Define the Problem Before Tools – Ep. 498

Mike and Tommy tackle a mailbag question about defining problems before choosing tools — and why goal setting matters more than ever in a world where AI lets you build anything. Plus, the January 2026 Fabric feature updates drop with 17 noteworthy items.

News & Announcements

• Fabric Connection inside Notebook (Preview) — Microsoft now lets you create and manage cloud data source connections directly within Fabric notebooks. The feature supports multiple authentication types including Basic, Account Key, Token, Workspace Identity, and Service Principal, making it easier to securely access data from Azure Blob Storage, PostgreSQL, S3, and more without leaving your notebook.

• Fabric January 2026 Feature Summary — A massive 65-page roundup covering updates across the entire Fabric platform. Highlights include AI Auto-Summary for semantic models, parent-child hierarchy in OneLake catalog, high concurrency mode for Lakehouse operations, result set caching going GA in Data Warehouse, and FabCon Atlanta registration (March 16–20, 2026 — co-located with SQLCon for the first time).

Main Discussion: Define the Problem Before Tools

A listener mailbag question kicks off the core topic: “I just finished listening to the adopting Copilot standalone for Power BI episode. You may have addressed this before, but I enjoy hearing a conversation about addressing what is the problem you’re trying to solve for.”

The January Fabric Updates — Quick Hits

Before diving into the main topic, Mike and Tommy spotlight a few standout features from the January 2026 update:

AI Auto-Summary for Semantic Models — Tommy’s first pick. The OneLake catalog now shows AI-generated summaries that help users quickly understand a semantic model’s purpose without opening it. Tommy ties this to his core consulting recommendation: every report must have a defined purpose with two to three questions it answers.

Parent-Child Hierarchy in OneLake Catalog — A quality-of-life improvement that shows how items relate to each other (e.g., a lakehouse and its SQL analytics endpoint). Tommy calls it “just needed” — the language already existed elsewhere in the product, and now the catalog properly reflects those relationships.

Mike and Tommy also riff on features they wish existed: model view diagrams inside the OneLake catalog, pinnable favorites in the sidebar, metadata descriptions for lakehouse tables, and annotation capabilities in model views.

Technology Shifts What We Work On

Mike opens with an observation: the last two months have been a tipping point for AI-assisted coding and development. But he’s quick to add a dose of realism — technology doesn’t reduce work, it shifts what we work on. Tommy agrees, pointing to the Apple movie Tetris as a parallel: 8-bit developers were doing complex work with the tools available to them, just as today’s developers tackle bigger problems with better tools.

Hyperpersonalization: The Real AI Unlock

Mike introduces his concept of “hyperpersonalization” — the idea that AI enables building tools that are uniquely tailored to individual workflows. He shares a story of helping someone build a local HTML page that loads Excel data, applies business logic, and processes it — all vibe coded in a single session. No deployment, no infrastructure. Just a hyper-personalized tool for one person’s specific problem.

This leads to a pointed question about Fabric: where is the equivalent experience? Mike looks at tools like Lovable, Replit, and Base44 and sees vibe-coded app creation that Fabric doesn’t yet offer — despite Fabric already solving two of the hardest problems (authentication and data storage).

Just Because You Can Doesn’t Mean You Should

Tommy pushes back and recenters the conversation on the mailbag question. Most people listening work in organizations where AI access is limited — they can’t freely authenticate external tools with company data. The real challenge isn’t what you can build, it’s understanding what you should build.

Tommy’s framework: every organization has two constraints — money and resources. The explosion of AI tools has reduced the cost of creation, but that makes goal setting more important, not less. Without clear goals, people fall into the trap of building things for the sake of building them.

“The greatest gift that these new tools give to us is not the ability to create things, but it’s the time to be more strategic.”

The Excel Expert Analogy

Mike draws a powerful parallel: just as consultants can walk into any organization and immediately spot the “Excel experts” — the self-starters who understand problems and use tools creatively — the same pattern will emerge with AI. Organizations will start recognizing their “AI experts” who know how to prompt, build tools, and solve problems autonomously. Those individuals will become force multipliers.

The AI Tool Belt

Tommy shares his “must-have” list if he ever went back to full-time employment: Claude, OpenAI, open-source models, and Perplexity — arguing he’d work like five employees with those tools. Mike agrees but adds a nuance: he’d run those models through Azure AI Foundry rather than sending prompts directly to third-party APIs, keeping data within Microsoft infrastructure.

