AI, Actually – Episode 11: Open AI’s Playbook for Scaling AI, Why Generalists Are Winning, and Revenue-Driven ROI

Do you guys agree? Is the old playbook dead? What does this rhythm sort of look like ⁓ in the real world?

Jim (01:22)

Generally, I agree with that premise, but maybe I’d go a step further. The old playbook was sort of about the relationship between IT and the business. And this is a fundamental sort of misunderstanding is that this is an IT problem. It’s an IT and the business relationship problem.

And IT, in my view, needs to go much further in owning and understanding the business problem than they ever did before.

because so much of this is not an underlying technology question in terms of bringing AI solutions to life or intelligence solutions to life. It’s sort of being able to express what the business wants in a functional way in the context of feeding it to the model, et cetera. We can talk about all that. The other side of it is the business has to go much further than they ever went before.

They sort of in a world where they sort of threw requirements over the wall at IT, that doesn’t work anymore. There’s a different kind of chasm and the business has to go much further in embracing the opportunities and that large language models provide or that generative AI provides and in understanding how to bring it to life. And right now they’re sort of sitting on one side and IT sitting on the other side trying to run the old playbook.

And the chasm is both of their problems and both of them have to go much further. ⁓ react to that a little bit guys, but I think it’s not just an IT problem. I think IT sort of got it wrong. think they got to go further, but this is not just an IT problem.

Pete (03:08)

Look, I am sympathetic to IT organizations right now because it’s got to feel like completely Wild West. And frankly, it is. Unfortunately, some of the instinct is to sort of control, right? And to sort of, you you can’t use this and you can’t use that or whatever. And I think the answer is a little bit the opposite of that, which is, okay.

We know everybody’s going to use this. Here are the ground rules. You can do this. You can do that. sort of go from sort of control to much more shifting into enablement. I think that’s first a mindset shift that I see. One is the speed that this stuff is moving at is just very hard for large enterprises to sort of even comprehend, to be honest. And so there’s got to be just ⁓ much more emphasis on

Jim (03:47)

Yes.

Pete (03:59)

How do we go fast? How do we stay very nimble? And how do we really start to take advantage of some of these things? And one of the ideas that we’ve talked about is you’ve got sort of the technology teams and you’ve got these business teams and the technology teams feel like they’re going to, I want to pick one platform and one model and I’m going to make it available to the business and the business is going to do what they want to do. Well, that doesn’t work for a bunch of reasons. First of all, there’s not one tool right now. And by the way, if it’s right today, it might change next week.

So that has to be very nimble on the business side. They don’t, they don’t have a clue necessarily. They know what they want to do. They understand the problem really well. They have this domain expertise, which is critical, but they don’t understand how to leverage these technologies. So I really think you need sort of these tiger teams where we’ve talked about forward deployed engineer teams with some mix of technical capabilities and you know, business acumen and domain expertise and the vision to sort of go to this next, the next place where this technology can take you.

I see very few organizations sort of set up that way. And that’s just, that’s one way I think you could really embrace this technology at the speed that you need to.

Alon (05:06)

the I guess.

Shanti Greene (05:06)

Yeah, to

steal an analogy from like supply chain, it’s a last mile problem where getting something 80 % complete using the AI tools that you have available is pretty straightforward and pretty easy. And IT can do that, but actually solving a real business problem, you need to understand that problem really well and what it means for a solution to be good enough to use. that closing that last mile gap is really challenging because you can hit

Pete (05:11)

Mm.

Shanti Greene (05:34)

a large percentage of requirements and get most of the things done, but still not have something that’s truly useful. And that’s where I think we’re going to see a lot of development happening on the agentic side and the model quality side is really trying to close that gap. How much closer can we get?

Alon (05:50)

Yeah. Maybe I’ll take a slightly different direction. The thing that’s occurred to me probably more recently is, if you think about who’s most productive ⁓ in this age when AI capabilities are just kind of moving along and giving you more and more power, I think it favors the folks who are more generalists. So if I’m able to think broadly, if I’m able to process

Pete (06:14)

That’s interesting.

Alon (06:18)

Here’s the problem, whether it’s a business problem or a technical problem. And I’ve got enough of a framework and a mental model for how do you solve for this problem? Then suddenly the AI is a great assistant. It sort of shores up the places where I’m weak. It makes me faster in the places where I’m strong. But it’s a great teammate. And so in…

one way or another, you’re trying to kind of figure out, what’s the team, what’s the smallest possible team that allows me to get broad coverage of all the domain knowledge required to solve a given problem?

