A Lesson in Recent History Cures Today’s AI Uncertainty

Listen Now:

Welcome to this week’s edition of Raw Data, where we’re diving into a timely discussion on the evolution of AI and its comparison with revolutionary technologies of the past, like Power BI. This episode features host Rob Collie and co-host Justin Mannhardt engaging in a deep dive into the hype cycles that have surrounded transformative tech over the years.

Drawing on Rob’s rich experience, we take a step back in time to explore the transformative journey of Power BI, a tool that redefined the landscape of business intelligence by making sophisticated analytics accessible, swift, and economical. This story of innovation and foresight is juxtaposed with the current state of AI— a field bustling with potential yet shrouded in a mix of anticipation and ambiguity for many businesses.

In this episode, Rob and Justin dissect the realities behind the AI hype, urging a pragmatic approach towards technological adoption. They highlight the importance of mastering existing analytics tools like Power BI, which many organizations have yet to fully leverage, before being swept up in the whirlwind of AI enthusiasm. This practical perspective is essential for businesses aiming to make meaningful advancements in their digital transformation journey.

As we navigate these discussions, the episode serves as a beacon for those looking to understand the true impact of AI in the context of proven technologies. It’s a call to action for focusing on tangible business problems, employing a strategy that prioritizes impactful solutions over the allure of the latest tech trend.

If you’re excited by the intersection of AI, analytics, and business strategy, you can’t miss this episode. And as always, if you enjoyed this episode leave us a review on your favorite podcast platform. Your feedback helps shape our content and reach more listeners. Additionally, for those eager to dive deeper and share your thoughts, we invite you to join our LinkedIn Steering Committee. Your insights and suggestions are what shape our journey forward, illuminating paths to future episodes and discussions.

Want to know more about Fabric and AI?

AI Isn’t Scary When You Break it Down: Five Use Cases Approachable Today

Ten Business Upgrades Within Your Reach in 2024

Fabric – It’s GA (GO) Time!

Should We Be Skeptical About Microsoft Fabric?

Mailbag #1: Python, Fabric, and Mullets oh my!

Fabric is Nothing, But it’s also Everything!

Rob Collie: Hello, friends. Today's episode begins with a history lesson, but ends with a lesson. When we sat down to record, the history lesson was sort of the goal. That was the topic that Justin picked for us and we did not expect it to lead us where it did.

And I know you'd probably expect me to say this about every episode, but this is one of those episodes where you really need to stick around for the end because that lesson that we arrived at, it's more of a guideline or a rule actually. It's something that I think is super useful for helping keep your head on straight when thinking about what your data strategy should be, what your AI strategy should be.

These are questions that are on everyone's mind these days. And I feel comfortable saying that where we landed, this lesson, this rule, I feel comfortable saying that it's insightful because even I am going to be using it going forward. And in some sense, it's thinking that I already had, I just hadn't reactivated it, hadn't applied it to today when I absolutely should have been.

Now, the lesson was the unexpected destination, kind of the payoff of the journey that was the history lesson. And the history lesson I think does a lot to give you the framework for why the lesson is trustable. So yeah, you got to eat your vegetables to get to your dessert.

I happen to think that the history lesson vegetables are actually quite tasty themselves, but there really is an inflection point near the end where the lights really, really just kind of came on. I left this conversation feeling a lot more confident and comfortable, and I sincerely hope that you do as well. So let's get into it.

Announcer: Ladies and gentlemen, may I have your attention, please.

Announcer: This is the Raw Data by P3 Adaptive Podcast with your host, Rob Collie, and your co-host, Justin Mannhardt. Find out what the experts at P3 Adaptive can do for your business. Just go to p3adaptive.com. Raw Data by P3 Adaptive is data with the human element.

Rob Collie: All right, here we are. So last night, I realized that we hadn't selected a topic for this week's show, and so I did the responsible thing and I delegated it. I shot you a Slack and said, "Hey, Justin, lucky you, you get to come up with our topic for today". You told me this morning that you have one. Well, I'm all ears, aren't I?

Justin Mannhard...: Normally, how these Rob-Justin episodes have gone is you start asking me a bunch of questions and we have a discussion about fabric, or AI, or FOBO, or something, right? Well, it's my turn. I'm going to ask you questions today.

I thought what would be interesting is to talk about you and your story and how that applies to today because you've had lots of people on this podcast. You've interviewed them and you've talked about that. And the reason I thought of this is one of the things I'm always very keen on doing is helping P3 as everything changes in the world, continue to be the type of thing you wanted it to be when you started it.

So I thought we could get in the time machine a bit and talk about your experience maybe at Microsoft and what you were seeing. Because I've heard bits and pieces of this story just working with you for the last five or six years, but I don't know that I've heard it all the way through in one conversation.

And I don't think our employees, or our listeners, or our customers have either, so that might be kind of fun. What do you think?

Rob Collie: Well, I mean, first of all, put a microphone in front of me and say, "Hey, let's talk about you." I'm like, well, okay, file this under don't threaten me with a good time.

