Knowing WHAT Formula You Need is More Important than Knowing HOW to Write It

Rob Collie: Hello, friends. In today's episode, I sat down with Justin to get his thoughts and reactions to my solo episode from two weeks ago, the one in which I shared the inflection point in my own thinking about AI, which by the way, if you haven't listened to that episode yet, I highly recommend doing so. I think it represents some of the most important big thinking that I've done in a number of years. Now, as a bit of a peek behind the scenes, I want to emphasize to you that this show has become [00:00:30] an integral part of my own thinking process, and indeed an integral part of my business interactions, my business thinking with Justin. Setting the podcast aside for a moment, Justin and I talk basically every day, of course, but there's only so much time in the week, and our non podcast interactions are more sharply focused on short and medium term business issues and initiatives.

Now, that's just the nature of business. There are so many known things that we need [00:01:00] to talk about that discussing the unknown is kind of low ROI by comparison, and so by default, that kind of stuff gets neglected. It takes a lot of discipline to go and explicitly call out and reserve time for this "big thinking." Most people never find that time to be honest. And while that still allows you to be short and medium term successful, eventually this sort of neglect does come back to bite you, especially when the world is [00:01:30] showing signs of changing like it is right now. So we've kind of organically allowed the podcast to evolve into filling that role for us. We use it as a means to figure out important things, even for ourselves. We almost never sit down to record an episode knowing what we're going to say, but instead we treat it as a process of finding out.

That solo episode that we released two weeks ago was informed by many weeks of podcast conversations, [00:02:00] both with each other and with guests. But when I sat down to record it, I hadn't had time to run it all past Justin. He heard that podcast and the thoughts within it at the same time all of you did. In some sense, of course, there's a little bit of danger in doing a portion of our business thinking in a public format like this. You're not going to see Deloitte executives hashing out their evolving picture of AI and recording the meetings for everyone to hear, but the fact that companies like that aren't behaving like we are is really [00:02:30] just confirmation that I like that we are. If we're being honest, the top execs at big four firms understand the world of data and ai far, far less well than anyone who's listening to this, even though when they go to the podium at some big conference, they do have a speech that's been written for them.

Now, I think the "risks" associated with using the podcast as a thinking mechanism, I think those risks are minuscule compared to the benefits. One, like I said, allowing ourselves this degree of freedom means that the thinking is actually [00:03:00] going to get done, two hours blocked every week for recording the show with minimal agenda other than having frank conversations that wander wherever they need to. And the fact that they're shared publicly puts a valuable pressure on us to be sharp and engaged about it. If we recorded these and never released them, the conversations would be less valuable even to us. It's a good pressure that yields better thinking, which I guess is all a long way of saying that today's episode [00:03:30] is an authentic reaction from Justin and a continuation of that ongoing and mostly public thought process. I hope you appreciate the transparency of it all and find it as helpful as we do.

As always, if you'd like to support the show and help that transparency, reach more people, you can leave us a rating and/or a review on your favorite podcast platform, or just tell a friend or colleague. All right, let's go find some things out, shall we?

Speaker 2: Ladies and gentlemen, may I have your attention, please?

Rob Collie: This is the Raw [00:04:00] Data by P3Adaptive Podcast with your host Rob Collie, and your co-host Justin Mannhardt. Find out what the experts at P3Adaptive can do for your business. Just go to p3adaptive.com. Raw Data by P3Adaptive, down to earth conversations about data, tech, and biz impact.

Hello there, Justin.

Justin Mannhard...: [00:04:30] Hello, Rob.

Rob Collie: Back for the jaw session format.

Justin Mannhard...: The jaw session. Is that formal now? Can we make it formal? I feel like we need a logo.

Rob Collie: Do the mullet people have jaws? I don't think they do. I think they've got an undefined chin.

Justin Mannhard...: The chin is knocked out.

Rob Collie: We're going to have to give them a chin. If they're going to have a jaw,

Justin Mannhard...: We'll get our people on it.

Rob Collie: All right. Yeah, top people. All right, so you said that you had some things in mind that we thought we should talk about.

Justin Mannhard...: I like your [00:05:00] solo pod sharing your perspective on AI and the clarity you've reached on a variety of things. I have some thoughts about what you said and ideas we could expand upon. I think that'd be fun.