Pete (06:46)

Mm-hmm.

Alon (06:54)

And it could be a team of one, it could be a team of three, it could be a team of 30, you know, in very complicated areas. But it’s very hard to do like if you’re doing it in silos where you’re handing off, we did our part, like here’s the specs, we built in with AI and now they’re very deep specs and you got to hand this somebody else. Well, we’re going to feed that spec to an AI and the AI is going to do something with it and stuff’s going to get lost in the translation. where historically I would say, know, there is, whether it’s software development or whether it’s, you know, project

Pete (07:10)

Mm-hmm.

Alon (07:24)

management, there’s always like, you know, sort of specialists who take a lane and operate in that lane. You’ve got to be able to ⁓ sort of fuse the people who are working on the stuff enough so that they’re sitting, you know, if not physically next to each other, they’re very tight in their workflow.

Pete (07:41)

Any

other

Alon (07:41)

And then AI is sort of ambient in that scenario where it’s just part of the entire process. Everybody’s relying on it to accelerate and to double check and to create faster, higher quality results.

Jim (07:58)

Success in recent years, sort pre-generative AI, was grounded in having individuals who had tremendous depth in their space.

Maybe it was software engineering, or maybe it was sort of functional business context ownership or, ⁓ you know, whatever, but sort of just that, that, that depth that we just had to have now we’re in a place and I’m, I’m, I’m agreeing with you here Alon that somebody who’s sort of medium depth in their functional understanding and medium depth in their understanding of sort of AI skills and have some depth in a little bit of it. That person is the glue and just so critical.

and to your point about, one person who has medium depth in all those things may go a lot further to delivering a solution than a team of three who have depth, great depth and exclusively in one domain or another. And it’s what that one individual can do. Just to sort of make one other point here, I think calling something an AI project.

Alon (08:53)

Yeah. Yeah.

Jim (09:05)

or what, you know, after sort of core enablement of our team, like we want to, and I agree with that. We want to roll out enablement to our organization just to give them environments that are safe and to allow them to build skills. But when we start talking about enterprise use cases, characterizing something, this is an AI project or how is your AI project going? I think it’s a total misnomer. These things need to be characterized based on what the business problem is we’re trying to solve or what the opportunity is, you know, we’re trying to optimize our supply chain.

Pete (09:06)

Yeah.

Jim (09:35)

or we’re trying to enable revenue lift through this new product. The fact that AI is part of what’s going to provide for that and because of its capabilities to bring intelligence into the problem that we’ve never brought before, or at least bring intelligence out of the humans into the computer in a way that we’ve never brought before, that’s awesome. But don’t characterize it as an AI project.

Alon (10:02)

I think it’s just easier to sell

it. It’s easier to sell if it’s got AI in front of it. There’s budgets for that. Yeah, if there’s no AI in front of there’s no budget for it.

Pete (10:05)

It’s a hot thing today.

Shanti Greene (10:08)

Yeah, you gotta get those buzzwords in there.

Pete (10:12)

Alon I think the point you made, like I always thought of, the ideal product team is a team of one. Like one part, understand the whole thing, soup to nuts, they can build it, code it, it, ship it, right? But obviously that hasn’t really been possible. That’s actually possible now. And the one thing we sort of touched on, but it’d be interesting to maybe kick it around with you guys is.

Alon (10:18)

Yeah.

Jim (10:26)

closer.

Pete (10:34)

What’s the impact to organizations? Because if I’m a product manager, like last year, my job was, I’m like, cool, I can write specs faster. I’m a product manager today. Like I’m going way further down the path, right? From spec to, first of all, I get to understand the code really well. Then I can write the spec. Then I can probably make a lot of the changes. So it feels like ⁓ organizationally,

Alon (10:44)

Yeah.

Pete (11:03)

I need to be thinking about the capabilities of people much differently, and be much more aggressive on what I think people can get done if they really lean into this stuff. And it sort of blows apart some of the traditional, I think, organization structures, which is going to be real challenge, I think, for a lot of business leaders. And I just say, to really get the benefits that we’re talking about, you’re absolutely going to have to lean into that.

Alon (11:30)

Yeah.