There are things about decisions that I've made and things that I've seen in my career obviously do have some personal interest value for sure. But what those imply or portend for people who are in this space, whether you're a practitioner or you're a business leader, trying to make sense of it all, I think there are a lot of relevant points there.

And understanding and having a perspective on all the things that are coming at you, having a framework to plug everything into is really important. And I didn't have that for the first part of my career like my early days at Microsoft. I just was experiencing tech as all these islands and stuff coming at me, and it was just overwhelming. I don't feel that way anymore.

Justin Mannhard...: I've always perceived as sort of a catalyst for the whole thing. You've described it to me as the experience where you discover fire. We're talking about this thing that used to be called power pivot for crying out loud. What happened for you that you saw, okay, the world's about to go in a different direction? Because that's happening now.

Rob Collie: We're at a point now where the waves are starting to compress. We're still in the last one and the next one, AI, etc, etc, is at least signaling its arrival.

Justin Mannhard...: When I imagine just for myself maybe what you were thinking about or seeing in the AI wave today, there's been people that they've been working on this stuff for years. They've seen it coming, and now it's just we're in this rush of like, oh, and the world is now seeing it.

Rob Collie: Yeah. When I joined Microsoft in '96, there was the internet happening.

Justin Mannhard...: Oh, yeah.

Rob Collie: That was kind of a big deal. Literally in '96, we weren't sure. The way that AI feels today, you can't ignore it. It's going to be a big deal, but we're not even beginning to see what the steady state looks like. In '96, there was still this idea that maybe the Netscape browser was going to be so important, that it was going to become the center of the universe and it was going to usurp Windows.

Well, that didn't happen. I mean, an entire industry battle in an antitrust case was fought over this notion. It's not like Windows Edge browser, or Safari, or Chrome is ruling the world today. No, the browser never ruled. But setting the internet aside, which I know is a big thing to set aside, my experience for the majority of my Microsoft career was that of just incremental slog in terms of technological advancement.

It was more just like, well, you come up with a new incremental idea, sort of like the better bold button. The big breakthrough in Microsoft Office, in my opinion, it happened in 1995, the Office 95 release where Word added the red squiggle underlined feature for spelling.

Justin Mannhard...: I've never known a world without it, Rob.

Rob Collie: Exactly, right? For years, I would laugh and tell people, Microsoft. I'm like, "Well, why does anyone need a new Win Word? Why does anyone need a new Microsoft Word? They did the squiggle feature already." And honestly, that probably is the last big thing you had to have.

And then the ribbon interface, there's a lot of people working today who've never known a world without the ribbon. The ribbon didn't rely on tech. It was just a good idea. It wasn't some technological breakthrough. It was just like, let's stop doing this menus and toolbars jazz and let's give people a more humane experience.

So the ribbon probably is the other thing that advanced Word. So we were in this stalled place in general. Not much was getting better, and this is even in the era where CPUs and everything were getting twice as fast every 14 months.

We should have been in the era of rapid advancement, but we weren't. It was kind of slow, incremental, not that big a deal. And the first time in my entire Microsoft career that I encountered anything that seemed to feel like science fiction was the Power Pivot data engine, which we now know as SSAS Tabular, and it's the data model technology that is underneath Power BI. And is also now the one lake storage format, overwhelmingly so.

That was a real breakthrough, a way to actually harness these CPUs and memory and all this stuff that had been getting cheaper and faster and bigger. Even then though, that was just kind of like, gee whiz. Cool. I think the real inflection point that we've been living for a long time now was just this notion, this dangerous idea that IT technology could be made fast and accessible.

We had lived an era where there was end user tech, aka Excel, and then there was anything more responsible than that. Anything more industrial strength or IT blessed was from a different planet. You couldn't do anything in less than months. And I'd been introduced to business intelligence when I worked on Excel.

The whole business intelligence industry as constructed in, let's say 2005, was 100% like that. It was super, super, super slow, super, super, super expensive. I had customers tell me things like... When they were being really honest with me like, "Yeah, my team spends three months to put a dot on a chart." That was a customer confession.

Justin Mannhard...: That's soul crushing.

Rob Collie: It's real, right? The industry metrics of how many business days it would take from requests to delivery to add a single new column to a single already existing report, it was at least two business weeks.

The equivalent of clicking a checkbox. And that's the thing that Power Pivot that became Power BI really kicked off. Ignore the breakthrough, crazy, crazy compression and speed, the ability to just chew through enormous volumes of data like it does. It's kind of like in a way, like the ribbon.

Justin Mannhard...: Yeah.

Rob Collie: Let's make it easier. I bet we can, and no one had tried. No one had tried to bridge the gap between the heavy duty IT stuff or power user tech like Excel and Access. It just so happens that the team that tried it at Microsoft was the team that owned the single most complicated of all of the offenders in the slow and complicated and expensive space.

The team that tried it to make it easy was the team that owned the worst offender in that category, analysis services, the OLAP, MDX, behemoth, right?

Justin Mannhard...: I remember the... I'm having my own flashback because I did not ever do a significant amount of work with multidimensional AS. But I remember my first encounter with a block of MDX code and just being like, "Nope." I'm all this young cocky Excel guru. I'm like, ah, let me have a crack at it. Nope.

