogramming how I interact

54:36 With computers because I’m shifting my mindset about what it can do and what do I what how much information I need to give it for example let me go back to some development scenarios here working in working with agents and building things. Now, I had to be more explicit in conversing with an agent. I had I had to give it a lot more details around, okay, I want you to build this. , , I’m going to have a button. It’s going to look like this. I want you to place

55:08 It here. I wanted to go get this information from this specific area. Here’s the connection string. Here’s how we connect to it. U provide authentication this way. Don’t store it this way. So you’re being much more explicit and you’re I can still get the gist across, but instead of me doing that myself and writing the code out individually or building the solution myself, I can now communicate or doing a lot more orchestration guidance, right? And so this is where I think the information developer or the intelligent developer is coming from is you’re going to be doing a lot more guiding of the

55:41 Agent to build those things that you care about, right? It’s the same concept. So, we’re going to need to figure out how to get agents to build a pipeline, right? Hey, agent, I want you to build a pipeline that’s going to look like this. It’s going to have this activity, this activity, this activity. The people that are going to benefit the most from agents are the people who already understand how it works. We’ll need more years before like Tommy for example like if I asked you Tommy describe to me a data pipeline that

56:16 Would load an Excel file from SharePoint and you could basically tell me what you would want it to do right you could say okay open up a data flow on the dataf flow here’s the SharePoint URL use the SharePoint URL find the folder that contains these files okay the files are all the name. I want you to merge all the files down to one thing. Use the combine data feature in Power Query to merge all the files together in one thing. Promote as headers. So, we know like you and I, Tommy, know the terms we should use to prompt an AI to go to

56:50 SharePoint, grab the files, promote the headers, and merge it all together. That’s the skill we’re going to need to have to continue to have. We have to understand the fundamentals there. And where I see AI in the app development space is AI is incredibly powerful for the senior and experienced developers because they already know what they’re trying to build. And once the agent builds something, I know it’s wrong or right. What’s going to be interesting here, Tommy, is in three years or five years from now, what does that look like

57:23 Moving forward? and the agent will still build systems and things for you, they won’t actually be doing like we’re going to have a new wave of developers that have never known how to build everything from scratch. We’re going to have a new wave of developers who’ve never built M language before or have minimally invested time on learning Power Query. Instead, they’re just going to describe to the agent what they want to do and then the agent will interpret what will happen inside Power Query. Again, we’re

57:55 Still going to build systems with the agents. The agents are the builders of things. It’s just going to help us go faster. So, I guess to answer your question directly, Tommy, where where do we spend our time? You spend your time figuring out where are agents inside the fabric ecosystem? How do you use them? How do you integrate with them? you will get better at prompting, communicating highle goals and objectives to agents. you need to understand deeply the business requirements and being able to interpret them for agents to build systems you can go use. I think that’s the gap that

58:28 We’re we’re trying to close right now is from the technical knowhow to AI system building. That’s what we need to understand and that’s why we’re going to this like intelligence developer piece because you have to understand what what is intelligence for your business. What does that look like? Where does your data come and what supports that intelligence? That’s something the agent just can’t know. You have to have people know that and you have to have people work with that data and structure to get the agent to build you the automated

59:00 Thing that can then load that insight every day. Does that make sense what I’m saying, Tommy? No, I I I completely agree from the like I we’ve been saying this for two years now, how do not underestimate prompting or context engineering, which is the new one, as a skill is as a hard skill that you need to be able to not just be able to put on your resume, but prove because it not just from optimization, getting things done, but also to your point, it’s it’s costly. So, it’s a huge part of what we do. I I would even go a step further, Mike, and

59:34 Say too that this intelligence developer, some of the skills that you also want to expand on is dimensional modeling, what we’re doing from a semantic point of view, if you’re not perfect, you’re in terms of feel like you’re near expert at that, that’s where you start. But I think too there’s also something that we’ve already done as business intelligence professionals where we’ve had to be that liaison that think like the business act like it. I think the ability to communicate with the business on what they’re trying to

60:08 Do and being able to translate that now to an AI project or an agent is never been more important. Now, the good news is if you have been working in PowerBI for a long time and you’ve already been a report developer, this is something that you’ve already gotten a good hand on in terms of making sure you’re asking questions the right way, not overwhelming people, but making sure that you extract what the scope is. And I think that’s going to that’s never been more important. And u especially

60:40 Where we’re going now. I think there are a few other things that we’re looking at as well when it comes to how our titles and roles are all wishy-washy especially in the BI space. I think it’s going to get a little more how do you say gray because I really I don’t think it’s going to get more gray. I think I think the roles are they’re already clearly defined. I think I think it’s I think the roles are already there. I think you’re going to define a lot of the roles around what

61:13 You’re doing on the particular technology stack. I don’t think I think there’s new roles appearing that are going to be more multid-disciplinary, but I don’t I think the roles are I don’t think it’s wishy-washy at this point. Well, I think for you and for you and I it’s it it’s not. But go again, go look on LinkedIn. go look at the job applications or jobs that are available and how many different positions say the same thing in our space. we do not have a structure around that. I I but again