Shanti Greene (11:30)

Well, I think traditional timelines

are out the window at this point. It doesn’t take six months to deploy a software project. It takes six months to QA the project and make sure it’s working well, but that should be like five months of testing. It’s just, everything has changed in terms of what it takes to get a POC out, to start hardening it, and then to incrementally build on it. A lot more resources now can go into like the QA, the UAT.

Pete (11:33)

completely.

No.

Shanti Greene (11:57)

be hitting all of the actual business requirements that used to be done in functional testing and unit testing and all of the pieces are, can we get functional code? Well, we can get functional code, but now testing that it actually solves your problem becomes the real issue that you’re gonna deal with.

Alon (12:12)

So for me, I think it starts with just thinking much more ambitiously about what you can deliver. So I’ll give you a, instance, we’re ⁓ looking at a segment that we’re not in and talking to prospects. And historically we said, great, let’s go out and do.

Pete (12:17)

Right. Yeah.

Alon (12:35)

a little bit of a market research and requirements gathering. Let’s put together a slide deck. Let’s show it and get feedback. And then let’s maybe sign up for kind of a focus group. And then let’s do sign up for some kind of a pilot. And this is weeks or short number of months that you’re trying to nail the business case for building a thing. ⁓ The experience I’ve had more recently is like, well, if you can ⁓ imagine what it is that you’re trying to solve as the business problem.

and you have a decent sense for it, you’re building a, you know, call it the specs and the requirements kind of, you know, in some number of hours, and you’re building a prototype in some number of hours, and you’re showing it, you know, for the follow-up call, right? So if you have an intro call, you’re meeting and sort of nodding your head about, we understand what the problem is, and we’re going to solve it.

So, you know, a day or two later, you’re back on a call and you’re saying, okay, here’s how we ⁓ conceptually solved it. And depending on the starting point, if you’ve got

So now again, it’s some of the gory detail. If you’ve got the basis, like a reference architecture that you already like for solving that problem, and you’ve got roughly your call of the agentic rules that predefined for your coding assistant, mean, functionality is just coming to life. And it’s not coming to life in a formula, like, okay, I’m not talking about like, doing something in Figma or even like Bolt or Lovable, or where it’s like, okay, here’s a prototype and it sort of works.

Pete (13:52)

Mm.

Alon (14:13)

I’m talking about, no, this is the online architecture. You’re going to use in a project going forward. You’ve got the separation of concerns and you’ve got the strong objects, data types and the schema. That’s the thing you’re brainstorming at the beginning. That’s not the final one. But the architecture from day one is actually an enterprise caliber architecture because you’re borrowing from references that already are tried and true and are running in other projects. And so the crazy thing there is ⁓

you’re talking about in a couple of weeks, you’re already iterated what would have taken you a couple of months. you’re in users hands on keyboard playing with the stuff. And that becomes a part of your QA process at that point. So it’s not like, hey, we’re going to QA the obvious and we don’t want to put in front of anyone because it’s going to be really embarrassing when they find obvious things. You’re probably more finding business like.

you know, friction between what the user thought they were going to be able to do and how you design the user interface. And so that creates just huge acceleration. The other super recent experience I had on top of this is, this was like last night, so 5.2 Pro is available as of yesterday. Let me go back and some of the things that took me a bunch of iterations to do. Let me just try with 5.2 Pro.

Pete (15:28)

Right, yeah.

Alon (15:36)

I mean, this thing like sits in like, you know, half an hour later, it says something as opposed to, you know, a few minutes later, but it’s like, oh, it like did a lot of research. Like it came, you know, it held in this context, enough information to have a very, uh, thorough view on how to implement something where before I may have been going back and forth for a couple of hours, kind of independently and trying to decide and

maybe not even considering all the facets that I should. So I’m like, okay, there’s a new toy here where ⁓ it’s kind of like, you’re not sure about how you want to do something. And you want somebody to spend a lot of power on processing it and just kind of give me the plan that I could understand which way I should go. ⁓ That was a new experience for me. okay, I haven’t…

Pete (16:26)

Right.

Alon (16:29)

Like let it go for an hour and come back and see what it did kind of, you know, short of it actually generating code. Like, so this is not code generation. This is just planning, if you will, of like ⁓ very detailed how to go about doing something.