Rob Collie: I was a young cocky Excel guru who asked multiple times to get training lessons.

Justin Mannhard...: Let me edit.

Rob Collie: The dark arts of MDX and OLAP cubes of the day, analysis services, and boy, did I face plant every time, couldn't learn it. I believe that still today, I still probably would really, really, really struggle with it. Even though I understand so many of the concepts better now because I've absorbed them through the lens of this much more humane approach that is SSAS Tabular and DAX and everything.

I said that like it was a ironic accident, that the worst offender was the team that tried to make it easier, but I guess it makes sense, right? Completely in the financial best interest of that team to make the experience of building good interactive data models to make that. Better because they're in the business of selling the servers that run those models, right?

Microsoft isn't a consulting company. And so to them, how hard it was to build these things, even though they're amazing when you get them built. One time, someone described something to me as like the Ritz-Carlton, it's great to stay there. You don't want to show up and have people tell you that you need to build it.

Justin Mannhard...: Right.

Rob Collie: I just want to stay there. I don't want to build the place.

Justin Mannhard...: You've described that in a way to me in the past of this idea that that team got the chance to do this. It's extremely rare.

Rob Collie: Yeah. When you build something successful... And the thing is, analysis services was by any measure, was a successful product. It was the market leader. It had more market share than any other competitor in its space. And so often, you become a company or a team in that position becomes entrapped by their own success.

Justin Mannhard...: Yeah.

Rob Collie: You are collecting an income stream from customers of that technology. And rather than deploying your engineering resources and your best and brightest minds to continually improving that, you're going to just stall development on that other thing completely and build something else?

That by the way, when it gets out that you're building that thing, people who are debating adopting your technology are going to pause their decision. Not only are your existing customers going to get grumpy at you, you're not doing the next thing for them. The incremental customers that you are about to win are going to say, "Nah, I'm going to hang back. I don't trust you guys."

Justin Mannhard...: "I'll just see how this goes."

Rob Collie: Yeah, "I'm going to hang back." Most product teams are not given this opportunity to completely reimagine their successful product. And those that are, usually reimagine it really poorly.

There's even a thing called the second-system effect where you give an architect, software engineer a second shot at something and they just absolutely blow it nine times out of 10. Anyway, that didn't happen. This was a true wise re-imagining when they sat down to build the new data model engine and it wasn't even knowable that it would work.

Can you make that concept simple enough that it becomes faster, that it becomes more accessible? Let's just fast-forward and say, yes. It was a smashing success and the fact that it was the apex predator of difficulty really inspired them. It's like becoming a mountain climber and wondering if you can even climb a mountain and climbing Everest first and being like, let's go climb the others.

Justin Mannhard...: There's a cool connective tissue in that team tackles apex predator problem, makes technology easier, more accessible. Now world tackles apex predator of analytics problems. It's pretty cool.

Rob Collie: Whereas I think we were in a very incremental phase of computing for the majority of my Microsoft career from '96 to 2010. This change, I believe is the thing that kicked off the inflection point wave of technology and advancement that we've been living for the majority of the last 15 years.

Whereas honestly, I think the 15 years before that, if you set aside the internet, I know huge asterisk, was completely incremental. It was a major breakthrough that Excel went from 64,000 rows to a million rows, but that's exactly what I'm talking about. That sort of proves the point.

Justin Mannhard...: You feel that in recent years as well, for a while, you read the monthly releases on stuff. I'm like, oh, okay. They moved some menus around. There's this new feature, but it's not dramatically shifting my landscape.

Rob Collie: Yeah. Think about the internet, most of what it did ultimately, it led to cloud computing, which you could say is another inflection point,.like Power Pivot launched before the cloud computing wave did, cloud services.

Going back before that, the PC was the last one, just like the introduction of the personal computer. Because we're all still sitting at PCs, it's hard to notice that there's something as big going on.

I actually think that the wave of technology, like the hybrid IT Power user technology, the citizen developer stuff coming out of Microsoft is as big a deal relative to where we were in 2010 as the spreadsheet was when it was first introduced as compared to pen and paper. But you don't notice it. When the spreadsheet arrived, the PC also arrived. Something physically entered your space that wasn't there before.

Justin Mannhard...: I see. Yeah, that's interesting.

Rob Collie: If we'd all gotten a completely different form factor of device in our lives with the arrival of Power BI, we would've noticed it more and given it much more proportional and accurate importance. But no, we're all still sitting here with mouse, keyboard, monitor, right?

I mean, the monitors at some point in that timeframe went from four by three to 16 by nine, and they got a little thinner, but that's it. That's the only noticeable difference.

This is happening all over the place, and I think there were a few differences about, let's just call it Power BI because Power Pivot was the first foray, but it was largely unsuccessful.

The dangerous and successful idea and the technology behind it that went into Power Pivot got reskinned as Power BI, but it's successful in the market. The switch from Power Pivot to Power BI wasn't some huge leap of faith, whereas the creation of Power Pivot was. Anyway, so let's just call it Power BI from now on.

Justin Mannhard...: Yeah, that's fair.