Rob Collie: That sounds great. I've kind of gone into the "vision lab" alone. I have to go and synthesize things on my own, and then I have to go and synthesize things, that same thing, sort of like re-synthesize it with others to clarify it. So yeah, I'm all for this.

Justin Mannhard...: That's a human experience. [00:05:30] You got to put something out into the world, and then find out what everybody else says about it.

Rob Collie: On the scale of things like that, I remember in the early days of Power Pivot, pre-Power BI, I would put thoughts out into the world kind of tentatively like trial balloons, and then over time, feel more solid about them, solidify them over time as I got the right kind of feedback. But on the scale of travels, I really felt that moment, that point of clarity that I've been coming around to in last week's episode. I had a much higher confidence than normal [00:06:00] in that intermediate state. I'm not expecting to be wrong about it. It actually just feels too clear. But yeah, there's always refinement, also expansion. Let's add some more clarity around the edges. Do you have a place you'd like to start?

Justin Mannhard...: I realized you and I have been doing the podcast together for a good while now, and we've been talking about AI quite a bit. I've gone through the experience that I think a lot of people have gone through. The hype cycle caught me. [00:06:30] Oh my gosh, this is crazy. And then I think you and I both, we went through some period of honest skepticism and cynicism about what was going to happen, and we've been talking about FOMO. Your thinking that came forward last week, sort of a much clearer sense of what seems most likely, most realistic, where things could potentially go, and you start to see things maybe where they could potentially stabilize.

And I think it was just interesting. We've been riding [00:07:00] this journey to this point. I feel like we're closer. I don't know that we're there yet, but we're closer to maybe understanding where some of this stuff is going to come home. And I kind of want to start where you finished. This is the piece that if there's any one thing that has calmed me down through this whole hype cycle, is just remembering the world is not going to have a shortage of business problems to solve. You might have to correct me how you phrased it exactly in your episode, but the idea, it's more important to know what formula to write [00:07:30] than it is. How did you phrase that again?

Rob Collie: The real trick is knowing what formula to write or what code to write. What should that formula do? What should that code do?

Justin Mannhard...: What does it need to do?

Rob Collie: That's more important and more valuable than knowing how to write it? Now, obviously you need both.

Justin Mannhard...: I was thinking about this. I had a meeting with someone, and we were both just kind of sharing our own experiences, learning how to develop Power BI, learning how to build data warehouses, all these things. And in our experience, [00:08:00] everything we ever had to learn was informed by some catalyst. And the catalyst was always understanding this is the problem I need to solve, this is why I need the code, this is what I need it to do. And then, okay, yeah, I have to go figure out how to do that. Maybe I'm using functions that I've not used before. There's hundreds and hundreds of DAX functions. I don't walk around every day with them all committed to memory. I have referred to documentation quite often when I'm building solutions. That's what's given me comfort, especially the way you framed that, [00:08:30] is there's still going to be problems to solve. AI is likely to help me in the pursuit of that.

I think it's true that I will be able to do work faster, or I might even be able to do work that is of a higher quality than I am capable of today on my own, but I'll still be solving a problem.

Rob Collie: Yeah. And this is one of those places where this AI thing is the same as previous revolutions. It's not the same in every way.

Justin Mannhard...: That's [00:09:00] right.

Rob Collie: That's something that was very clearly called out at the beginning of the podcast. No, folks. It is a little different, but one way that it is the same is that we've experienced a revolution in BI that literally at least took one zero off the price tag, one zero off the elapsed time, sometimes more than one zero.

Justin Mannhard...: Yeah.

Rob Collie: Sometimes it's more than 10 times faster, more than 10 times more affordable. 10 is modest in a way relative to what the results actually look like. And that did not kill the industry. [00:09:30] In fact, it massively expanded it. It massively expanded the market. And anything that is a further accelerant, when you see an accelerant like that coming down the road towards you, this is something human beings do, we come to binary conclusions about things we have to. We're just always trying to come to a binary conclusion, even though it does not fit reality. That's what we want. We've got to be really careful about this, right? So it's coming down the road towards you, and you either say, "Ah, it's the asteroid that killed the dinosaurs," or "It's a [00:10:00] hoax." But we have experienced a 10 to 50x accelerant in our industry, and only good things happened. I honestly don't think anything that we're seeing coming down the road AI wise, even in the technical code generation space, which I do think is going to be one of its absolute strengths... It already is.

Justin Mannhard...: Yes.