Pete (16:42)

Yeah, I’ve had this experience and I know it sounds like you have too, where it’s surprising how much you can get done. I’ll tell you a story. had an example the other day where I’m having to put this presentation together and I’m thinking, man, I need like a three hour block of time to knock this thing out. And I thought, all right, let me just, let me just, so I’m going to get this done. And I sort of time boxed it. I’m going to get this done in 15 minutes. And.

Alon (17:08)

Right.

Pete (17:11)

Actually worked. mean, I actually was done with with my intended final product in about 15 minutes And so I just continue to find myself I have to almost change my mindset going into a project of okay Forget about that what you think you you know Give yourself to you get it ten times done but done ten times faster and it’s shocking to me how often that that that actually

Alon (17:33)

Yeah.

Yeah. I think the other thing that happens is you just sort of start to get a feel for what things AI can do well. And so it’s surprising. Some of the stuff I go, historically, would never build, like in the first pass of the product, we would never build a whole series of, let’s say, utilities or robustness or ⁓

just stuff that comes later when you start doing the polish. And now it’s like, well, just put it in the rules file. Make sure that it understands that we want batteries included in this thing. Like we want out of the box product analytics in the first pass. ⁓ And you just get a completely different level of maturity out of what used to be prototyping sessions to like, OK, these are more like version one of a product as opposed to something earlier.

Pete (18:24)

Yep. Yep.

Alon (18:26)

The funny thing is though, it’s so uneven is that you end up with something on the screen. Like if I was trying to do like some custom UI arrow looking thing. And that took longer. Like I probably spent an hour trying to tweak the like, no, no, I want the edge to be sharper. want an outline. want to, you know, and that like the mall is just like, just, know, like struggling to do exactly what I had in my head. And cause it doesn’t know how to nudge these things like so slowly, but it could build like, you know, the 90 % of the functionality, the same amount of time that it took me to like, you know,

Pete (18:41)

Right.

All

Shanti Greene (18:49)

Mm-hmm.

Pete (18:53)

Amazing.

Alon (18:56)

make an outline arrow show up in that way.

Pete (19:00)

So one of the topics in the white paper was this idea of around fluency and building sort of an AI literate workforce. I you’re into this stuff every day. Everybody on this call is on it every day. Jim and Shanti, weigh in a little bit on, all right, how do we build this an AI literate workforce? People that have the insights that alone has and then sort of can bring them into the enterprise. What’s your perspective on that?

Shanti Greene (19:25)

I mean, I think

part of it’s a mindset shift. Like you’ve just, you’ve got to be curious and you have to want to experiment and try things. There isn’t a, this is the right way to do it. I can learn it once and it will stay that way. It’s going to change. It’s going to change really frequently. And to be fair, that just weirds some people out. They aren’t ready for that. They don’t have that kind of like continuous learning mindset where they always want to be doing that. But if you do, you’re absolutely going to be.

Pete (19:30)

Yeah.

Yeah, that’s right.

Shanti Greene (19:54)

that 10x engineer because there’s so much you can do and it keeps changing. I’ve got the same personal project that I haven’t been able to complete for going on a year and a half, two years now, but every three months I start from the beginning, give it the same style prompting and say, how much further can AI get now than it could three months ago? And we’re getting closer every time. There’s still some integration issues because like authorization across services is still a problem, but we’re getting there. I’m going to get there without having to.

Like keep having these long sessions. One of these days I’ll be able to give it the initial specs and something will run with it and go.

Jim (20:30)

⁓ You know, Pete, I know you’re a fan of Crossing the Chasm. It only comes up three times a week in conversations with us. AI world, non-AI world, I don’t know that anything changes about the bell curve that is described there in terms of innovators, early adopters, ⁓ you know, the next wave, the early… Yeah, unfortunately, it sort of is what it is and it’s sort of human nature. ⁓

Pete (20:48)

Mm.

Yeah, that’s just human behavior. is what it is. Yeah.

Jim (21:00)

And AI presents sort of a ton of fun for those innovators and early adopters right now. But we know the values there. We’re seeing it. ⁓ We’ve got to be tight about it in terms of enterprise use cases. I do think there are things that we’re seeing, that we’re working with organizations on, and that we’re seeing companies work with that sort of try to push people along that curve.