Rob Collie: It's multiple things. It's accessible. I've been focusing on that, meaning there's a wider crowd that can learn how to use it. It's still not everybody, it's really not, but it's much more accessible, orders of magnitude are more accessible than the previous technology. It's faster and more flexible to use even by the experts. So when we talk about accessibility, it almost makes it sound like a preschool thing.

Justin Mannhard...: Yeah, it's a dumbing down.

Rob Collie: It's not. It's anything but.

Justin Mannhard...: Yeah.

Rob Collie: It's accessible because the concepts are simpler and the easy things are easy to do. Whereas in MDX world, the easy things were still hard to do. You had to pay a mighty cost just to get the equivalent of hello world.

So that kicked off a lot of things, right? So a few years later, they did it for ETL. They did it for the data transformation stuff that again, the power of analysis services, the original analysis services was so great. Again, once you got the Ritz-Carlton built, which by the way, almost no one ever, ever, ever got the Ritz-Carlton built.

If you ever got there, it was amazing, but it was so abstract, so expensive, so ivory tower that even at Microsoft Business Intelligence gatherings... In the year 2010, I went to a user group that was Microsoft BI and no one in the room really understood what analysis services even did.

Justin Mannhard...: That's bizarre.

Rob Collie: Isn't it? It's crazy. You couldn't even explain the benefits of it to people. What do you get if you deploy this, right? Whereas now, you don't even think about that under-the-hood thing. It's like, no, you get this dashboard.

Justin Mannhard...: Yeah, you get this stuff.

Rob Collie: To get this dashboard that shows you end-to-end across your whole company and you can get it pretty quickly. So it opened up the value proposition basically to the world. It basically rounded to zero the number of people in the world who understood what they would get out of a tech like this.

And furthermore, you can now afford it. And this is I think one of the most important things to emphasize is that even though this tech has been out for a long time now, I still think the world is in the midst of waking up to what it can do for them.

Justin Mannhard...: Agreed.

Rob Collie: And it's a whole family of things that now follow that same philosophy. Like I said, so I was talking about Power Query. SQL Server Integration Services was another big monster, scary apex predator technology, not quite as apex as analysis services was, right? It wasn't the T-Rex, I don't know.

Justin Mannhard...: It was a pack of Raptors. It was pretty...

Rob Collie: Yeah. It was pretty bad. And they made that accessible, and again, also scalable. It scales all the way up. It's part of actually multiple places in the Microsoft ecosystem now. It's not just a Power BI thing.

Justin Mannhard...: Yeah. I think at one time, I counted 10 discrete implementations of Power Query across Microsoft in different places.

Rob Collie: Wow, that's amazing.

Justin Mannhard...: Yeah.

Rob Collie: The place where this comes back to me for a moment is that I was used to the incremental progress in software. Those are only model I'd ever known.

Justin Mannhard...: That was norm, that was the normal.

Rob Collie: I got my first PC in 1992 when I went to college. Honestly, red squiggle underline was the only thing that appeared on the PC itself between '92 and '96 that I cared at all about. Okay, that's not true. Doom also, that's a pretty big one.

Anyway, I didn't expect Power Pivot, which became Power BI, I didn't expect it to be... Even though I was working on, I was one of the first few people at Microsoft working on this thing. I didn't expect it to be any different.

I expected it to be incremental, not that big a deal, given the level of its ambition, probably not even all that successful. But then I saw much to my surprise that it was amazing when I finally deployed it against a real problem. I had the advantage of deploying against a real problem that I had paid an expensive consultant to do for me before, so I got to relive it.

Justin Mannhard...: For yourself though?

Rob Collie: Yeah. And I was hoping that it was going to be 80% as good as the result that I got when I paid the professional because I wanted the technology to work. I was somewhat patriotically invested in it, although cynically expecting it to be bad, seriously. It's a weird place to find yourself.

And I didn't discover that it was 80% as good. I discovered that it was like 1000% better, and that was in one of the earliest, earliest versions. It was even in a beta release of it.

Justin Mannhard...: Yeah.

Rob Collie: And the signature moment was I wrote a formula on my own, and I didn't realize at the time, but when I got the formula done in Power Pivot and it was probably 20 minutes and it wasn't a hard formula to write, it was just debugging it, peeling the onion and getting it closer and closer to the right answers.

And eventually landing there 20 minutes later, that it hit me that I'd walked that exact same path with that exact same formula using the consultant and the old and slow technology, and it had taken about two weeks.

Justin Mannhard...: 20 minutes versus two weeks.

Rob Collie: 20 minutes, and not paying a consultant and also not having to explain over and over again to a consultant how the business worked, and that kind of blew my mind. And that's when I started to rewire what my career was going to be going forward.

Oh, this is an inflection point. Oh, the world of consulting around all this stuff is going to completely change. We need a different kind of company that's able to take on smaller, faster projects at lower price points because those just didn't exist.

And it's really exciting. We can do things that explicitly go after and target the mid-market as a place where we really like to help people. That was completely priced out of all of this before. And honestly, there wasn't even a value proposition to it that you could explain to people back in the day, even if they could afford it, which they couldn't.