Mike also flags Google’s aggressive push into the AI space as a potential disruptor, noting that Gemini and Google’s free tooling could squeeze out competitors. The coming months will see intense jockeying between the big tech companies — and ChatGPT may even start serving ads in threads.

Looking Forward

Episode 500 is two episodes away, and Mike and Tommy are planning giveaways, surprises, and a celebration. They encourage listeners to spread the word and join live for the milestone episode.

The bottom line from this episode: define the problem first. In a world where AI makes building almost free, the winners will be those who invest in understanding what to build and why — not just how.

Episode Transcript

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

0:00 Good morning and welcome back to the Explicit Measures podcast with Tommy and Mike. Good morning, Tommy. And how goes your Thursday?

0:37 Dude? I’m pumped. I feel pumped. Maybe two episodes away from freaking 500 or maybe just we’ve been talking offline about a few things, but I just feel I feel good.

0:49 Excellent. Yeah, there’s a lot of things changing right now. Again, Tom, I think we said this in the last podcast. I feel like the level or fever or the voice or something has leveled up here in the last two months around code, code development, building things. I’m feeling much more optimistic around where code can help me build out my tasks and supply me with additional support. It’s interesting. It stinks that we have to get — well on one hand it’s very annoying that we have to get to this state where there’s all these like remedial tasks that we do.

1:23 Like the whole idea of technology was to make our lives easier not more difficult, not work about having more time on our hands and what the reality has done is technology has just shifted what we work on. It’s now something different.

1:36 It feels like we’re in that same excitement of AI is shifting again what we work on. Like oh it’s going to provide all these benefits — I don’t think it is. I think it’s just going to shift what we work on.

1:48 And no, I would completely agree with that. The technology or whatever is available to us just gives us bigger things to do or more complex things to do. I watched the movie from Apple called Tetris. It’s a great movie.

2:02 It’s hilarious because it’s basically showing how they developed it. They’re like eight bits, and it was basic. It was still complex to do, but that was the technology available to them.

2:13 Yeah. You and I are doing things that they could only dream about, but it’s because we have it available to us. So, we’re not just going to keep that level. That being said, I was telling you yesterday, I am so happy that I’m being able to live through this era now with what technologies available to us. Granted, I’m even grateful we have the internet or this would not be possible.

2:47 Just having high-speed internet, I think, is amazing because most of the time I look at this going, “Oh, wow.” If we didn’t have high speed internet, we couldn’t do video podcast. I remember downloading games that would take a couple megabytes and it would just wait. I would just wait for things to download because it was so stinking slow.

3:08 Now I go to my Xbox and I download a game that’s like three gigs and I just wait. Oh, it’s taking 5 minutes to download three gigs. This is ridiculously slow.

3:17 Brother. I remember Kazaa the torrent thing for MP3s. It would take forever. I don’t do it anymore. So, if you’re listening — but that was back in the day.

3:48 This morning I woke up singing songs that I’ve been creating myself for me. Sometimes you get that earworm I guess is what you would call it. Tommy, I’ve made so many songs about coding and data and things and now I don’t sing pop songs. It’s now my own songs that are like the way I like to hear them. The AI has really been able to help me produce a very personalized and specific genre of music that is specific to me.

4:40 I was going to ask you, Mike — did you write a song about DAX?

4:44 Maybe. It came up on my feed and I’m like, who in the — I was like, “All right, I got to look at the creator here.” I don’t always look at the artist. I’m like, “What in the world — oh, okay.”

5:09 Your music taste I don’t think is my music taste.

5:12 I know what you like, Tommy. I believe you like a lot of the Italian songs.

5:25 I want to hear a Sinatra version or give me a classic rock — how about a Guns and Roses and Sinatra mixed together?

5:47 All right. Not only will I do that, I will — if you have a song worthy enough, a mix of both of them, I’ll promote that on all the social medias.

6:01 Tommy, I’m going to promote it anyways.

6:04 For those of you who are on the internet and have seen some of our feeds, there is a song that I just recently published out on YouTube that is all about DAX. And I’ve been slowly making new songs and putting them out there on YouTube.