Even we do it within our own organization, things of channels to share, things like our own Gen.ai NerdOut channel and some of the learning channels. And I think there is an obligation by companies, even departmentally.

to A, do that end user enablement and as much as possible, facilitate those people who are the innovators and the early adopters to get in front of it. Then the question is, okay, how do we get them to sort of pass along those learnings to the rest of the group? There’s a lot going on out there in organizations from a standpoint of designing their incentive systems, designing sort of their evaluation of how they ⁓ sort of look at their employees to sort

or pick up AI to share it. ⁓ I think there’s a lot more to happen. This is an organizational change question about how you drive the behaviors. And those organizations that are embracing some of those behavioral modification sort of tools and techniques are going to benefit from it. But there’s still human nature in this that is a challenge to overcome. And I don’t think the org change group, if you will, has figured this out exactly yet.

But man, those groups that are figuring it out earlier, I’ve talked about this before. We have the opportunity in what we do at Answer Rocket to work with mid cap clients.

mid-sized companies and large organizations. And there’s an advantage ⁓ that I don’t think we’ve ever seen before potentially for mid-sized organizations to take a leap forward. And those that are sort of embracing as much as possible this learning curve and then how it translates from individual productivity to enterprise use cases, ⁓ that’s how this story is going to be driven. So we’ll see.

Pete (23:15)

Yeah, I would just maybe add two little things. One is, ⁓ look, it’s going to be a lot different. Your organization is going to treat this stuff much differently. If your CEO and the leadership is saying, hey, we heard about this AI thing, and we should probably do something about it, to the CEO that walks in and says, hey, guys, let me show you this app that I vibe-coded over the weekend. It’s sort of tracking our performance or something. There’s this idea if the leaders are going to have to lead.

and sort of demonstrate and model this behavior. And the other thing that just sort of hit me alone, you talked about generalists and Shanta, you talked about curious and I thought, oh, that’s who you need to be hiring. You need to be hiring the curious generalists as opposed to hiring for, oh, you have all these skills, right? Because you can learn, oh, you can learn so much so quickly. I talked to Jim, I was like, this technology reminds me of the scene in The Matrix where,

Jim (23:42)

of personal.

Pete (24:08)

know, Neo’s like, I know Kung Fu, you know? You can get up to speed on just about anything very quickly if you know how to use these tools.

Alon (24:15)

Yeah. All right. Let

me throw a counterexample at you just because to stay balanced. without naming any names, I guess.

Pete (24:28)

Are we on this call? Okay.

Jim (24:29)

But yeah, if we’re on this call…

Alon (24:31)

You guys are safe.

we’re, ⁓ we’re, we’re, building a, a product and you know, there’s this, this, ⁓ the sort of this technology barrier, guess I would say at some point when you’re vibe coding something, if you don’t really understand the code that it’s writing at all, it all seems pretty good. Like there’s no, there’s no notion of like, Hey, this makes sense. Like, and to me, like, you know, generally speaking, when I’m reading a plan and I’m like, okay, you got

Pete (24:51)

Right. Yeah, right.

Alon (25:01)

basically the general stuff, right? There is a point at which though, when stuff gets written, and in a particular case where the technique that the AI decided to use would have made things really inefficient as the data grew in scale. Not at all obvious though, it’s a small data set that you’re playing with when you build it. And if you’re not technical, it’s like, ⁓ it looks like it’s getting some records, it’s doing some manipulation. And it’s like, no, no.

Pete (25:04)

Right.

Yeah, demo time.

It seems fine. You don’t know the difference.

Alon (25:29)

If you do that, the minute you hit like, you know, a thousand, you know, records in this thing, it’s, know, like the algorithm itself is going to be spinning its wheel for like, you know, this is a screen UI. It’s going to spin this wheel for like 30 seconds. And your response that should be under a second in this scenario. there’s some, and so like the only thing was like, you know, the AI wasn’t, ⁓ it was prioritizing something in the requests around like how to do it that had nothing to do with, with a scale. Now, ⁓

And so it’s easy to inject a whole bunch of code like that without ever knowing that you’re carrying that problem. ⁓ And so this, so it’s easy. say, you know, if you’re curious, that’s a great start and you could teach yourself and you can build all kinds of interesting things. I think we’re going to have a lot of slop basically prototypes being built because that’s what they are. And so we mean that, that prototyping and saying, we’re on a trajectory to build a product where

Pete (26:02)

Right. There.

Don’t you think a lot you could…

Jim (26:21)

Bye.

