Start with the Why – Ep. 477
In this mailbag episode, Mike and Tommy return to the foundational question that should precede every data project: what is the problem you’re trying to solve? It sounds obvious, but the rush to adopt new tools—Fabric, notebooks, AI—often skips the most critical step. They break down how to frame problems properly before jumping to solutions.
Mailbag: What Problem Are You Solving?
The Anti-Pattern
Mike and Tommy see this pattern repeatedly:
- “We need to move to Fabric”
- “We should use notebooks”
- “Let’s build a lakehouse”
None of these are problem statements. They’re solutions looking for problems. The first question should always be: what decision are you trying to make, and what data do you need to make it?
Starting with the Why
Before any technical work:
- Define the business outcome — What changes if this project succeeds?
- Identify the decision-maker — Who acts on this data?
- Map the current pain — What’s broken, slow, or missing today?
- Scope the minimum viable answer — What’s the smallest thing you can build that delivers value?
Tool Selection Comes Last
Once the problem is clear, the tool often selects itself:
- Need real-time monitoring? → Real-Time Intelligence
- Need historical trend analysis? → Lakehouse + semantic model
- Need ad-hoc exploration? → Notebooks
- Need executive dashboards? → Power BI reports
Starting with the tool forces you to justify the tool rather than solve the problem.
Consulting Reality
Mike shares that the most valuable conversations with clients happen before any technical work starts. The “requirements gathering” phase isn’t paperwork—it’s where you prevent months of wasted effort building the wrong thing. Tommy adds that clients who skip this phase almost always circle back to it later, having spent time and money on misaligned solutions.
Looking Forward
“Start with the why” isn’t a new idea, but it’s one that needs constant reinforcement—especially when shiny new tools create urgency to build before thinking. Mike and Tommy encourage listeners to push back on tool-first mandates and insist on problem-first conversations.
Episode Transcript
Full verbatim transcript — click any timestamp to jump to that moment:
0:03 Welcome back to the Explicit Measures
0:34 Podcast with Tommy and Mike. Hello everyone and welcome back to the show. All right. And yeah, it’s weird, man. It’s just you and me. It’s just us. We’re just doing this is a this is a Tommy and Mike only episode. This episode also is being pre-recorded. So for those of you who are already on watching or if you’re part of our membership, you can already see this video early. But if you’re not, you’re going to see this when it’s recorded video. Tommy and I are are traveling a little bit. Again, this is around the holidays. We do the best we can to keep consistent with episodes even through holidays and when we’re traveling and
1:06 Speaking and things, but sometimes we miss a couple. So, so far we’ve done good. We’ve in four years we haven’t missed any episodes. Yeah. The only reason you may not see an episode come is because of yours truly here forgetting to upload the dang thing. But it’s always on YouTube. So, we got a lot better. We’ve got a lot better. Well, [laughter] we never missed an episode. Technically, true. The numbers all are all consistent. We’ve even we’ve even flipped numbers around once or twice to like because we’ve screwed up some things. But I think we’re we’re good to go. This is a recorded episode. We’re going to jump in. before we jump in,
1:38 This is not really news items. Maybe it is a little bit Tommy, but maybe let’s let’s get into the main topic and maybe I have a comment around here around the the initial part of this question today. Today’s a mailbag. Tommy’s going to read our mailbag today for us and then we’ll get into the topic. Tommy, to you. Oh my goodness. Do we have a good mailbag? So, we’re going right into the mailbag question then. Sure. Because I think my I think my introduction or my notes or my my thoughts here are related to like the first part of this question. So, I got immediately distracted by this question
2:11 And and saw a word in here that I wanted to just hang on for a bit and talk about and then we can probably get into the actual question. So, let you let let’s let you read the question first. and then we’ll get into my little anything we were talking about offline. Yeah, it’s literally this is going to be like the first Here’s the interesting question. I’m going to give you a rabbit trail immediately on this conversation. Lost our keys already. So, exactly. Tommy had a weekend issue around his car keys. But we found that we literally found them five minutes before we started recording. So, this is great. But hey, for those who are putting mailbag questions in, if you were like, “Hey, no
2:44 One they haven’t answered my question lately.” We had to update the Excel spreadsheet. So, we’re seeing all the recent mailback questions come in. So, expect a lot more mailbacks. So, we see your questions and we’re going to be answering them. We do. So, let’s get into it. I just finished listening to Adopting Copilot Standalone for PowerBI episode 428, which is crazy because that’s 50 episodes ago. You may have addressed this before, but I’m I enjoy hearing a conversation about addressing what is the problem you are trying to solve for. You can have many
3:18 Tools and be highly skilled with but do you understand the issue for which the business needs improvement against? Oh, this is a great topic. Okay, awesome question. This this really goes down to our, , our our core ethos, our core mindset, which is if this report doesn’t make you money or save you money or if your data engineering doesn’t make you money or save you money, why are you doing it? But before we get into that one, I want
3:50 To I want to quickly go on a rabbit trail here. So this question here was around adopting co-pilot. Copilot was the note here. And I just want to just generally pick on co-pilot here a little bit just Tommy from your perspective. First time. Yeah. The first time we’re ever picking on it. So let me Yeah. Right. Tommy, have you used are you using C-Pilot today from like the mic? So there’s a couple co-pilots that we’re talking about here. We have like GitHub copilot, we have PowerBI co-pilot, and then we have Microsoft M365 copilot. There’s probably more things that are
4:23 Co-pilotesque. But my question to you, Tommy, is are you using the Microsoft 365 co-pilot today? Currently, the only instance that I find myself relying on it, and that’s what I’m going to consider using here. Okay. When I say use doesn’t mean I dabble with it is a part of my workflow. Yes. Is the facilitator in Microsoft Teams for meetings that I cannot interesting. So, okay. Yeah. And that’s but that’s really the only use case where it is part of if I do a meeting I immediately check the
4:55 Recap I have a confluence page I put all the notes with I send it to my clients that’s what I use and that is my use case of co-pilot today. So to be clear then let me I said three co-pilots. I said Microsoft 365. I said PowerBI and I said M365. I don’t know what the three were where whatever. they are you just brought up a fourth one though because the that one is not it’s still co-pilot teams premium but it’s teams premium. So like that. So the and the reason I’m asking this
5:26 Question is I’m like I’m trying to look like all the licensing that’s going on. So like you could have copilot for M365. That’s one. You can have co-pilot for PowerBI. You can have this. Anyways, this is another co-pilot that’s on top of all the other co-pilots. Correct. Correct. Okay. So, I will also echo your sentiment around that co-pilot as well, Tommy. I really like the facilitator one. It does a great job. I love the fact that it takes notes in line with the meeting and it summarizes the meeting and then it kicks out a one note or a loop document at the end of the meeting and puts it
5:58 Back into the meeting for people to see if they have access to it like right really. that facilitator feature is actually topnotch. I did like the the team’s co-pilot experience. I’ve I’ve kept it and left it on for a while. It was okay. Like I like the fact that it puts videos in my stuff. I like the fact that it translate the videos with transcripts. My only down my one of my downsides here is I feel like I I don’t I’m not smart enough. I’m sure you can get to it. I just haven’t figured it out yet. I’m not smart enough to be able to say when I put a video in, , Clipchamp or Stream, it takes the video,