6:36 Anyways, that being said, Tommy, let’s talk about our main topic today before we jump into the news.

6:50 So today’s main topic is a mailbag. Defining the problem before your tools.

6:57 This is a really good question. We talked a little bit about this in a previous episode, episode 428. We addressed like this is a solution looking for a problem. So are there other areas of fabric that are doing this? What does this mean to define the problem before you start building the tooling on top of it? Or do we just build the tools and try and find problems to fix?

7:39 Yeah. So, we actually have here — let’s start off with our fabric updates for January. Microsoft just released, I believe, yesterday, January fabric feature updates. And let’s have some fun with this. There’s a few worthy items. We have about 17 good ones.

8:09 Well, we’re not gonna talk about all of them. It’s 65 pages. Every blog is basically a book, which is insane.

8:23 Yes, Tommy, I am. But like on the other hand, it was Christmas. What have they been building through December and January? Usually in the past when we looked at PowerBI, typically you get no updates in January at all and this year we got an update and it was minimal but it was like maybe a couple fixes maybe some features that didn’t quite get out.

8:59 All right. I am going to start off. It may not have the highest impact, but this is a feature I’m going to enjoy and it is auto summary for semantic models. What this feature is in the OneLake catalog, you can basically see AI generated summaries that help users quickly understand the purpose of a semantic model without actually opening it.

9:39 The way we do that is it has to have a reference page with two to three questions that report answers. And so this is very near and dear to me in terms of the importance of this. Now how good those summaries are, it’s going to have to take a look at this. But the more clarity we can provide the reason for something to exist and having that easily seen without having to open it is better for everyone.

10:15 I’d agree. I think we’ll see how this plays out.

10:28 I’m going on potential for this. It’s like drafting a quarterback in double A. I don’t know if he’s a blue chip, but the potential here is pretty great.

10:46 I’m probably going to go after — this is going to sound maybe a bit vanilla or maybe a bit too simple. I feel like this should have existed before. So it’s the idea of auto summary for semantic models. I got to play with this feature a little bit. But Tommy, I am really enjoying the OneLake catalog in general.

11:18 I think it’s a solid tool. I like what it’s doing. For me, when I look at the OneLake catalog and what it can do, it’s really for data discovery. I was just in a client’s tenant. They gave me access to some semantic models and I was able to go discover what tables are in them.

11:59 I really think this is an underdeveloped feature and I’d like to see some more development around data discovery, who has access to what things, what’s governance. I think this is very important to be part of the BI team. So I feel like the move here should be when you come into fabric, I’m going to unpin a lot of the left-hand navigation experiences like the real-time hub. I don’t really use that very much. I’ve even unpinned Copilot. I’ve pinned workspaces and the OneLake catalog because I feel like those are the two main places I want to be.

13:16 The other thing that I would pin that I still can’t believe is not a feature yet is favorites. How is that not pinnable to the left-hand side? Like literally a home screen for favorite things.

13:45 I’m hesitating on does favorites become a part of the core product. Maybe I would argue if you made favorites a little bit better — like here’s things that you have favorited and then when you go to the favorites page, you basically provide all the statistics and details around what’s happening on your favorites list. Like how many times is a report viewed and it’s in your favorites.

14:56 It’s like a favored thing, but I need to pin it. I want this thing to stay there. A lot of what Microsoft does is they let you have things and you put them on a page, but sometimes the page just ebbs and flows. There’s some reports that I’m like, I just want that report to stay there.

17:30 Speaking of UI updates and making things easier to see — if you go to the OneLake catalog, it shows a list of everything and usually I have to filter by lakehouse or semantic model because it’s showing everything as its own entity. But we know that most things have interdependencies. If I have a lakehouse, there’s a SQL analytics endpoint and there may be a semantic model. So what they now allowed is a clear parent-child structure to understand how the data relates to each other. This was just needed.

18:54 One thing I’m seeing around OneLake — semantic models are really nice. I can make measure names, descriptions, see data types. When I go to lakehouses, there’s no ability for me to add metadata to that table. How it got there and what those columns are doing. Does the Delta format support column descriptions?

19:50 I did like this — going to semantic models and doing Explore on top of your data. Major win. You can go all the way into the column level, there’s a little binoculars icon, just click it and it just works. This is nice. If you have access, click the binoculars and it shows up and you have the edit and explore my data experience.