And it's almost a blessing to these people that they were priced out because, oh my God, it was awful. If you did it right and managed to navigate all the obstacles, it would deliver results that were positive. But most of the time, this was something that you would give it to your enemies to inflict upon them.

There's this thing that goes around on the internet periodically now, which is the sabotage manual that was delivered to resistance forces in Europe in World War II.

In addition to all the usual stuff like pouring sand and gas tanks and blowing up this and blowing up that, they had a knowledge worker, essentially like a knowledge worker sabotage guide, which was turn everything into a committee, call for meetings on everything. Have you seen this?

Justin Mannhard...: No. This is great.

Rob Collie: We've got a link this. When I saw this, the first thing I did was like, oh my God, that is the old school IT project handbook is to do all the sabotage things.

Justin Mannhard...: I need a steering committee. I need project review board.

Rob Collie: So we've been living that and Microsoft's investment and dedication to this vibe. We're seeing it like Fabric. We're not going to rehash it all here, but if you want to know why Fabric is like the absolute explosion continuation of this strategy, go listen to our handful of podcasts on Fabric.

And then of course, while we're in the middle of this, we're still playing this out, waking up to it, that's when AI is now signaling its arrival. But it's weird though, right, because when I saw Power BI for the first time and saw its promise, I knew exactly how it was going to be applied. I could see more data projects.

Justin Mannhard...: Faster, cheaper, better.

Rob Collie: At different levels of the market, need a company that's able to operate like that to help people rather than using this saboteur manual. And so all of that was so clear. It was clear to me anyway, not because I was super smart or something, it was just that I'd had the right set of experiences.

Justin Mannhard...: Sure.

Rob Collie: I'd lucked into the right set of experiences so that when the new thing came along, I could connect the dots and go, "Oh my God." I'm not getting that same clarity at all from the AI space. I'm working on it. We've done some episodes about decomposing AI into most promising subcategories.

Justin Mannhard...: A couple interesting points of perspective. One is the perspective of, I actually started using this thing in a very first on the scene type of way. And you had this belief that you could go out on your own in a way and start this company and do these things. That's one perspective, like, hey, I'm going to go out and I'm going to go do this thing. And how that might apply to how you're feeling today about AI, for example.

The other perspective is, and I don't think you were necessarily thinking this way 15 years ago, you're a business leader, you're a CEO of this company. And how do you balance the, hey, we're in this hype cycle, this new thing is coming, should you jump on it? Should you wait? The world's still waking up to BI.

You were an influential figure in the early days of all this stuff. What would you say to people wrestling with this dilemma today? Because I think we're all wrestling with that. Do I skip BI and go to AI and what is it?

Rob Collie: Well, the first thing I'd say is you're in good company. The people who tell you with confidence that they know the things that you absolutely need to be doing and they know what the future holds here are lying to you. The only question is are they lying to themselves as well? Do they know their frauds, is the only unknown.

I have seen so many examples of this now where the internet was breaking in 1996, not breaking down, but sort of breaking onto the scene in '96. And I sat in so many big powerhouse brainy rooms with absolute demigods of software pontificating on what it was going to mean, and they were wrong about all of it, just completely wrong.

I remember one of the big badass architects at Microsoft who was in charge of damn near everything, going all in on adding a folder to Windows called My Pictures. This was his calling card for 18 months. The previous decade had been about text, and the next decade was going to be about pictures. Okay, he wasn't wrong. Yes, imagery did explode, and then guess what? He didn't have the foresight to say my videos at that time, did he?

Justin Mannhard...: No.

Rob Collie: We went right through pictures into videos in an eye blink and he missed it. I've watched, again, same type of people, people with IQs that are... I'm not shabby, but I've watched people with IQs so much greater than mine say things like, "Yeah, we need to turn the Excel grid into a XML shaped surface because that's the nature of what data is going to be in the future."

And completely missed the fact he was right that data was going to be less structured in terms of how it was stored, but he was wrong. That analysis was always going to be rectangles. You only analyze based on things that are in common, right.

Justin Mannhard...: Yeah.

Rob Collie: It literally doesn't make sense to compare a breed of dog to a holiday. The one time in my life, it won't happen again. It didn't happen before. It won't happen again. The planets aligned. I was able to see with reasonable clarity, maybe not great depth perception.

I thought that this would've been over in three years, the world's adoption of all of this and adaptation to it, and here we are 15 years in and it's still happening. But I was right about what Power BI was going to do. I don't know what the steady state of AI is going to look like or even if there is a steady state. Are we entering perpetual inflection point mode?

We're now at least a full year into it being front page news everywhere. And look around, the walls are all still the same color. I'm still sitting at a PC. It hasn't even coalesced around particular use cases that are super, super, super clear and disruptive.

Justin Mannhard...: I caught a clip of an interview with Sam Altman and Satya Nadella and Sam was saying something to the effect of at the beginning of all of this, two weeks into all of it, the world had a freak-out. Everybody's going to lose their jobs. Robots are going to do everything. It's going to become self-aware, yada, yada, yada, and then very quickly, all we start hearing is like, and it's too slow, right?