6:32 It transcribes it, and even gives you all the details of the video, like who was talking when, like a lot of good analytics, but I don’t know how to get access to the video or the transcript to go use in other programs. I don’t know how I don’t know where it I know it puts it in one drive. I just don’t know how to access those things. And I think I’ve heard you can do it through Microsoft Graph API, which is great. I just need to make another like series of API calls to go get these more advanced features out of the transcript. But facilitator I do like and loop is my favorite now. I’ve totally ditched I
7:05 Don’t I no longer use OneNote anymore. One note’s dead to me. Like I only use I only use Loop now moving forward. It’s so much better. Yeah. And I’m not gonna I’ll try to talk or not talk too much about non-Microsoft products here, but there’s an equivalent to Loop called Notion. I’ve heard about this one. I haven’t played with it yet. And I don’t know why I’m non-Microsoft on some things, but for me, it’s it’s the same idea. I actually Loop got a lot of their inspiration from Notion. The slash, the commands, the blocks, the
7:37 Segments, so to speak. Yes, their AI is incredible in terms of with both of those circumstances, facilitator and what I’m doing in notion or what you’re seeing in in loop. They specifically solve a problem that you already had and you because that’s what’s the change with facility. Correct. I always had issues with meetings and the notes and then you have to recap that either took time or I didn’t catch everything. Correct. Those are two problems I always have. Most people always have. It takes time to do the recap or you didn’t catch everything. Facilitator doesn’t do more
8:11 Than it needs to do. It does exactly what it should. And that’s where the big win is for me. I don’t need to do 18 things. I really need it just to do two. And this is my my challenge with this one, Tommy, is like if as I as I start building like again me as a as a contractor or a business owner here, right? I have I form lots of relationships with many different customers all at once. One of the things that I feel is missing is I I’m starting to like centralize all of my communication into one platform, meaning I go to like dynamics and centralize things. Co-pilot helps me out there, too. So, the reason I bring all this up,
8:43 Let me get to my point here, cuz I’ve been dancing around topics here. So, we talked about adopting Copilot. Really like that episode, but one of the things I just recently watched was All Things AI with Sam Alman and Satia Nadella, a Halloween special from the podcast. , I think it’s like B2G and there were some really interesting Brad Brad Gartsner I think is the the host. I saw some clips, some shorts of this over the weekend that were like, “Oh my goodness, things with co-pilot or
9:17 AI in general.” Like there’s a lot of this, , is it hype? Is it going to is are we in a bubble? Is the market over pushing all these things for these AI pieces? And one of the things that Satia said and I don’t think Microsoft is actually in this predicament but it sounded like Satia was saying the investment that companies are making directly into AI is so large that companies have the chips they can they can use them but they have like
9:48 GPUs or processing units on order sitting in the in the wings but they can’t use them because the facility the building that they need to go in can’t be built. And if they did buy build the building, they don’t have the capability of being able to turn on enough power to run these huge data centers. And we’re talking like data centers of, , it’s not just a data center. It’s a data center with like running, , half a million homes.
10:20 It’s it’s something absurd. Like the amount of electricity that’s required is humongous. And so I think we’re we’re hitting this point of this there’s a huge in influx of like all these organizations, large organizations. Amazon’s Palunteer has just I think just announced they landed the the largest 500,000 GPU processing facility somewhere in like Tennessee or something. I don’t know, Kentucky somewhere. But these these buildings are massive and they’re huge computers and they are consuming an inordinate amount
10:54 Of power. And I guess back to the point here, Santi was saying like we can buy the chips, but the chips are literally going to sit on the shelf until we have enough power to actually power them. So they still need to get power. Like there’s all these points in time that you need to have, okay, we got to have the chips. And then when you have the chips, you got to have the building. Once you have the building, you got to have the power, the transformers, all the other. There’s all these downstream effects of like things that need to be in place for these co-pilotike experiences to run. I don’t think us paying 20 bucks a month is very far removed from what we’re doing. I find it interesting that you’re bringing
11:25 Up basically the solution provider side and then when we’re talking about the I think what we’ll be diving into about the the the internal side the one who’s going to actually consume this right so you’re talking all about the people who are creating it the provider of the solution Microsoft open AI and other companies who are going to be creating solutions Nvidia all the people play all all the people in the AI like pipe stream like like but I really do think one side of the problem. Correct. That’s one side of the problem. And I think what I’m looking at from I
11:58 Look at from a technical standpoint. Like I look at that side of the world and say, “Wow, that’s such a major big problem to solve and it takes a lot of capital to do that.” So smaller companies are going to be very challenged to get into this space and compete. Oh yeah. Tommy’s not going to go stand up a company and build a data center. Like there’s just too much. My PC may be pretty strong, but it ain’t going to do that. Yeah. No. Where I think things get really interesting here is when these models get efficient enough or people are able to then start running models locally or on their local devices. I the day we
12:30 Start running models on our phones, I think it’s like game over. Like that’s that’s going to be amazing. I I completely disagree with that. I think this is let’s go back to something we’ve mentioned before on the podcast where what we’re getting into right now when it comes to what we’re trying to solve is this conductor’s era. And I I love this phrase here of what we’re getting into because I am not going to be building my own local model as a service ever. Most companies aren’t. You misheard what I said then. Okay. Sure. You’re you’re taking this down a route that I did not say. So
13:02 I didn’t I didn’t say we will be building models on our local machines. I think that’s going to be left to like the big companies with the lots of processing units. Like I’m not going to go be able I’m not going to ever be able never going to train them. However, once the models have been trained, running the inferences of them, that’s the part running the models, not training them, that’s there’s two parts of this. The running side, I think that’s when like you still need a graphics processor unit. You still need these big machines. I don’t think we’ve gotten there yet. The technology hasn’t gotten far enough yet where we can start
13:33 Running the inferencing side on our local devices. And I think our devices are going to continue to get more powerful and they’re going to change. The architecture of our phones and computers is going to continually change such that there’s going to be more generally available GPU pieces inside machines that are going to let you run these models locally. And I think to me that’s the major unlock, right? That’s when the price on the that’s when the price on the floor these things can just drop through the floor because now I don’t have to have I have to have the
14:05 Big data center to build the training for the item. But now that I can run it locally on these other devices and get fairly good performance out of them or they’re like models that are hyper tuned to a specific thing, right? It’s not these massive very generalized models. it’s very specific to a thing that when we get to that stage, that’s when I think things really change again for us because now we don’t need these massive paid services to use them. You just pay your licensing fee or you get the app or download the app and then it runs all the AI stuff locally for you. So, so let’s go down that that particular path here and let me ask you