Rob Collie: Even in that space, I don't know that there's room for 10 to 50x again. The human inertia in the equation is not going to allow for [00:10:30] that. We even see, when we work with some of our larger clients, just like the bureaucratic human inertia at bigger companies is already the limiting factor in how fast we can run.

Justin Mannhard...: Right.

Rob Collie: I don't think 10 to 50x is possible, but if we survive 10 to 50x, and in fact thrived, why should we look at this and think planet killer or hoax? There is absolutely a nuanced other result. If you think of yourself as a Power BI developer, yeah, this is a problem for you.

Justin Mannhard...: Yeah.

Rob Collie: A version of copilot that doesn't exist [00:11:00] yet that can write the DAX is a real threat to a power bi developer's business model, a power BI developer's career. By definition, the way I look at Power BI developer is you take specific requirements and translate them into data models and formulas.

Justin Mannhard...: Right.

Rob Collie: That whole model doesn't work very well. Writing requirements and translating them, you never find out that the requirements could have been much better, that the requirements sucked. You don't get that iteration that you [00:11:30] need. Our business model allows the stakeholders, the domain experts, to stay close to it, and we could iterate fast and we can improve and discover what the requirements should be rather than pretend that we know them.

But anyway, if you're being handed the requirements and the requirements are somehow correct, that's what copilot is going to do. That's the thing that copilot is going to do a large part of, is translate those requirements. So last night, I experienced an example of this that drove it home for me. These data models that I've built, it's really one data model to rule them all [00:12:00] to help manage my wife Jocelyn's health conditions. Having a personal example that I can share, this is so helpful, so I'm going to just briefly describe it to you. There are two medications that she takes. One of them we're trying to reduce. We're weaning her down off of it because it has long-term not great side effects, right? It's used to manage a condition, but we think we can get her off of it, and then eventually the condition won't be a problem either.

But when you start cutting down on this medication, side effects from the cutting, from [00:12:30] the reduction, the tapering start to come up. So there's another medication. Sounds like the I don't know why she swallowed the fly story, where the second medication is used to dampen those side effects. That second medication that dampens the side effects, you can't take too much of it. It has its own side effects that get bad, right? So you've got to titrate just perfectly. Okay. These reductions in medicine one, as we're tapering down, I had this chart that showed the reductions each night. Some nights we reduce. Some nights we don't. And I'm comparing that to symptoms, [00:13:00] and also to how much her blood level of the mitigating medication is. And it just occurred to me last night looking at these streaks of when she was cutting and when she isn't, so not helpful.

What's really helpful is to look at the window of recent nights. How many cuts have we made recently? How big were they? But also how recently? I arrived at this revelation. What we need is a 14 day trailing window, and I multiply the cut amount on the most recent [00:13:30] day by 14. I multiply the cut amount from yesterday by 13 all the way down to one for 14 days ago. 14 days ago is less important, less impactful than recently. And then I threw the other measure of how much have we been cutting, just threw it off at the chart because the buildup of the cuts is what really gets you, right? And so now I'm in a completely different space. I'm looking at a completely different set, and now I can look back in time and say, okay, truly four weeks ago, we experienced exactly this point in time as today, [00:14:00] and I would never have been able to do that before.

Now, a really, really good version of DAX copilot doesn't exist yet, but would've helped me write that measure faster. It involves dates in period function and the SUMX function. It's a good intermediate level of complexity. But even then, it definitely wouldn't have told me to write it. It definitely wouldn't have told me this is what you need. By the time we get to that where it can tell me that I really should be doing the other thing, I think that's artificial general intelligence, in which case all bets are off. [00:14:30] We're in a completely different world, right?

Justin Mannhard...: Country borders are going to start breaking down at that point. Yeah.

Rob Collie: Maybe the earth itself, right?

Justin Mannhard...: Yeah.

Rob Collie: The old joke is going to turn everything into paperclips. So that insight, knowing that I needed to look at the problem differently, if copilot had written the formula for me, it wouldn't have saved me that much time. Now, the difference is that I might've been the person who couldn't have written the formula.

Justin Mannhard...: That's right.

Rob Collie: That'd be a big deal, if I didn't know how to write that formula. And there are formulas that I don't know how to write, so I look forward to the day that I have an inspiration [00:15:00] that it fulfills for me. You really, really, really need to think of your job, and I love that we're already there at P3. You've got to be a business problem solver.