Because I'm using ChatGPT, I'm using Copilot now from Microsoft. There are these very powerful moments like, wow, I can't believe I could do all this, but it's almost very subtle. It is very interesting in that regard.

I think there's a real challenge on how do you make this stuff a reliable part of the technological landscape? You could think about the hallucination problem or its creativity landscapes. It's like, how is this really going to play into things? I don't know.

Something you and I have talked about a bit on the podcast too is specifically in the analytics world, there's some things where we think generative AI has some real legs, but are we also accidentally just going to forget about machine learning applications along the way here too? I don't know where we'll be.

Rob Collie: Right there. You just made a distinction between AI and machine learning that most people don't even-

Justin Mannhard...: Well, because we used to call machine learning AI.

Rob Collie: I'm just now calling it AI all the time and just painting with a broad brush, even though we just had an episode where we talk about decomposing it into all of its components and they're all very different from each other. Okay, so what to do about today's uncertainty? First of all, just breathe. No one knows.

Justin Mannhard...: Sam knows.

Rob Collie: No, he doesn't. He doesn't.

Justin Mannhard...: He knows he wants $7 trillion for chips.

Rob Collie: Well, right. I've been in rooms with people like him at moments like these and seen just how wrong they are. It's not like I was right. I wasn't even making predictions about these things that I'm making fun of.

Some of these people were being paid $10 million a year at the time. Maybe if they were paid $50 million, they would've been right, but $10 million clearly wasn't enough. So if someone says that they need another $7 trillion.

Justin Mannhard...: $7 trillion.

Rob Collie: It's even hard to imagine what a trillion dollars is. I saw something one time where if you stacked a hundred dollars bills, a trillion dollars would get to the moon, and he doesn't want one of those.

Justin Mannhard...: Seven.

Rob Collie: He wants seven of them. I mean, it's not to the moon, I forget, but it's way out there into space. It doesn't stay in the atmosphere, the stack of hundreds. So number one, relax. If we need another $7 trillion, Sam doesn't know where it's leading either.

Number two, focus on the things that are knowable. Are you done instrumenting across your business in a way that you can see everything at all times, right?

Justin Mannhard...: Right. Are you done?

Rob Collie: My observation looking around is that we aren't close to done. The episode we did with Austin Senseman, his thought experiment of just go walk into a downtown somewhere, walk into a random office, and you will discover if everyone was cooperative with this weird stranger that just walked in, you will discover there an opportunity, multiple opportunities to just absolutely revolutionize the way that they operate. And sometimes in the space of a few days.

So even "just" in quotes, "just" really leaning into Power BI and taking it to its natural conclusion and applying it optimally, which we discussed in our previous episode, I don't think the world's done with that.

Justin Mannhard...: No.

Rob Collie: It's not a task. It's not a slog. It's like an absolute gold rush. It's good news if you're not done because there's so much left, incredible, multiplicative value to be created. And by the way, that sets the stage for all the stuff that Microsoft is doing with AI and machine learning is centrally predicated on good data models.

And good data models are what make Power BI work. So while you're getting tremendous short-term impact on your business, you're also setting the stage for whatever comes next. Even though there's a lot of variability, we don't know everything that's going to come next. You're still well-prepared.

Justin Mannhard...: There's a word of caution in a way. We've just very recently had some interactions with some of our customers. And I think this is a trap that's very easy to fall into where they realize there's these imminent opportunities or challenges in our business, things happening in the market or the economy, and we think that we really need to level up into all this AI stuff, so we're prepared to deal with these problems.

Rob Collie: Completely natural.

Justin Mannhard...: Completely natural.

Rob Collie: That's what the market's been telling you you need to do.

Justin Mannhard...: I fall into this trap. You could ask my team and be like, "Yeah, Justin gets all whiz banged out about this stuff all the time." But what I always try and bring the conversation back to is, okay, is this really just a little bit better analytics problem masquerading as an AI problem? Are we getting what we can get and what we know we can get from tools like Power BI before we just leap and say, oh, we just need to be in a different ball game altogether?

I think that's especially important with the hype around the generative AI stuff. I don't see anything clear yet on how that's going to thread the needle with analytics. I've seen amazing things with image creation and people making video games and all this cool stuff, right?

Rob Collie: Yeah. Over the past year, one of my friends, Chris Ray, who's actually been on the show, I've watched his AI generated cars get a lot better on Facebook.

Justin Mannhard...: Yeah.

Rob Collie: Okay. That's helped me feel a bit more like, ah, I've been to this rodeo before. No one knows what they're doing. No one knows where this leads. I would've thought that based on the hype, a year later, he'd be driving the AI generated car. You know what I mean?

Justin Mannhard...: Yeah, he would've built it.

Rob Collie: He's just having a little bit better success with it. Yeah, so one of the stories I've been working on in the background for another project is if you're a business leader, your instincts about where your problems are, are so much more correct than you would ever dare believe.

All this noise from the outside inflicting FOMO and FOBO on you makes you think, oh, I need an AI strategy. Well, you might, but as we've talked about, it's not really AI strategy. It's like, what can AI do for you? And places where your business can be improved.