14:38 A question. When that is the case, what problem does that then solve? It’s a cost problem right now. It’s a cost problem. Right now the I think Open AAI is worth and again I was looking at some numbers on them. Open AI is worth $500 billion or something crazy, but the amount of money they’re the revenue they’re generating right now is like 18 billion. Yeah. So like there’s a huge disconnect on how much revenue they’re bringing in versus how much they’re spending and how much they’re building. So what they’re doing is they’re a loss leader. They’re losing
15:10 Money every quarter in order to keep building out the technology to make it better because at some point it becomes the de facto standard and then when more people adopt it, they get more money in the in the door. People clearly say this is valuable to us. Instead of paying 20 bucks per user, it goes up to like 50 bucks, 100 bucks and that becomes the standard. So they’re going to have to raise their prices on something or bring the cost of running everything down lower so it starts balancing out. The problem we’re solving here with this one is a scalability issue, right? Of all the people in the world who need
15:43 AI to do things, there’s not enough companies in the world that could build large enough data centers to do all of it. You’re going to have to start figuring out how to do this is the same thing with like IoT and distributed [snorts] compute. And so you start communodities and you can more commoditize the the asset that means people can use it more. it can be easier to deploy. And I think what happens then is you get a scalability remove or or blocker that’s removed. That’s what I’m that’s what I’m thinking
16:14 About. So for whatever reason, I just keep hearing the drum beat here when you’re talking about there there’s that one side of the coin for the company who’s creating these solutions. Correct. But that does not to me have any correlation with an organization adopting it. So and I’ll take this. Okay. This is a different question. Yeah, I like where you’re going here. Okay. Right. Because there’s two sides of the coin in terms of this being profitable or use having a use case here, right? So even take PowerBI, man. Like let’s go back to PowerBI. One of the things
16:46 You’ve been saying on the podcast since the beginning and has not changed is we’ll create more data in the next two years than the history of mankind. I still use that and that you’ve been saying that since 2021 or earlier, but the pro so we’ve never had a problem from a scalability point of view. We’ve had PowerBI for 10 years and yet a lot of organizations even today still struggle with adopting and finding PowerBI as an investment not a cost right so even and we’re not dealing with scalability or GPU issues with PowerBI and again I’m not even talking about
17:20 Fabric here we could bring fabric is another variable to the equation but we could literally just focus on the PowerBI situation and you and I both know very Well, a lot of organizations are still trying to get up the mountain and not understanding why people aren’t using it the way they should and why there’s still this mistrust and even though we don’t have any scalability issues, it’s an established product. All the things you’re saying, we are still dealing with the issue just around PowerBI. We can bring again happy to
17:53 Bring fabric into this, but I it’s a good it’s a good ex evidence here just with PowerBI. Well, I agree with you, Tommy. Like, yes, but it also depends on what industry you’re in, right? Are you in an industry that is having a lot of competition around data and analytics, or is your company comfortable just saying, “We’re going to keep doing the same old same old Excel and kicking things off.” So, if you’re in an area where your workflow is not really AI related, you’re not adding a lot of
18:24 Value out of it. , I would argue I I some someone’s going to come up with something cool like this that’ll do it. But like if I think about like the trades Mhm. like calling [snorts] people, relationship based, like you got to go there and do work on a site, those things, pling, electrician, like those are super important. But do I see a lot of PowerBI influencing a lot of like the day-to-day work for the line? Like probably not. But if the industry that industry takes on a lot more analytics space and does a lot more analytics around it, then
18:56 You’re more competing with other companies or doing more analytical things. And again, there’s probably like bright spots in each of these industries or areas. But I think pro talking to what you’re saying though is I agree there’s a lot of companies out there and to your point most of the large Fortune 100 companies I think all of them I think Microsoft said all Fortune 100 companies except two all use PowerBI like they all have them and I’m I’m guessing the two that don’t is like AWS and like [laughter] and Google like I don’t know maybe there may be some other ones that just don’t
19:27 Have it up at that highest level. I never thought Yeah, that’s pretty good. They probably I don’t know like maybe maybe there is I I’ve never anyways regardless though like when you look at what’s going on right in the space of like who’s using PowerBI like the top dogs are all competing on on the analytics like they all have that competition when you’re in your niches or your industries or your specific spaces the com the competition may not be as rigorous around data analytics but I think this actually pulls us back to our question today a lot of which is regardless what tool you
20:01 Use, let’s just ignore the whole tool aspect and let’s just focus on, , what is the why? Why are we doing things? And I think for me right now as becoming more of a software development company than just a pure consulting company, as I look at this, when you start really focusing on like marketing, marketing analytics, those things, I think it really helps you identify like what is the reason why people need your software? What is distill your idea down to like the most concise piece like what do you need to do in order to drive action and then once you like if your objective in sales
20:35 Marketing is get more sales, get more customers, get more like whatever that funnel looks like, you have to identify what is the funnel and then you have to be able to do things that make data that increase the end goal, whatever that end goal looks like, right? , and I think when I look at this, it’s the why, right? I only have a couple customers. The market market opportunity is very small. Well, therefore, in order for our business to make more money this year is to increase the amount of spend per customer. Okay. What actions are we
21:09 Taking to do that? That’s the why. Like so if we focus on the why everything else to your point Tommy falls out of the space from that and says okay now that we understand we want more revenue per customer in order to do that we have to build more products or sell more things or whatever those other objectives are that then informs you now understand the why you can now build reporting to help you track to those results those end games. I’m going to take this a little further and from a cultural point of view and I think you and I have both seen this in our space
21:41 For and it just seems to be always a continuous stream of very similar situations where PowerBI is becoming the tool choice but the reason why is it’s a and we I’m going to echo something we actually talked about with Chris when we’re talking about real time. We have a solution in search of a problem. And this is the continuing drum beat that I see at a lot of organizations where we h we know we have PowerBI. We want everyone to use it. And that’s just because we should, not
22:13 Necessarily because it’s going to solve all these different things. And the problem with that type of mentality, especially in a larger organization, is that mindset starts or the decision starts usually from the top where we’re going to again plug in any tool you want to your point. We’re going to use XYZ, we’re going to use Copilot, Tableau, Tableau, whatever, whatever. Maybe insights. It doesn’t matter what it is. Yeah. And of course, we’re going to use product X for our new processes. Well, the problem with that is all the people who actually rely on the previous tool, anything legacy, what the current