20:33 Question about the explore feature because I use that all the time too. I think about the adoption side. For you and I, that makes total sense because usually we’ve created the semantic model. But if I am a CTO or a director or a manager and I have that access, is that something we’re trying to promote as part of the experience for end users?

21:52 When we talk about semantic models, the semantic model in the OneLake catalog needs the model view. One of the tabs really needs the model view diagram. And I need to be able to see the tabs that are saved — I’m dealing with some very large semantic models now and I’m making multiple tabs at the bottom of the page.

22:49 Inside PowerBI desktop or the service, when I’m building these model views, I need like a post-it note on there — a comment on that page. Think about it, if I document it in the model with a text box that has comments and then save the view of how those tables are laid out, when I publish it everyone can see the tables. I can write documentation on how to use the semantic model. That completes a very difficult portion for people.

24:50 I think this is exactly what people want and I do think what you’re describing is a perfect place to put AI and models on top of this. I should be able to go to a dimension table and see the dimensions. There’s a relationship to a fact table. I should be able to communicate with the agent and say, “Using this dimension table, what kinds of things could I calculate?”

25:41 If you click on a measure, I should be able to see a diagram view of every column and where that came from. Like a mermaid chart — I’ve grabbed this column, this column, and a measure. It highlights the measures, the tables. I can see at a quick glance when I put this measure on this visual, these are the impacted elements back into the semantic model.

26:25 Like that would be huge. I might have to just build this thing.

26:32 I think we probably should get to the topic because we’re 30 minutes in and the last 15 minutes have been features that don’t exist.

26:42 All right, let’s go in. Our main topic. Thank you guys and gals for the mailbags, continue to send them our way. Here we go. “I just finished listening to the adopting Copilot standalone for PowerBI episode 428. You may have addressed this before, but I enjoy having hearing a conversation about addressing what is the problem you’re trying to solve for. You can have many tools and be highly skilled with it. But do you understand the issue for which the business needs improvement against?”

27:47 I think this is an incredible question. And I’m not going to pick on Copilot. I think this actually applies generally. If there’s not a need, why are you building it? I’m seeing a lot of people vibe code or build applications. I’m honestly myself doing a lot of building of things. In a world where everyone can create anything they want just by talking to it, the only winners are the people that can really solve a problem efficiently and communicate it to an agent or a team to get something created that actually solves the problem.

29:22 I just spoke with a team yesterday and they were looking into PowerBI. They haven’t considered fabric, but they were going to move to PowerBI. And they’re asking about fabric. I’m not starting with, “Oh, well, there’s lakehouses and data lakes.” If you don’t understand what a team’s current situation is, more importantly where they’re trying to go — not “we want lakehouses” because that’s not the right question. We have to understand what questions we’re asking.

30:07 I don’t have to have a problem to create something now. Usually I wouldn’t spend my time before AI focusing on something just to vibe develop. But even without the AI side, I could migrate everything to a notebook or a lakehouse, but for an organization, you have to understand the issues at hand or more importantly the goals. People already have organization data and now I’m going to give them this fabric environment. Where do you start?

31:21 This whole idea of hyperpersonalization. I’m going to coin that term. I was working with someone recently. They had a problem. We sat down and said, “Let’s build a little web application, a mini app.” Not the full-blown deploy — just an HTML page that does some things. We sat down and vibe coded loading an Excel document, reading the information, filtering columns, applying business logic. All done in an HTML page local on their machine. It was hyper personalized to what they wanted.

32:56 I really like this experience of having a vibecoded experience alongside where all the data lives. Where is this experience from fabric? I’m looking at Lovable, Replit, Base44 — these are doing a really good job. Fabric’s already got authentication solved. Where do you store the data? Fabric is an environment for that too. So where is the equivalent of a vibecoded tool that lives in fabric?

34:30 We’re going to need to start thinking about what is the experience I want to give to my user to interact with data. And the agent can take care of everything else.

35:02 I’m gonna challenge you to take a step back on the AI things here. Most people listening who work in a larger organization do not have the ability to use AI as much as you and I do because we have our own firm. Most organizations are not allowing this. People are still trying to integrate the products already available in fabric. They can only play in the Microsoft playground.