Justin Mannhard...: Period. I was thinking about the code aspect and knowing what you need versus how to do it as being more important. Back when I was still teaching our Power BI classes, one thing I would do at the beginning of our advanced class is I would just ask the group, "Raise your hand if you've needed to be able to do something, you [00:15:30] googled the heck out of it, you found some code, you copied it, you pasted it in your model and it worked, but you had no idea why it worked." Everybody's hand goes up. And so I would say, "Well, my goal in this class is that you leave understanding why it works." Putting myself through this thought process of, okay, everybody that's in that situation, we've all been there and we all get back to that point where we're on the internet looking for how to solve something.

I know what I need to be able to do. Now, I've got a copilot that can help me get it done right there on the spot without me pecking around. [00:16:00] But then you just hinted at this. Then there's the evolution of the person who has no knowledge of DAX whatsoever can also get something done. And then there's the evolution where we don't even realize the DAX is there. The reason I was thinking about this is I've been using GitHub Copilot for general purpose, code C#, Python, all that good stuff. And over my consulting experience, I've helped customers create notebooks or PowerShell scripts that would scrape their Power BI service [00:16:30] to do audits and inventory of their assets. Somebody asked me for one, and it's been like a couple of years and I couldn't find it, so I just asked GitHub Copilot like, can you write this script? And it did. Not only did it write the script, but I asked it to notate it with comments to explain what was going on, and it did that too.

It got a couple things wrong, but it's like you think of that world where not only can it write the code, but it can explain it to you, if we get to that point, things get different. It will always come back to not even [00:17:00] just knowing, but having a deep understanding of what you want to be able to do so that you can explain that to something that's going to execute on it.

Rob Collie: And even in my case, the debugging last night to make sure it was right, and it wasn't. I had to cut it down from 14 to three days for a moment just so I could look at the numbers and know what they should add to and do that math in my head, and then I discovered, oh, I'm subtracting the wrong direction. I was subtracting the max date from the other date as opposed to vice [00:17:30] versa, or things like that. The referee is still really important, but if you're pinning your future career on the fact that only you know how to write DAX, well-

Justin Mannhard...: Yeah, or insert code language here.

Rob Collie: I think DAX is going to be, for some reason, one of the last ones solved, it's so far ahead. Copilot is absolutely devastatingly good against even old languages like VBA.

Justin Mannhard...: Right.

Rob Collie: A friend of mine who's like, I don't know, he's in the top 0. [00:18:00] 01% of VBA developers on the planet. He posted his whole ChatGPT session asking it to write and improve some code for him, and he is like, "Yeah, this is better than I would've produced."

Justin Mannhard...: This made me think about the granularity of bets we could make about AI. So just to use a quick sports analogy, if we're all going to bet on whether there will or won't be a Super Bowl, yes, there's going to be a Super Bowl, right? Nobody's making [00:18:30] that bet. Yes, AI is going to change a lot of things. That's a bet. It's going to change things. Okay, which team's going to win? Okay, are they going to cover the spread? You get into the prop bets. And so I think with AI, that's kind of where I'm at myself. Is there going to be a DAX copilot that's really capable? Probably. When? What parts of the lifecycle are going to go first?

There's a lot of cool things happening with Python and machine learning, code writing. We just mentioned GitHub Copilot. We've talked [00:19:00] on some of our other episodes about how generative AI is helping with documentation or maybe some of the grunt work oriented aspects. That, I think is still a bit of a betting game. Which parts of the lifecycle is AI going to come in and take down first? Where our company is going to pour their investments for those types of capabilities? To me, that still reads as a bit of a crapshoot in terms of timing and what's going to happen first, but you see some of the sales presentations from not just Microsoft, but the other vendors [00:19:30] as well, if they get some of this stuff, it's pretty compelling that people are going to approach analytics very differently.

Rob Collie: My top line takeaways from all of this, number one, a version of copilot that helps a lot with power BI development, we don't have it yet, but it is coming. It is a solvable problem for them to build a version that's really helpful.

Justin Mannhard...: Right.

Rob Collie: It's not helpful yet, but it's solvable that they will build one that is helpful. And the impact on the industry [00:20:00] will be primarily just another accelerant. It's not going to remove the need for analytical expertise.

Justin Mannhard...: No.

Rob Collie: Now, that is one of the lies that companies are going to tell in their marketing because that is a very compelling lie. We've seen it.