22:47 Processes is. Well, you’ve not bought them in into why we need to use this tool, why this is better because any product or technology that you bring into organization is going to require process change and skill change. You’re either going to have to bring people in, upskill people. There’s no technology you can bring in without requiring those two. I’ll agree to that statement. That’s that seems Yeah, that’s right. That’s spot on. Mhm. Yeah. And I think this is the big thing that I this is it’s just more with fabric because right now I think fabric is very much for a lot of
23:19 Organizations a solution in search of a problem but PowerBI was too and and to a lesser degree PowerBI was absolutely too and AI is that with steroids Barry Bond style. for for those who are listening and for Mike. Barry Bonds was a really good baseball player, started doing steroids, did really great, but he also looked like he had a basketball for a head because Yeah. So, yeah. So, it was very apparent the man was, , it wasn’t all it wasn’t just protein on the roids. Yeah. On the rosids. That being said, AI is that. But we’re still dealing with
23:53 These problems where someone the CTO or even the CEO I’ve seen this where the CEO had a agency come and say, “Hey, you should do this.” He said, “We’re going to do it.” But there wasn’t an initiative. It wasn’t like, “Hey, we’re going to start using PowerBI, whatever it was. We’re going to start using PowerBI and it’s going to solve these problems. We have a year initiative to do so. Here’s what we want to change.” It was, “Hey, moving forward, we’re using PowerBI. Please yada yada yada. Move your stuff over, which for a lot of people, you they’re not robots. It’s not just changing agents or change, you
24:26 Know, directing the source. You are now telling people, oh well, okay, those Excel files are this. Now I have to change my process, change my day-to-day. You’re dealing with a range of people from age to skill which are may not buy in and happens all the time. Next thing , you’re six months into this and it’s a very convoluted space and it just like spending money as a cost on something like PowerBI, which is becoming more and more apparent with fabric. And I think this is the big I’ll start I’ll stop there but this is where
24:58 I see the big issue which is a very common stream with most organizations. I I like this point Tommy and I think this is very true. I think if you don’t if you look at the PowerBI adoption roadmap so in the PowerBI adoption road mapap it talks about when you’re building your first proof of concepts around PowerBI it’s very clear and I agree with this one. You should start with a clear objective in mind. Right? I have a data system, business objects, Oracle, whatever. We’re going to replace that with a new BI platform.
25:33 The new BI platform will be, let’s turn those things off. Let’s add semantic models. You the concepts I think sometimes people look at this going, oh, PowerBI, it’s this brand new thing. It’s all this new stuff. Like, no, no, it’s not. It’s still a series of data marts. the terminology and the tools that you might be building in might be slightly different but it’s still in my opinion it’s still a data warehouse you still have all the information coming to a place and if you have an existing data warehouse and fabric it’s not really replacing any
26:05 Of that your concepts still exist you still want to star schema or model the data as the business would understand it none of these principles change and sometimes I feel like to your point Tommy organizations show up and say oh we got to do everything differently this is all new. We’re going to use this new tool. It’s going to make everything that much more effective. But what was lacking before was process and proper business understanding and understanding of like how the data is needing to be presented to the business. And then what happens later on is well, if you don’t
26:39 Define the why, why we’re doing it, what tables exist, what how does our business run, right? What are customers? What are accounts? what are our products? If we don’t define those relationships between the simplicity of your business, doesn’t matter what data platform you put in in place, it’s still going to be a mess and it’s still people still will like try to export things out, get to Excel and go from there. So, I agree with you, Tommy. I think in a lot of ways the lack of vision for why we’re doing
27:13 Something distracts a lot of companies. And I think this is the same reason why co-pilot is struggling right now. So if we look at copilot as well as like another let’s okay let’s talk about powerbi but now powerbi and the idea of co-pilot where does that fit right and again I’m exploring a lot with AI AI is moving so fast right now it’s very exciting but if I look at the copilot experience when I look at all these other companies open AI the other one that we like is claude a lot of claude code things right
27:45 Anthropic poll doing all these really interesting things so all these all these AI companies are doing these really interesting things, but I feel like Microsoft is continually behind the mark of where these other AI things are being produced. And when I look at other products, non-Microsoft ones, things like Adobe Express, I look at Canva, I look at Miro, their integration with AI seems to be really interesting. It’s extremely impactful and it see and when I play with those, I’m wowed by the experience.
28:19 There there is a difference though there and and [clears throat] I don’t normally say give some Microsoft a little leeway here or or give some Microsoft super and but with Canva with each of those tools that product only does a really specific thing right like it’s only meant to do one thing what what do you do in Canva you do designs that it I understand but like it’s the same thing to like Excel it’s the same thing for like PowerPoint it’s the same thing for like PowerBI like all of those tools do a very specific And Canva does a very specific thing and
28:51 Adobe Express does a very specific thing. But in all those tools, I see other companies putting in AI in a way that it’s like it’s easy to use. I can understand what its purpose is. It seems to make sense with what’s going on. It feels to me like they understand a better why around what the AI is supposed to do. Like what’s the why? Why are we putting the AI into it? And on the other side, I see Microsoft doing a lot of things and it’s getting better. I think the products are getting better with AI pieces, but when I look at the Microsoft side of things, I’m like, there’s really no why here. Like, it it
29:24 Feels a lot more disjointed. , other than maybe it it adds some value maybe when I’m in an email, but even then, I’m in emails. It’s not really the worst. It’s not really helping me a a ton. Now, again, I’ll be admitted here. There’s there’s places that I do think it’s doing a good job. I think AI has done really well or Copilot is doing pretty well in what I’ve been using in Dynamics recently. I’ve been building dynamics out and putting my customers and information in there. So, I could do a little bit of chatting with that. But there’s other things where I’m like, I it doesn’t quite do what I want. It needs it needs more. And now I’m feeling
29:58 Like everything that Microsoft provides to me is like, okay, well, now that’s just an agent. Go build a custom agent that does what you want it to do. And then I have to provide all these extra additional instructions. the need though. But that’s I I think this is Oh, this is you’re making a great point here because with PowerBI and I think especially AI, Steve Jobs warped our minds and I think everybody’s mind when the phone came out because one one of the very common terms or very consistent things he Microsoft and or Apple would always say was it just works. You buy the you buy the Macintosh, you buy the phone, it just
30:31 Works. And most marketing messages are geared towards that. Here’s PowerBI five clicks the wow AI, , just it’s a chat. It’s a little open chat. Just type whatever you want and your problem solved. But that’s a f that’s a fallacy. That’s a fallacy on that. It just works that easily. Now again, I will make that distinction here that we have to consider there’s two types of AI for the consumer or for a consumer. There is the individual the micro side of this Mike Carlo using chat