36:25 Yeah, Tommy. I’m gonna agree with you and slightly disagree. Are large organizations actually limiting what you can physically build on a website? I’m not writing — I’m not vibe coding with software. I’m going to a website and everything I’m building is in the website. The accessibility to put AI where I want it, I think it’s there.

37:28 These last eight weeks I’ve really seen an effect and I’ve shifted much of my mental model towards this is going to happen.

37:59 A friend of mine has employees in sales. They’re using ChatGPT to write prompts for sales opportunities. He’s collected multiple prompts and any that work well, he feeds back — “make it better, more of these.” After months of work, he’s now almost one of the lead sales people in his area because he’s been using AI just to bounce ideas off of.

39:05 If you can increase your sales by a lot for 20-50 bucks a month, people better take notice.

39:41 But Fabric only has Copilot. I know it’s coming, but you’re doing exactly what the mailbag question is asking about — talking about all the cool things out there, but what does it actually solve right now?

40:20 A lot of people are still trying to adopt PowerBI. I just had a project scope yesterday. They are looking at AI agents, sure. But they’re still doing what happened 10 years ago — migrating report objects, integrating Snowflake. They’re a large organization.

41:30 What I feel like I’m trying to articulate is there are so many problems to solve. There’s been a tipping point where the gap between what I can dream up with problem solving and the ability to execute has almost disappeared. The cost to build hyperpersonalized tools has dropped to almost nothing.

42:48 I feel like in some ways Microsoft is severely lagging behind in the AI generated space. But I’ve had a really successful experience inside Azure Foundry. I like that I have models — any model, all models basically — everything from HuggingFace, all the big producers.

43:51 Let’s just look at PowerBI. Where is PowerBI Pro Copilot? Where is the creation experience? I think Microsoft has missed the mark substantially. A lot of people think Copilot is used to look at data and give information about data. I think that’s a very hard problem to solve. But what’s much better is how you throw an agent at building a single visual, a report page, a background image. Those are things agents are amazing at.

45:00 Are you going to recommend right now to vibe code semantic models?

45:20 No. But I’m using the PowerBI MCP as part of my regular workflow. It allows me to pull desktop up and start talking to it. In order for me to get the knowledge of Marco Russo and Daniel Oair into my model, I can point an MCP at it and say, “Go look at my model, suggest some measures for me, but only build DAX patterns. If we can’t make a DAX measure fit a DAX pattern, we don’t build it. Period.”

46:42 There’s really two big themes here. One — just because you could doesn’t mean you should. But the bigger thing is one of my biggest assets when I work with clients is the ability to prioritize goals. Every organization has two limitations — money and resources. They want to do everything under the sun, but someone has to prioritize what needs to get done first.

47:45 Where we’re going with AI is that limited resource is drastically reducing. We have more time to build things. A lot of people are now dealing with — I don’t have to focus as much on goal setting because I can quickly start building something. But I’m going to argue the importance of goal setting because I fall in this trap too.

49:36 Well, I haven’t set what my projects are, what my milestones are, what we’re trying to do in the long run. All these things just make it easy to forget that.

49:53 As I work with agents, one thing I’m going through is trying to understand what the agent can do. If you do a better job of PMing and breaking down the problem to a much smaller state — the business analyst space has to spend a lot of time talking to business and doing technical things. If I had to write no DAX, if I could just have all that become task-level things I could offload, I could sit back and think about the big picture. That would be great.

51:28 The vibe of this conversation — we need everyone to focus on solving problems. There’s a shift happening on what tools we’re using. What if I took the time to make a semantic model and cut it in half? What if I was able to take that time and cut it by 75%? That’s the stuff we’re talking about with AI and agentic things.

52:28 I’m looking at GitHub Copilot with sub-agents. Also Claude Code with their bots — things where you can let them autonomously think and program. Tommy, I could literally buy a $10,000 desktop computer that can run these models locally. I could treat it like an employee — exactly like a new hire. Agents are now building their own memory. They’re creating information and storing it locally so they can recall it when needed.

53:31 I could give every one of my employees an intern for a couple thousand dollars to buy a machine. This thing works 24/7. There is no insurance. There is no downtime. It just runs. The AI is doing two things — reducing the time it takes to build things and making the cost a fraction.