61:45 That’s neither here nor there. I do think though we’re going to see some additional roles for people in our space where we’ll basically have like a semantic architect. We’re gonna have someone who’s going to be just like we have a solutions architect now who does basically everything in PowerBI and for data projects. I think that’s just going to expand to you do everything in PowerBI and fabric and you’re also building the cop co-pilot agents and you manage that front to end. , to me, I think that’s really going to be a single role. And Mike, in the

62:17 Same way that you have a PowerBI data engineer, right, where I do PowerBI reporting, but I’m also doing data engineering. They usually are, , they can be two different careers or two different roles at a company, but a lot of times they’re meshed together. I think we’re going to see a lot of this space where people when companies are beginning to look for people who are experts or have the skills to build and help implement AI at an organization. I don’t think they’re going to be looking for someone who’s

62:49 Built their own local model on a computer. As much as that pains me to say it, they’re gonna they’re going to be looking for someone who has a strong grasp on the technology, is great at prompting, and knows their data and knows what the data is supposed to look like because most of these tools now, especially with fabric and what Microsoft’s doing. All you really need is the right data and understand the structure of like for example, Copilot Studio, the things that it needs to get fed and what the business needs. I don’t need to necessarily develop the hugging

63:23 Face model and all the Python behind and create our own model. I just need to have a great idea on the technology stack what it’s capable of and what our data has and then working with the business to change that process. Now again I may be head in the clouds there thinking that’s I but I that’s what I expect to happen. So I think for a lot of people if you want to get started in AI and you want to actually say hey what like just like when PowerBI came out Mike

63:55 We’re like I want to be an expert in this field. It’s very new but there’s a low barrier to entry and we set ourselves up from the beginning just because it was so new where I’m like I’m going to learn everything I can. I think the technology is already available to us especially if you’re an office 365 shop where there are the things that you’re that that’s you have available to you the tools the technology and the data already to get

64:29 Started and I think it’s just really beginning to market yourself that way but also again it’s just learning the technology that’s out there it’s going to handle a lot of it. So to me, that’s where I see the future as well, but I think it no matter what it’s gonna start with semantic modeling. I think we will never get away from that in our space. I don’t think the ch You said a lot of things there, Tommy. You didn’t really focus on a single point. You rambled a lot. So like [laughter] No, I guess that was the head in the

65:00 Class. I just I want to do AI and data. I I know you do, but I don’t think the roles are changing. I don’t think all all we’re seeing here is we’re seeing an expansion of an existing role, right? The data analyst is getting more capability. I think the data analyst is going to need to be understanding how agents work. All this is doing is it’s giving more capability to the analyst to do more things, right? Nothing’s changing here. I don’t think it’s wishy-washy. I think the roles are very clearly defined. We have data engineering. We have data science-l like

65:32 Roles of study and then we have the analyst building reports and or working with agents regardless how you roll these different roles out. They all needed context. So around agents and how they’re going to change their workflows. So all we’re introducing here is in the data engineering realm, here’s an agent to help you build pipelines. In the data science realm, you’re building agents or you’re creating things with that world. In the data analyst realm, you’re still using the same things

66:04 You’ve always been doing. You’re talking to the business. You’re understanding how the data relates. You’re now just adding the flavor of an agent to help you build things quicker. So, I think that’s where things are going. I don’t I don’t think any of the roles are changing. I don’t think it’s any more wishy-washy. I think we’re just adding more capability to each of these individuals and helping them get their job done better. So, I I disagree with your point around it’s wishy-washy. It’s all over the place. I do think what you’re going to be doing day-to-day will be changing. I think semantic models are still very important to what we’re going to be doing long

66:36 Term. Where I think the trend should start moving forward towards is we should start focusing more on using agents to help us build regularly repeatable visuals and reports and stop trying to get agents to try and give you all the insights from a single answer because I think people are just naturally not going to trust it. like I can trust having the agent build me a really complex thing and then I can go back in and look at the Power Query and the tables and the

67:07 Relationships and what did it build and then I can vet it from what I know of what should be built and then we can you like prove it out that this is what we want. So I think that’s that’s the space that I think is going to be the most relevant here for agents. things that I can build and physically test will help me produce more things more efficiently. All right, with that being said, let’s go ahead and wrap up here. We could probably talk about this for like another whole hour. So, that being said, thank you all very much for listening to us ramble about the intelligence developer, , what they may be doing in the future and how

67:40 Agents are going to potentially assist us building more regular components or things inside PowerBI. That being said, we do appreciate your listenership. you guys could be spending a ton of time with your family or doing other important things. we don’t want to take you away from that anymore. Thank you very much for listening. We do appreciate your listenership here. If you like this content, please make sure you share with somebody else that you think might also like it. I think we’re on a pretty good clip here building out lots of really great content. That being said, Tommy, where else can you find the podcast? You can find us on Apple, Spotify, or wherever you get your podcast. Make sure

68:13 To subscribe and leave a rating. It helps us out a ton. Do you have a question, idea, or topic that you want us to talk about in a future episode? Well, head over to powerbi.tipsodcast. Leave your name and a great question. And finally, join us live every Tuesday and Thursday, 7:30 a.m. Central, and join the conversation on all of PowerBI.tips social media channels. Thank you all so much, and we’ll see you next year.

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.