31:05 GPT on his own and then Carlo consultants or Mike Carlo the company using an AI they are going they have to serve different uses and they have to do different they have to have different services for them to work okay Mike probably has a great benefit you probably have a great time using chat GPT on your own you probably have no issues with that or very minimal but trying to use it in your organization That’s where probably we’re talking about the generality of this. And again, when we’re dealing with AI, if it doesn’t do
31:38 99% of what your expectations is, you’re probably not going to use it on a consistent basis. This is why the beginning of the conversation, you asked me about those co-pilots. And the only one we said was facilitator because anytime I do a meeting, I’m checking the correct? I I do not use the Outlook one because not every time do I create an email am I using the AI and the times I have I’ve not been satisfied enough not saying not satisfied at all but satisfied enough to meet my expectation. You put this on an organizationwide
32:11 Scale you’re going to run into some some issues. And I’ll start with that. And if I can take it a little back here too because we’re I know we are we are a PowerBI podcast or fabric podcast at the end of the day. And it’s funny we’re talking about we’re talking about technology right now. We’ve talked about AI, Copilot, PowerBI. I would make the argument and I want you to consider what we’ve talked about for the last half an hour. This is the same problems people have even when they say they want data governance or or projects like that in
32:44 Their organization. These are the same situ these are the same issues or complaints that I get. People want data governance but they don’t know what it’s supposed to solve and they don’t know what it’s supposed to look like. I think you can plug that in when we’re talking about co-pilot and I think you can plug that in when we’re talking about fabric. People want fabric but they don’t know what it’s supposed to solve and they don’t know what it looks like. And I I was just dealing with a client that we were doing a data governance project and the major the first half of the entire project was just getting people on board the
33:16 Communication on the expectations of what it’s going to do, who’s going to be involved. Now that it’s actually set up and up and running like they’re good like I can monitor this but they needed so much on what are we actually doing here? That’s how I’ve seen things fail. So, you can plug in a lot of our own initiatives that we’ve done for the past 10 years with what we’re doing with Copilot today. , yes, I agree with I agree with you on that one. But where I’m still having some trouble on
33:49 The gaps here is like I’m just when I look at like the landscape of other AI tools, it there’s there’s a lot of other different it feels like the the why as what they’re doing. Going back to our topic which is why does this AI exist? What does this tool do? What am I doing in the like what are the actions that I need to perform and how is this AI making my stuff smarter, faster, easier? Those seem to be articulating better in other places and not so much in the Microsoft space. It just really feels like every feature that’s coming out is trying to
34:20 Get tied onto something that is AI and we’re just trying to build a bunch of features and see what sticks is what I feel like it’s happening right now at Microsoft. I don’t feel like the same thing’s happening on these other tools, right? So, , I I have a So, another one that I’m really enjoying right now is NAN. NA is another tool. It’s an automation. It’s almost like Power Automate, but it adds the ability for you to add a large language model in the middle of it where you can call APIs, make a connection, do things behalf with with an agent on things. And
34:53 So I feel much more confident using that program to stitch multiple actions together. Okay, I need something. I’m going to send a message to the nan. It’s going to then take that message and from there it will then process the message down. It will then understand who the summarize the message in the email body, create a contact, create an account, go search for those two contacts and accounts in my existing system, write them in and like do all this other stuff. So to me, there’s this missing part of like yes, Microsoft seems to be building like a framework of
35:27 Things, but for actionable tools that come out of this, I don’t feel like there’s they’re not really addressing the why. And so I think you’re finding a harder time to adopt co-pilot and getting co-pilot out to the organization because I think to your same point earlier, Tommy, when we’re talking about this one, you said just [snorts] like PowerBI is corporate decides we’re going to use this tool. it’s going to happen. Let’s do it. Instead of saying stepping back and saying we need to identify proper use cases where the AI or PowerBI should be used, this tool solves this
36:01 Problem. And I think there’s very clear especially around PowerBI. There’s very clear objectives you could clearly point at and say this solves a problem. We have a lot of data silos. We need them all in the same place. PowerBI and fabric solve that. Let’s identify some data silos that we have today. let’s move those teams and their data into fabric. That’s a great use case. So start with something like that that identifies a clear why of what you’re doing and then from there you can let the creativity of the business after you get that initial milestone completed
36:35 Then the company can go in and say let’s build all these reports let’s build a bunch of pageant and stuff and that now facilitates I think the the allowance to spend money to train people to use this new thing. And again, we’re we’re right now dealing with and again a study by Gartner said 95% of Gartner or AI projects are failing at organizations. Exactly to your point. So that means there’s only a 5% success rate. And most organizations who are trying to adopt AI tools for the organization, the majority of people are still using their own individual
37:07 Subscriptions for AI rather than using the organization’s tools. So we’re there’s this huge divide right now and and again I think a big part of this we talked about the solution to search of a problem and it’s an initiative based thing to me where I think the common issue is where you just mentioned that the organization as a whole says we’re going to do AI we’re going to adopt X Y and Z or co-pilot whatever it is and let’s let’s just focus on co-pilot I think this is probably just a really
37:39 Good place to focus and the problem is it does require especially for an organization extra finetuning whether’s especially with data a with co-pilot studio I don’t think a lot of organizations realize that the a generality is a general solution is going to answer a general question but you’re not going to have a specific problem answered by a general solution if I just use co-pilot the chatbot or the conversational piece. It’s meant to
38:13 Be general. And I think even the ones in Excel or the ones created by an organization, if they use that general co-pilot, if you ask a general question, you’re going to get a general answer. And that’s what co-pilot’s meant for the top level, but they don’t realize the investment needed first. First off, actually, I’m not even going to start with the investment or the skill. to your point and I think to the mailbag here have has the organization gone through an initiative of what they see co-pilot solving for the organization in a year’s time what is this going to
38:47 Solve where are we going to find these breakthroughs whether it’s through our operations team with all the paper they deal with or the sales team to be more efficient with their accounts what do we foresee in a year’s time that we are going to be a better off as an organization where the teams are going to be more efficient, then we can invest in the co-pilot studio and what data agents we’re going to create and what tooling and skills these things are going to have. But you can’t go the other way. We can’t just say we’re going to create a bunch of agents and expect that to work.
39:20 You need to start with a business cultural point of view. How is my sales team going to be better off? And that’s going to take time. That’s discovery. This is the same crap that we did with PowerBI adoption. Start with that discovery. what are people’s pain points around data? It becomes a lot more general because AI can achieve so much more where I can go what’s their process, who are they calling? , what we can solve for is incredible, but we can’t be general about that. Generality is going to be the enemy of any AI solution.