54:41 Here’s what you’re trying to say. The greatest value, the greatest gift that these new tools give to us is not the ability to create things, but it’s the time to be more strategic. A lot of people fail to see that. The greatest skill I have for an organization is not my development or prompting skills — it’s to take all these tools, the arsenal, and spend more time thinking about the strategic side. At the end of the day, you can build 10 things, but if the business doesn’t see the value, what’s the value?

55:45 If you’re in the space now — learn from someone on what it means to create value for the business. What does it mean to be strategic? The best way to find out the problem — there’s really two things: an issue or a goal. I always ask clients — imagine it’s a year from now. Your team is high-fiving. Why? What are those things that made that happen? Not what products are in use, but why would your team be so happy?

56:48 I can now use AI to focus more on getting to that vision. But all these tools are a means to some end. That’s the biggest problem with all the new things. I go on GitHub and I’m looking through the explore — “oh that’s cool” — and I’m building that as the end. You need to focus on these tools as a means to some solution.

57:41 100% agree. And this stuff’s all changing so fast. I’m going to go back to my earlier comment. There’s an Excel expert version of people and there’s an AI expert of people. When we step into an organization, we can identify almost within minutes — who are my Excel experts? They stand out. The way they talk, the way they communicate about what they’ve been building.

58:31 Tommy, we love working with those people because they’re prime. They already know how to data engineer. They’re just doing it in a medium that we’re going to change. We’re going to move you from Excel to Power Query, from Power Query to pipelines and lakehouses. The analogy holds true for AI — the same exact pattern.

59:31 Those individuals are already solving problems. They’ve discovered AI. They’ve figured out how to prompt things. Going back to my sales example — his results were way above everyone else. At some point you sell so well that your role changes — you become a leader and manager giving the prompts back to your team.

60:34 We’re inflecting. The inflection point is starting to happen. I don’t think AGI is happening anytime soon. But the tools and systems — we’re finally treating agents and the context window more like a human mind. We’re going to do a better job documenting, skilling, giving it tools, giving it capability.

61:36 When I look across these other companies, there’s just a lot of other things I’m seeing that are like, man, these other companies feel like they’ve got it right. And I am just frustrated that I can’t do it in fabric.

61:49 I completely agree. If someone ever wanted to hire me, if I ever went back to the FTE space, I would walk in and say, “Look, I’m not working without you giving me Claude, OpenAI, open source, and Perplexity. And if you do that for me, you will see the most efficient — working as five people in one.”

62:29 Say the list again. Claude, Claude Code — I need that access on my terminal to build. And Perplexity. You do that and I’m working like five employees.

63:00 I would not — just working independently now and being able to be as productive and as efficient and strategic with projects as I am now. It’s like going from five screens here to someone giving you a laptop, no screens. “I’m going to tie your hands behind your back. Type with your nose on the keyboard.”

63:22 I’m going to give you my set of tools. I think I could get away with everything you’re doing with Claude Code, OpenAI, Proxy. I’m very bent on staying inside the Microsoft shop. I’m very adverse to sending prompts to other companies where I know they are looking at my prompts. So would I run their models on Microsoft hardware? Yes. So I get access to all the Claude models through Azure AI Foundry.

64:30 I don’t use Perplexity, but I just saw an announcement from Google that’s almost like a Perplexity killer. We are in the throws of large software companies jockeying for position. Google is coming in saying we make enough money on advertising, we missed this AI wave, we need to crush it.

65:35 I feel like a war is brewing across the big tech companies. The next three to six months — there’s going to be a lot of brand new tools that are free or almost free. I saw an advertisement that ChatGPT will start serving ads in threads. That’s coming.

66:54 We are getting over time. Thank you all for listening. Great question from the mailbag. Should we be looking for tools or should we be defining the problem first? 100% we should be defining and understanding what the problem is. The tools at this point are becoming immediately irrelevant.

67:30 We are two episodes away from episode 500. We want to make it fun. We will be doing giveaways. We’re going to have some free stuff. Make sure you send the message around for episode 500. I’ve got a couple surprises too.

68:01 You can find us on Apple, Spotify, wherever you get your podcast. Make sure to subscribe and leave a rating. Got a question? Head over to powerbi.tips/empodcast. Join us live every Tuesday and Thursday at 7:30 AM Central.

68:53 Thank you all so much and we’ll see you next time.

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.