The biggest obstacle I think any business leader faces today when thinking about data is that you've been trained subconsciously to think so many of your problems aren't solvable. So they don't even really consciously rise all the way to the top of your brain as a problem or as an opportunity to fix.

You get really good at assuming that those are just part of the fabric of your existence, lowercase F, fabric, and there's just part of the background and there's nothing you can do about them.

If you get a few small wins, a few small projects under your belt, you'll suddenly realize all of those problems, all those inefficiencies, all those missed opportunities, all those leakages, they are solvable and they're solvable on relatively short timeframes.

And Power BI, as amazing as it is, is just read-only. It's just seeing the things that are already happening in all of your systems. It doesn't change any of the ways you operate. It changes the way you think. It's visibility, right?

Justin Mannhard...: Yeah.

Rob Collie: It helps you see what's already real, and the thing is, you can't see it. You can't see what's already real today. BI is like massive, massive, massive leverage on your visibility.

Justin Mannhard...: Was talking with a customer just yesterday, we were just sort of checking in. We've been working with them for a long time now and we've done all sorts of things, and that's sort of similar with you. Let's go back to some of this catalyst moments and it's like now I walk around with my tablet and I've got everything I would ever want to know right here.

I forget how amazing that is. I just know there are so many companies, so many people out there where that's not their reality right now. You don't know everything you need to know or want to know, and you don't have it at your fingertips.

Rob Collie: So even the person in what seems like the late stages of completion on this, knows that they're just scratching the surface of opportunity.

Justin Mannhard...: Yeah.

Rob Collie: Okay, so that tells you a lot. If you want a simple version of this, two different ways to approach it, number one, what would you do with perfect vision? If you had God-like vision into everything going on in your business, what would that change?

I've taught some people who found that's an interesting way to think about it, but what would be the business impact of that? And once you start them at the business impact, now you can actually start to get crystallized about Power BI is the God-like vision tool.

Another way to look at that same thing is like, well, what are all the things you can't see today? What are all the things you don't know? And again, certain personalities will do better thinking about it one way versus the other, but it's really just the same question asked from flip side of the coin. But that's just the read-only vision thing. Then there's all of the business processes that ultimately devolve into manual processes.

Justin Mannhard...: Just air gaps, right? The air gaps.

Rob Collie: Yeah, air gaps between systems, right? Someone's got to do the equivalent of take the bucket of water from this pipe and walk it over and dump it into this other pipe.

Justin Mannhard...: Some splashes out of the bucket and muddy path in between pipes, real mess.

Rob Collie: There's error. Did I dump it in the wrong pipe? Yeah. Happens every now and then, right? The same spirit of up-tempo short timeline type of project has basically come to everything like that. Automated workflows and triggers. There's rules, branching workflow rules.

There's power apps that allow when those pesky humans do have to have some input, but you can deliver these apps and activate them only at the times that are necessary for the right people with the right constraints to reduce error, but also to make things faster.

Who wants to hand type in the names of customers? Not only do you get them wrong, but it takes you a while anyway. It's really more I think about identifying the problems and opportunities than it is about mapping the tech.

Justin Mannhard...: I agree.

Rob Collie: It might turn out that once you have that catalog of problems and opportunities, and again, it won't be complete, but just a lot more complete than what you dare to make today. Once you have that, it might be that AI or machine learning is the right answer for some percentage of those even today. Don't work at it backwards.

It is tricky. I am sympathetic to it because in order to identify the problems and the opportunities, you kind of have to know what's fixable. You have to know what's doable. You have to have some sense of the possible because again, I really do think that most things like that, you're just going to unconsciously filter them out. You're not even going to put them on your list of problems or opportunities because you wouldn't dare think that that was addressable.

Justin Mannhard...: It's funny too, with this backwards thinking you're describing where you think about how can I use a large language model or how can I use generative AI? Just log into LinkedIn and you'll find 100 companies telling you how that's going to change the world for you.

It's funny, I'm having all these near-term experiences that relate to this conversation. Yesterday, somebody asked me, what is P3 doing in this space? Oh, what's P3's service offering around Copilot going to be? I was like, well, I'm just waiting for Copilot to not hallucinate. And that's why I always come back to there's just so much more that's known and valuable and doable and attainable.

I think just bringing the conversation back all the way around, that idea of, well, we need to build a company that's wired for that way of thinking, to move fast, to be flexible, be adaptable, like we say here. And help people understand like, oh, no, that problem that you thought just wasn't even touchable, let's do that in a couple of weeks.

Rob Collie: Yeah, there is no shame in demanding that problems be tangible and solutions be understandable.

Justin Mannhard...: Good. It's a good thing.

Rob Collie: One of the jokes I used to tell when I taught classes was a Bill Jelen joke originally, but then I added on. So Bill's joke was there's only two industries in the world that refer to their customers as users, the software industry, and the illegal drug trade, and-

Justin Mannhard...: That's good.

Rob Collie: And I added onto that, which was as a software engineer, I had one of the only jobs in the world where when I made a mistake, my customers would blame themselves.

Justin Mannhard...: Right.

Rob Collie: When something would go wrong. When you get an error message on your screen from software, most people say, "Oh my God, what did I do?" No, you didn't do anything. That should never have happened.