39:53 This is interesting, Tony. I’m trying to unpack what you’re saying here around the general pieces of this, and I think you’re I think you’re right. Okay, I’m gonna I’m gonna what you’re saying I’m going to react to. Okay, let me unpack a little bit what you’re saying here. So the why when we look at when I look at AI or I look at these other copilot or even PowerBI or fabric for that matter, I’m looking at these situations where I’m I’m standing in a particular problem. I have something I need to accomplish or resolve. And while you
40:26 Were talking, I was thinking, wow, we could throw all these things at the AI. We could have AI build all these different agents or small tasks or solving simple problems. Right. So to your point, there’s this really interesting, very knowledgeable large language model, agents, whatever you want to call them, that can do a lot of things. Right now, I’m looking at like I have Microsoft M365 showing up. There’s an agent called a researcher. There’s an agent called the analyst. There’s an agent called workflows, right? There’s all these different things that Microsoft is producing that you can go use and help you build. Like,
40:59 I have this problem and this agent will help me build something that solves that problem. Right? Great. This is interesting. But I think it’s highly inefficient for us to to assume like if I’m building the agent every single time and I have to run a prompt back to the agent in order for it to build something that then produces an output. like even just searching for documents or so think about your day and all the tasks that you’re doing and what of those tasks can you start automating
41:32 What of those tasks can you start handing over to something else that make to build a little mini process or a little mini application around it. So this this is going to sound very maybe not sketchy, but this is going to sound very maybe like okay the weird ideas, but at the end of the day, Tommy, what you need is a lot of customized mini applications that are doing repetitive tasks for you. So whether it’s that whether it’s the agent itself like so you’re talking to the agent, hey agent, do this thing. Well, the reason you’re using an agent is because it abstracts away like the APIs.
42:06 It can think a little bit. It can do some stuff. It can it can basically take some information and interpret it for you and put it somewhere else. Once the agent understands what’s going on and once it understands enough of what’s happening here, why can’t these agents start building little mini applications that you use and then can then string together? Like once the agent understands the the game, right? , I was just recently working on a project where we had to take a URL and how you have a URL and you like
42:37 Encode a URL for HTML. So there’s like certain special symbols that go in or out. Okay. Characters. Yeah. So I was using an agent to say decode this, decode this. It’s the same effect as going to like Google, searching for decode URL, going to the website page that has the very simple tool of like paste it in here, get it decoded or paste the encoded version in and decode it. Right? This is a a very simple tool that is encode or decode something. Okay, what I think is missing here is
43:09 The orchestration part. There’s I can build this agent. I can build this little mini app. I can build this all these little things, but what is lacking is stringing a lot of these little task based things together over and over again to then automate them all together? So, does that make sense what I’m saying there? Yeah. And it it does, but I I can you tie that in to the the general part of that that you you wanted to talk about here because I see what you’re saying and I I would say that yeah, absolutely.
43:41 We want to tie things in together, but at the same time, what are you trying to answer there? Well, I think the general the general space of what am I trying to accomplish, right? So, at the end of the day, I have a problem I’m trying to solve. My workflow in my day is do some marketing, get some data in, move some data from here or go to this website, get this data out, put it somewhere else. Tommy, I can’t tell you how frust how absolutely frustrated I am with Google Analytics, YouTube Analytics, LinkedIn Analytics, and Twitter or X’s Analytics. Like getting any of that stuff out from
44:15 Any of those systems like it’s like stinking impossible. And none of them have good APIs. I can’t just easily get out with the information I need. I’m getting to the point right now, Tommy, where I’m I’m literally going to say, can I find AI based browsers where I can sign into these various sites and have it have the AI browser navigate each of these pages and just page through each page of data. That’s a good idea. I might try that. So, I’m I’m I’m testing it now. It works okay, but the challenge is the browser
44:48 Doesn’t technically have access to any of your operating systems. So, it can go to the page, right? It can go to the page, it can read page information, but it can’t do anything with the information that it reads. It can only like summarize and get information out. So what so again, let’s go back to like this. What is the why here, right? The why I have is I need analytics on the podcast. We have the podcast running on so many platforms and all the platforms report their data a different way. So unless you put a human in the loop and go navigate to this page, go to this thing, click this button, get the
45:23 Excel file downloaded, take that Excel file, rename it, put it somewhere else. Like I think an agent could do this. This seems dumb enough that an agent should be able to figure this thing out, but I can’t get it to work. Well, you’re still talking about an individual Mike Carlo problem. And this is this is the again I I need you to the What did Morpheus say? free your mind right thing this and I’m going to challenge you here I think this we’re talking about a human so you’re talking about an individually I Mike Carlo has a specific problem
45:55 I have a identify why a very specific why for a specific purpose which is where AI usually actually does the best however when we’re talking about AI in an organization or or in a larger scale this is a human problem this is not a technological problem and here is how I’m going to make this argument I want you to go to Facebook market and go search for bikes and look at the hundreds hundreds of bikes that are available to sell resale from someone. Okay, this is a problem that we have when we buy a bike man this is not just an AI
46:28 Problem where and again I want to expand this why Mike if you were going to buy a bike today what bike would you buy? , I would buy whatever Tommy tells me to buy because Tommy does much more bikes than I do. Right. But but that would be the wrong bike though for you. That would be the wrong bike. I I don’t know. I don’t even know. Like, right. Exactly. So, because you don’t know the exact thing that you do with the bike, right? And a lot of people deal with this. Now, let me expand this and let’s say you were buying a bike for everyone
46:59 On your team and you had to buy the same bike. It definitely would have a battery in it. Like, for sure it would have to have some battery. But that’s a but that’s a good start. But if you had to buy the same bike for everyone in the organization, you can’t buy a Bianke, which is like the high-end scale $15,000. Really? , I could buy it, but would it be worth my money? Probably not. Would everyone use it, right? ? No. Correct. Yeah. And then you don’t want to buy the Walmart version either where it’s just general and like I for me, for Tommy and if I was going with people who are biking, I have a bike that I invested
47:32 Money into. It’s not the it’s not a Bianke. Let’s be here. Tommy doesn’t have that much money. , it’s a really nice bike though. , but I have a bike that suits the purposes that I had a problem. Like Mike, I was like, I want to lose weight. I actually want to stay in shape and I’m going to put a lot of miles into this. I had those measurements already set out way before I knew which bike I wanted. I didn’t have the three options. I knew that I like I want to do about 100 miles a week. I want to be able to be comfortable. I want to go fast thing, , and I want to beat some