Justin Mannhard...: Yes.

Rob Collie: Tableau became a very, very, very highly valued, publicly traded company, based in large part on being the most effective teller of that lie that analytics was [00:20:30] now like finger painting, mashing bar charts, and getting different answers, right?

Justin Mannhard...: Follow me this way for your Alteryx licenses.

Rob Collie: Yes, that's right. Yeah. I'm frankly pretty excited about it. I'm ready for a version of copilot that speeds things up.

Justin Mannhard...: Right.

Rob Collie: Again, I've lived this revolution where there was so much drudgery in a project and not enough thinking. The citizen developer, the Power BI revolution just [00:21:00] dramatically cut down on the drudgery, which gave us more time for thinking and more ROI on actual good business improvement thinking, something that makes formulas faster to write and/or, actually both, makes formulas easier to research. Yeah, I completely welcome that. The way we've constructed our company around that previous revolution that's still playing out puts us in, I think, a great spot for this next round. I've been amazed at how long the traditional [00:21:30] model of data consulting has hung on.

We still have a tremendous number of competitors who operate on that awful, glacial, predatory, slow, expensive model. I've been shocked at how resilient that has been. You can't kill it, like Rasputin. Maybe this will be the thing that breaks it through, right? So the idea that thinkers aren't going to be needed. For a moment, even when I started thinking about what DAX copilot would be like if it were really good, there [00:22:00] was a moment of misplaced uncertainty, but gosh, just even that formula last night, I'm just like, oh my gosh.

Justin Mannhard...: That was so good.

Rob Collie: The computer is never going to do that for me. And by the time it does, none of us will have any human physical needs that are unmet. We'll be living in a world of Star Trek, right? Material needs just don't exist anymore.

Justin Mannhard...: I'm curious what you think about this. I'd be willing to predict that when these technologies mature [00:22:30] and they get to that level of capability, if you look at from a budget point of view inside of a company, the budget that's put aside to staff data analytics and AI teams will be the same or higher, meaning there's just going to be more throughput.

Rob Collie: The ROI is better.

Justin Mannhard...: The ROI is going to be better. There will still be analysts. There will still be data scientists. We might start calling [00:23:00] these things different things over the years, but there's still going to be... I love the way you said that. There's still going to be thinkers that are still probably going to be using tools like Power BI that look and feel a lot they feel today. The throughput's going to get better, the drudgery is going to go down. All of my experiences that I've had where I finally cracked the nut on a tough formula or a tough metric or something. Yes, there's a relief because you're like, oh, good, I got that done, but it's not like I'm going to write [00:23:30] that same formula a hundred times in the future. Projects are going to keep going. It's going to be sometimes gradual evolution. Sometimes we'll probably feel like acceleration of change, but we're all still going to be needed.

Rob Collie: Yeah. When we got this previous gift in the form of Power BI, if you just looked at the state of the world at the moment when Power BI arrives, all the activity that's happening at that moment in time, a lot of those activities were at risk because of Power bi. The net impact though is that we now have many, many, many [00:24:00] times more BI professionals in the world. Overall, BI spending is up.

Justin Mannhard...: Right.

Rob Collie: Even with Microsoft's software is affordable and projects move faster, and still spending has gone up. Like you say, the ROI is there. It's provable ROI. It's a good multiple. It wasn't before. So it'd be pretty silly of me to be someone who absolutely championed what that first wave was going to do and become like [00:24:30] a curmudgeon about the second one. So I'm not going to do that.

Justin Mannhard...: Right. There's just so much to be optimistic about.

Rob Collie: Something else I'd like to reemphasize, this two-thirds, one-third rule.

Justin Mannhard...: That's great.

Rob Collie: The human tendency to try to reduce something to a binary conclusion. So we went through, you and I, if we go back over the last year, there was a moment of falling for the hype, or at least being afraid of where it leads, and then we started to have our more skeptical thoughts. I credit us for being [00:25:00] early there. The wave of articles that's out today are saying it's just a farce. AI is just a farce. It's this total waste of time and money. It's crazy how strong the backlash has been.

Justin Mannhard...: Yeah, it's amazing.

Rob Collie: But that's wrong too. And so it's just a question of what truths you're connecting with. And I do think that two thirds of the hype and the activity and everything around this is a social phenomenon. It is hype. But you think about what the one third remaining is, that's [00:25:30] a massive thing. Because think about how big the hype is.