If you get some mysterious error message and you ask yourself, what did you do? Some software engineers somewhere made a mistake, so the parallel to that is what are you doing about AI? And you don't have an answer. It's kind of like saying, what did I do wrong? No, what did the software do wrong?

The software world owes you a clear value proposition explained. If it's mystical, you almost have a duty to tell them, no. It's on them. It's on the tech. Don't make me feel uncomfortable about what I'm doing or not doing. Explain yourself better. We should hold everything to that standard.

Justin Mannhard...: That's a good summary of my anxious feeling around AI.

Rob Collie: Yeah. Why do we do that? Why? Why do we feel uncomfortable that Sam Altman has yet to explain truly what it's doing?

Justin Mannhard...: Yeah.

Rob Collie: Show me some absolutely middle of the road, widely applicable use cases.

Justin Mannhard...: I think my favorite use case so far, Copilot's summaries of meetings, for example, is actually very good, but I'm just like, okay, but that's not moving my needle.

Rob Collie: Incremental.

Justin Mannhard...: Incremental, incremental.

Rob Collie: That is the game changing, world beating technology of AI being used to deliver an improvement that is more incremental by far than all the stuff we're living right now because the other stuff is tangible. It's directly applicable to so many problems, and it's so clear how it can be applied to those problems, and it's so clear what the results look like.

Justin Mannhard...: This is fun because this is in this phase of our conversation, not where I expected we might end up, comparing these two inflection points. One inflection point, the one where you decided you wanted to change your career trajectory, you could see very clearly the application and what it was going to do, and you told us all about it.

You had a blog and you were showing people. Now the world has seen and it's Power BI and all that sort of stuff. This inflection point is similar, but very different. I think that's really interesting because it's not obvious and it's sort of like [inaudible 00:45:47]. I know you're big on fabric and you're a big believer in the team and all that sort of stuff, and maybe that's a different piece from generative AI.

I think that's a really important insight for business leaders is if you can't see clearly the how and the what and the result you're after, that's a clue to either be okay with the reality that you're entering some level of experimentation zone where failure is likely, or you just don't worry about it, don't worry about it right now.

Rob Collie: I think the rule that I just invented for myself or reminded myself of is when it's unclear, it's not on you.

Justin Mannhard...: Yeah, that's great.

Rob Collie: Before you even decide what to do about it, get your head in the right space first, which is this is not your fault. The nerds, the software nerds, they're letting you down. They might be letting you down because even they don't know.

I had one of Microsoft's biggest, baddest brains, people were terrified of this guy telling me over and over and over again that we needed to change the shape of the Excel grid, and I just kept saying to him, "I don't get it."

Justin Mannhard...: Yeah, and there's a lot of people that feel pressure to just nod and say, okay, we'd all be better off if we would just say, "I don't understand. I don't get it."

Rob Collie: Oh, man.

Justin Mannhard...: So good.

Rob Collie: This reminds me so much of what David Gaynor used to do to me when I was his young protege at Microsoft. I would do the techno babble thing to him and he would go, "I don't get it." I get angry, I get frustrated, but damn it, he was right every time. And by the time I was done spending a couple of years with him, I had not only adopted that philosophy, but I'd become his new high priest. For him, it was just one of many positive aspects of his personality, but no, no, it changed me forever.

Justin Mannhard...: That's an interesting insight here too.

Rob Collie: It comes back to the Einstein thing too, right? If you can't explain something simply, you don't understand it well enough. Do your homework, Sam Altman. How many trillions of dollars do you need before you can explain it to us?

Justin Mannhard...: Whether it's software or anything, if there's a failure to understand, it's not the person receiving that's at fault. It's the person that's communicating or building a tool or whatever it is. And I think we could take both angles on this as just being okay with saying, "Rob, I don't understand."

And at the same time, when someone says that to us, realizing, oh, okay, that's on me. We talked a little bit about this with Looch on his episode, right? About the empathy towards the user. When something's funky with the report and it's like, oh, it's not their fault. We probably need to change something about how this works for them.

Rob Collie: And the humility that these software gods would require in order to realize what they're doing and straighten up is not coming because they never get the signal. How many times a day is Satya Nadella or Sam Altman told one way or another that they're amazing?

Every time Satya walks into his house that's probably made out of diamonds, he's getting a signal that yeah, you're doing everything you're supposed to be doing, right? And Sam Altman can vocalize publicly a request for $7 trillion.

Justin Mannhard...: Which by the way, would definitely get to the moon. Well, according to the AI machine,

Rob Collie: The 100... See, we found a practical application. All right. What do you think? We solve the world's problems.

Justin Mannhard...: I appreciate you being game for it.

Rob Collie: We ended up in some interesting places. I learned some things today.

Justin Mannhard...: That's good.

Speaker 3: Thanks for listening to the Raw Data by P3 Adaptive Podcast. Let the experts at P3 Adaptive help your business. Just go to p3adaptive.com. Have a data day.

Subscribe to the podcast
  • This field is for validation purposes and should be left unchanged.

Other Episodes

Copy link
Powered by Social Snap