48:05 Records, right? And then I’m going to look at what bikes meet those requirements. Most people when they buy a bike are not even doing that, but we’re talking a bike now. And I think you’ll bring this back to this AI problem where people like, “Oh, I want AI. I want to be able to ride a bike.” Well, that’s very general. What are you going to do on that bike? Are you as an organization Mike wanting AI to improve the applications that you’re building? Is it for communication? Right. the best cases I’ve seen AI work where they’ve
48:37 Been adopted is honestly operations teams when you are dealing with people especially on the chat point of view to say hey has Jim put everything in the PO is this ready for approval and AI can go through soft data like the shareepoint information to make sure Excel files are where they need to be and then provide that answer or we have all these papers we need to input but those are very specific problems. If if you want to start with AI with your company, start with operations. They have the easiest easiest solutions that AI can solve for
49:09 And very specific. But everything we’ve talked about and I think a lot of the and not in a bad way, but I think a lot of the situations that you brought up were very specific to my Carlo, which is great, which is fine. And we can find AI working very well. But co-pilot again when you’re dealing with hundreds of people at a company like if you don’t micro it down and if you are just staying general you’re going to get general answers and that’s where the problem is. Okay Tommy while you were saying
49:42 Yes. So while you were saying this I tried Comet. So, Comet is a AI based browser and I did exactly what you said, Tommy. [laughter] I’m literally trying. So, but this is this is where this is where I’m trying to struggle like where does this stuff fit, right? So, there’s this general AI that’s like can do things, but I’m trying to solve specific problems or a a very detailed problem. And so, to your to your question, let’s talk bikes. Let’s just give a real practical example. Hey, there’s a number of bikes. Go to
50:15 Facebook Marketplace, look at a bunch of bikes, click on the links, find a bike that is comfortable, can handle a good three to four mile commute. I just like see I literally said research Facebook. I said research Facebook marketplace for bikes. Find a good bike for me to use that would be a good 3 to four mile commute. It needs to be a comfortable ride and in the $500 to $1,000 price range. just some gener generic things and just do something and then it goes through the Facebook page and actually
50:49 Scans and skims through all the different items and goes bing bing bing bing and start showing me all these different items which I think is incredible. I I yes I get all that like that’s what AI general AI should be doing, right? But it’s like when does this become like okay that was a a random request at a random point in time. When do when does some of this when does something like this this is a task, but let’s say I’m I’m regularly let’s expand the idea a little bit. Let’s go. Okay, now I’m a marketing agent or I’m a marketing person and I want to go to a a a
51:23 Narrowed focus area. Here’s the page. Here’s the pages I want to go scrape. Here’s like I want to start taking these tasks. And I think where the AI and agents are supposed to really help out here and what I think is missing right now is the automation of this, right? There’s there’s a whole pile of this stuff that’s just very singular in nature. A one-off. Like this example here, the bike example, a one-off. Like, okay, I’m going to buy it once. I’m going to move on. I’m not going to use this thing again. So, this is this makes sense to be a good agent or a an AI general model. Now, let’s take this to a
51:57 Reusable, right, piece of tooling, right? The input to this tool is some scripting things, but now it knows what to do like and does it really need to have an agent behind this running this thing like or can it start can the agent build a little mini application that understands okay I’ve got to go to this URL I’ve got to put in these parameters I’ve got to search by this stuff like can the agent build a little mini tool that is now repeatable and where I’m going with my
52:30 Mind here is okay, we’re looking at AI, agents, all these other things, even PowerBI, right? There were things I was doing that was very manual. We’re now putting automation in place of that. And so now we want that automation to be cheaper than what it was for me to run it before. And I don’t want to spend a ton of money to have it run all that extra stuff. So now I I go from this generalized common one-off use cases to how do we take that general use case and turn it into a productionalized use case? Something that I can I can automate in and out of to get that happening over and over again. And I think that’s what
53:03 I’m missing right now. I’m missing that part of the AI space in Copilot. I’m missing part of that AI space in like fabric and and agents. So I have again going back to the question of the the thing here. We’ve talked a lot about AI and how things are are playing here, but I think going back to is I’m still always trying to solve a singular task or a task that will be repeated over and over and over again throughout my week or the months that I’m working. Right? And I I think this is going to be close to my closing thought. I think
53:35 Honestly anyone who’s been working in PowerBI is going to be very prepared for the AI revolution or these data agents because it’s we’re tackling the same problems where I realized the best value I had at a company when I worked at FPE and what I do now is not when I build an individual report. It’s when I’ve been able to uncover what I say like looking under the rocks where I I would notice the processes that took a lot of time. It’s like, well, how can we save money here? One of the best projects we did was honestly building.
54:08 We looked at this large Excel file that people were using. And we saw all the different inputs and time it took just to get that Excel file right. It’s like, well, how many hours does this spend? And we actually looked at each individual team, the time it took to get that Excel file did, and we’re like, this alone can save the company hours, hours of time on a weekly basis. That’s where the best value has been for us. and BI. It’s not just been through a bar chart. It’s where can we automate and save that time. And I think we’re going to see we’re going to see a lot of
54:41 People from the business intelligence space. Mhm. Move into copilot studio move into a lot of these co-pilot and data agents because we have to if you were good at it, if you were successful with BI in that area, you’re it’s a very easy skill that translates over to how does the organization solve those AI problems. And that’s and this is where I think I’m getting with a lot of this stuff is there’s there’s a lot of pieces here where we’re trying to figure out where’s the right surface area of these different AI things to get in place. And
55:14 Right now what I feel like I’m lacking right now is I can’t quite get so for me the general use cases of things right the agents that I’m seeing today Microsoft doesn’t have the tooling in place to build very detailed agents that can then be turned into repeatable objects that get used over and over again. You need to look at Copilot Studio a little more then. That’s again admittedly that’s something I’m not playing with. So I’m being fully transparent and clear. I’m not playing with Copilot Studio. That’s something I
55:46 Have yet to experiment a bit more with. But I’m when I look at other tools out there, right, Copilot Studio is interesting, but can it do what I need to do for NAN? Does it integrate with all the Microsoft products the way I want it to integrate with? One other random side note, we had Chris on the podcast here and Tommy, you’ve been seeing my chats back and forth to Chris. Fabric is great. I can’t just randomly send data to Fabric. There’s no there’s no method to have like a web hook or an API call go into Fabric and just send me data. Everything