Justin Mannhard...: Totally.

Rob Collie: Is anything ever captured the tech space, the business productivity space, the amount of talk, the amount of total percentage of the pie of all conversation and activity and thinking, and probably angst, right?

Justin Mannhard...: Yeah.

Rob Collie: That is a monster, monster, monster pie, and a third of it is very significant. So don't get caught in either model. I'm committing to myself that I am going to be [00:26:00] using this rule of two-thirds, one-third.

Justin Mannhard...: I love it because it keeps you not necessarily in the center, but more or less grounded from getting too ignorant of how technology's going to change, or far too excited about it. That's a good rule. I'll adopt the rule.

Rob Collie: We've made multiple rules today.

Justin Mannhard...: The other topic, I was curious if you wanted to expand on a little bit about the AGI. I think you had a good take on what needs to be true. You believe it's possible and it's going to happen, but [00:26:30] the truth of what needs to come to bear just seems so daunting. I struggled making that leap, I think.

Rob Collie: When I realized in college that my brain was made up of these neurons that are essentially like analog circuits and that there's nothing special about the biological wet wear in my head... If you went one by one and were able to replace every neuron in my brain... And there's billions, right? If you replaced each one one by one with a transistor that somehow mimicked the function of that [00:27:00] neuron, I would never notice. By the time you get done replacing all of them, I would've thought nothing changed. And so we absolutely can build such a machine, and I think we absolutely will at some point. That is scary. Because even in college, when I came to that conclusion, my very next conclusion was, oh, and when we do that, we should never plug that thing in.

Justin Mannhard...: Right.

Rob Collie: Now, on the flip side, if we were somehow able to do this and have it be benevolent, [00:27:30] so many of the problems that kind of wig me out, the course that civilization is on seems a little quasi sustainable, not even quasi. It just... I've got real angst about where we're headed. People don't really think about what happens if sea level rises by the predictions, right? If sea level truly rose the levels that they're talking about, everyone just looks at that and goes, "Well, I'm in land enough." You are not in land enough for a world where New York City's underwater.

Justin Mannhard...: That's right.

Rob Collie: So many things break [00:28:00] down. We saw what happened with COVID. We're so interconnected that a disruption of that magnitude, losing coastal cities, ports not being usable anymore. We need to make new ports. Society runs out of the ability to cope. And then guess what? There's no food in the stores. So I welcome AGI. I really do believe that humanity's on this sort of all or nothing course. We kind of need AGI. If we get it wrong, ugh. And that's all I have to say about that.

Justin Mannhard...: [00:28:30] The collision course with environmental reality, whatever timetable that's on, compared to the timetable to solve that problem, and if AGI can be a part of that, because Sam Altman wants a crazy amount of trillions of dollars because he knows he needs computing hardware, both a sheer amount of it and it significant advancements in its capabilities. So you think about all that manufacturing, the electricity that's involved. It's like, can I [00:29:00] get there? And at what point something disrupts the supply chain and the interconnectedness of anything? We've been advancing as a species in a society so rapidly for such time now, it's like the brakes suddenly slam on. You can't deinnovate.

Rob Collie: Yeah. The other thing, Justin, that happened this week was you and I simultaneously coming to a very similar conclusion offline from the podcast, which is AI is not just copilot for DAX. It's so many things.

Justin Mannhard...: That's [00:29:30] right.

Rob Collie: Compared to BI, the whole concept of AI and what should AI do for your business is a lot more custom per business. Even the value proposition of it is different per business. BI and dashboards is relatively universal value proposition, actually being able to see everything that your data was trying to tell you but wasn't, that's pretty universal. Even there, obviously each dashboard is different, each business is different, right?

Justin Mannhard...: Sure.

Rob Collie: For the [00:30:00] topic of AI, decomposes so quickly into at least five or six different buckets

Justin Mannhard...: Easily.

Rob Collie: And it's almost always by definition, clean sheet, something that you've never done before, whereas you can imagine a dashboard that's a better version of your Excel report, or a dashboard that's a better version of these paginated reports that you're cobbling together. In terms of our role as a consulting firm for AI, a couple of things I've been crystallizing for us. Number one, this same clarity that we've been [00:30:30] developing for ourselves. No one's doing this for the world, right? Everyone's talking their own book, everyone's just trying to get clicks, and most people who are doing this have really no idea what's going on. That's the truth. The corporate boards that told all these software companies to freeze everything and go into AI, not one of them understood what was going on. Not one of them was an AI researcher or a software expert or whatever. The entire phenomenon is being led by people who either don't know or aren't being completely transparent.