56:20 That I’m aware of at this point has to be authenticated. I need web hooks to show up to Fabric and just do things. A lot of other services like PayPal and other services use web hooks very easily. Here’s a URL. Send data here. It’s up to you to validate if that data is actually useful to you or not to save it or not. But there’s there’s a whole missing world of like in my opinion in fabric again I can’t talk to an agent. An agent can’t build me a pipeline. The agent can’t build me the thing that I need yet. So we’re the the gap is getting closer to closer to being
56:53 Closed. We’re just not quite there. And I really do feel Tommy this whole agents copilot agents thing it just needs to live inside fabric. It I get that it’s in Azure right now. That whole exper like that agent builder like experience just needs to be moved out of whatever it is and just put into fabric. That’s where it needs to be. Anyways, that is frustrating. But I I’ll end on this. Speaking of agents and doing what they need to and we’re getting there. I forgot as we’re talk as you
57:25 Were talking I got all my notifications from Claude from Perplexity and from JCPT I have set up searches that I wanted to do whether it’s a deep research or just find me some information every Monday morning and they’re all coming in today. For example, I want to find new Claude skills that are coming in. I also want to know when the movie Leo is going to be out. My kids love it but it’s on Netflix only and it actually finds when can I purchase it. So the the automation, we’re getting there and I’m going to Wait, hold on. Where where are you doing
57:57 This though? Oh, so I’m going to those tools and I say, “Set a task for me. I want every Monday for you to do this, , research. Find me new open repositories for this new thing called Claude Skills and send me those latest ones, but do it every Monday at 8, , 8:38 in the morning.” And then it just does that search. It finds me this stuff. It sends me a message and saying it’s ready for you. That’s my new Monday stuff. And again, I So I have those three going up on three different tools. This is But I think you’re I think
58:29 You’re speaking directly to my problem, which is which is you have to go to all these tools and they have their own way of doing it and there’s not a centralized way like where’s the orchestr like to your point Tommy I love it. That’s a great idea. That’s a great example, right? set a task, do this thing. Again, what you’ve what you’ve described for me is the general problem that I’m trying to you take a general AI, you’ve made a general thing, I’ve built a task around that one general solution, and now you’ve been able to schedule it to do a thing
59:01 Repeatedly for you. Awesome. Great. My hangup on this is one, where is this same functionality in anything Microsoft at this point? Don’t see it yet. Yeah. And then two, what is the input to that task? And where does that task output its data to? It sends you maybe an email or put it somewhere. I need it I need it to land in fabric. I know honestly like it’s got to be in fabric somewhere. It’s got to land there. Like for me to effectively use all these AI and start with the Y be so easy.
59:33 It’s just I need the data to get somewhere. So you can’t tell me there’s a co-pilot agent I should be able to create saying hey search the web for new marketing data. put it into this lakehouse, done like, , go download where does that exist? Why is that not out there yet? Like I don’t understand why this doesn’t like and and so okay, we’ve gone very very far away from this topic which is like, , talking about starting with the why. I think it would Yeah, I think we’ve talked about a lot of good use cases of like, hey, here are tasks that we are doing manually that we’re spending a lot of time on. We need those
60:06 Tasks to be automated and landed in here. And so I think the starting with the why you’re trying to solve this problem is super effective. Yeah. And , again, I’m still trying to learn this out and figure out how this works. I do think there’s also a trend here. I’ve been watching some of my newer engineers on the platform that we build with do things. I think there’s a trend, Tommy, that people are going to start moving away from Googling things and they’re just going to start sending
60:40 Their questions or responses to an agent. So, go back to my earlier example of decoding or encoding a URL. You’re going to go put that decoded URL. You’re going to go to chat GTB5. You’re going to go to Copilot. You’re going to you’re going to go to these other tools and say, “Here’s what I want you to do, agent. What I’d like the agent to do or start doing for me is identifying these reoccurring tasks that I’m doing and have the agent be smart enough and say, “Hey, I see that I’ve decoded three things for you. Do you want me to build you a mini tool that does decoding for
61:13 You? Put on your new tap page. Let let the agent generally build you the tool that says here’s an input field. Here’s the two buttons. Put it here. Paste this. Click this out.” Boom. Boom. Boom. Okay. Now that I’ve got that done, here’s the app, the web app. Here’s the here here’s an API call you can use to get data in. Here’s the API call to get data out. And that can all be done once. And then you can take that efficient piece of code and just run it.
61:44 So, and I don’t need the agent every time I’m doing these things. So, that’s to me, this is what’s missing. I’m I’m going to make a note here on episode 4. What episode are we on? ,77. This may be the nerdiest thing you’ve ever said and I loved it. [laughter] URL coding data finding out mini web app hosting. So there [laughter] that’s a nerd recipe. Tommy, we we say words. It’s funny because like we talk in words. I know words that other people
62:16 We talk in words, but the words we use are like a different language to other people. If I had said those words to anyone else who doesn’t know exactly what we’re talking about, the context for each of those words are like, “What the heck are you talking about? I don’t even understand encoding my wife. Are you talking about a spy novel from the 50s?” Like just literally a HTML web page. Like it’s just like basic stuff. Anyways, there’s there’s a the future is really bright here. Yeah, I think there’s a lot more coming. I do feel like fabric is like the solution for integration, right? It’s it’s the solution to get data to many different teams in different ways. But I I think
62:49 Regardless of whatever tool you pick, it doesn’t even matter if it’s fabric at this point. You have to start with why you’re doing something. And if you’re spending time on things that are not important, the question has to come back to say why are you even doing it? And so regardless of all the tools, regardless of all these AI, co-pilot agents, adoption, a lot of that stuff hinges on do we have a proper question around the why? Why are we doing this? And once we think we get that phrase started, if you start with the why, then you can really
63:22 Start being effective with what you’re doing later on down the road. A really good point. Dude, this is very This is a very timely topic because I’m dealing with a lot of this and processes right now that I’m already building. So, this is very, very good. More to come. Very excited about the future here. Thank you all so much for listening to the podcast. We burned through well over an hour of your time today. So, we really appreciate you sticking with us and listening to the podcast. If you want to get this episode early, as soon as we record it, make sure you become a member on our YouTube channel. We have a membership area. Become a member. You can get this episodes or any episode that we pre-record. It’ll come out the day we
63:55 Record it. We’ll put it out there for you so you can have it. If you’re not a subscriber, you can catch it on all our normal media platforms. Tommy, over to you. Where else can you find the podcast? You can find us on Apple, Spotify, wherever you get your podcast. Make sure to subscribe and leave a rating. It helps us out a ton. And share with a friend since we really we really do this for free. And like today, do you have a question, idea, or topic that you want us to talk about in a future episode? Head over to powerbi.tips/mpodcast. Leave your name and a great question. And finally, join us live every Tuesday
64:28 And Thursday, 7:30 a.m. Central, and join the conversation all Power Data Tips social media channels. Thank you all so much, and we really appreciate your time. We’ll see you next time. out.
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.