There's [00:31:00] a reason why clarity isn't being trafficked in, so we're developing clarity for ourselves, and just about what the space looks like, what it can and can't do, how to think about it, how to spot opportunities. That clarity is a tremendous value that we can and should be providing to our clients.

Justin Mannhard...: Totally.

Rob Collie: Even just getting business leaders at our clients to the same level of clarity that we have worked hard to achieve is a very, very valuable thing that we [00:31:30] can do.

Justin Mannhard...: It's overwhelming. Five buckets with each bucket having hundreds of sub buckets. That's like the reality of this. It can be very overwhelming to think about where to go here because it's so fresh. When we say, "Oh, I need to do a BI project," you can pretty much imagine in your brain what the assets and the deliverables and what the work is going to feel and look like. With AI, there's so much, and everybody kind of needs to decide, am I trying to infuse AI into something I go to market [00:32:00] with? Am I trying to bring AI into how I get more efficient internally? Am I trying to make this process smarter? What's actually going to give me bang for buck? Where can I get started? We love to say this. Even with BI projects, I think a lot of our interest is, how can we help people get off the starting line? Let's get into the race because you can definitely get into the race.

Rob Collie: Yep. And something else that you said offline with me, and this is always true even with BI, but more true now, rewind a bit and imagine people were saying, "What are we going to do about Power BI?"

Justin Mannhard...: Right.

Rob Collie: It [00:32:30] wasn't really being asked that way. What do we do about AI? Well, that's where we're at. The answers to what should we do about AI, by definition, we, P3, won't have them. You, the customer, you will have them, but in order to have them, you're going to need that clarity.

Justin Mannhard...: That's right.

Rob Collie: So in terms of us thinking about our AI services portfolio, we've really turned the corner here, which is clarity is kind of service number one. What does that service look like? Is it a class? And we're not quite [00:33:00] done defining that yet, right?

Justin Mannhard...: Right.

Rob Collie: It's a really important pillar. And again, everything looks obvious in hindsight, but it was not easy to come to this point, to, A, develop the clarity, and then B, realize that that is the thing, the first thing.

Justin Mannhard...: It's interesting because we've been saying some version of this since the hype cycle got started. AI needs to find its way into our services portfolio. We kind of knew that from the jump. This is something that's going to need to make sense for us. We weren't quite sure [00:33:30] if that was totally true or where it would play,

Rob Collie: And again, we should clarify it means many different things. Machine learning falls under this AI umbrella. We've been doing that for a long time. But what about the generative AI thing? When you say AI in this case, you mean generative AI?

Justin Mannhard...: Yeah, I'm talking about generative AI. And again, there's so much breakdown. There's generative AI in the context of analytics. There's generative AI in completely non-analytical contexts, support bots or being more efficient in your job, or [00:34:00] adding capability to your software, whatever that might be. So yeah, we're kicking around some ideas, and I think we need to get more engaged with customers and business leaders on this topic and share the clarity we've arrived at and bring more people into this conversation because I think that's how we're going to figure out, well, what do we really bring out into the marketplace that compliments everything we do with our great team on Power BI, Power Platform, Azure, and more? I think that's where we'll find the answers.

Rob Collie: As a side effect, if this becomes the reason to [00:34:30] have you and I doing something together with customers... Have you and I ever given a talk together?

Justin Mannhard...: I think the closest we came was when it was my first year. I taught Power Query after you taught Foundations.

Rob Collie: Wow, that is amazing. We've managed to never cross those streams.

Justin Mannhard...: You came and sat in for a half a day at the Bar Keepers Friend jumpstart or whatever. You know what I mean? But we've never done something. Yeah.

Rob Collie: I mean, this podcast is obviously significant, but we've never done anything [00:35:00] that qualifies as a service for our clients as a duo.

Justin Mannhard...: Never.

Rob Collie: Or even been on the same stage like at a conference.

Justin Mannhard...: We should fix this problem.

Rob Collie: Thanks for listening to the Raw Data by P3Adaptive Podcast. Let the experts at P3Adaptive help your business. Just go to p3adaptive.com. Have a data day.