Alon (26:27)

It’s important that the technical underpinning is right. And it’s still going to take some degree of, you know, of review to make sure those things are aligned. I’m sure all that will get quite a bit easier. You have your architectural, you know, agent that can review and say, Hey, these are the trade-offs you’re making and so on. So, so somewhere along the way, it’s, it’s not necessarily about the human beings having to know all this, but

Pete (26:45)

Yeah, we’re

still going to need engineers.

Jim (26:49)

Yeah, good engineering doesn’t go away.

Alon (26:49)

We’re definitely

Pete (26:52)

Well done.

Alon (26:52)

going to but the most productive engineers are going to be able to lean into ⁓ the whole stack and be inclusive of what problem am I trying to solve and understand it deeply. think non-engineers will have the challenge. Yeah, so the challenge flows both ways, right? Non-engineers have the challenge of understanding the technology better. Engineers have the challenge of understanding the domain better.

Pete (27:01)

But I think all this says…

Yeah.

Right. But the gist, I think, is the set, the time cycles for this stuff shrinks just dramatically. So ⁓ the next topic, I just want to sort of shift to the next one in the white paper, Jim, and this one sort of right up your alley. And it talks a little bit about ROI and the idea that ROI is sort of shifting from, you know, I saved some money, so cost savers to

How do I generate revenue? How do I create new products? How do I create some very significant value beyond just cost savings? How do you see that? How are your clients sort of thinking about that and talking about it?

Jim (27:53)

Well, first of all, everyone wants to create the self-funding roadmap. And that’s a conversation that we always have with clients. yeah, it’s a good sales pitch, but I think it’s extremely plausible right now. And… ⁓

Pete (28:11)

Mm-hmm. ⁓

Jim (28:13)

know, fortunately or unfortunately, cost savings often comes back to a labor question. You how can we bend the cost curve there? ⁓ You know, the sort of intelligence systems, where does this either sort of allow us to not hire additional labor, but continue to grow the business, or where does it ⁓ offer the opportunity to take some cost out? I think that, you know, to say we’re moving from one sort of mindset to the other, I don’t think is quite, I don’t quite agree with. I think it’s an and.

And

I think that’s the conversation that we get into. What’s fascinating is we have run into clients who have said, hey, ⁓ in their business, they’re seeing revenue opportunities and they don’t know how to sort of scale it. So they’re coming back and saying,

The first question now is no longer just the cost takeout. again, don’t see it as we’re moving from one to the other. I see it as an end, but I see most senior leaders now are asking the end question because they’re seeing the revenue opportunities that this may unlock.

Which is awesome, right? That’s the sort of message that gets people excited as opposed to this, know, where can we sort of take labor out of our business? So I think it’s both. And I think the opportunities are there. And I think the executives are seeing both right now.

Alon (29:36)

Yeah, if you’re, if you’re imagining a world where intelligence is cheap and abundant, are all kinds of new products and services just about anyone can offer on top of whatever they’re doing today. ⁓ And so, you know, you start with that mindset of, know, all of the universe of things that I, I do, which things benefit the most from abundant intelligence that’s online and, you know, ready to go at any time.

And that creates the ability to start thinking about new offers and new ways of doing business.

it’s probably best done by thinking of a few ⁓ examples in scenarios, but the historical pattern of ⁓ I can’t talk to every one of my customers the way I want to talk to them because each customer ideally would be treated individually. ⁓ And if I can treat each customer as an individual, knowing their exact issues, their concerns, I could improve the service that I provide them. And so if you’re in a business of a health

desk, obviously there’s, you know, becoming cost efficient at doing that. But if you’re in a business where every interaction with the customer is an opportunity to, you know, surprise and delight them with something additional and some of those things, generate revenue we have, right? That’s a great intersection to be in where suddenly the addition of intelligence to that conversation where the customer is no longer waiting on hold, the customer is not waiting for somebody to return some information to them. They’re able to very quickly get the personalized response they need.

and move through the process to get to some results that they’re looking for.

Pete (31:17)

So if we try to sort of wrap this into maybe final thoughts, you you’re in a, I don’t know, you’re in an elevator with a senior executive that just trying to describe like, here’s how to be thinking about this and here’s how to talk, sort of approach the organization with it to get real business results. I’ll go around the horn, but Alon I’ll start with you. What’s your sort of elevator pitch on, here’s how you need to be thinking about this.

Alon (31:45)

How

many floors are we traveling on?

Pete (31:47)

As many as you like. mean, it’s a big building if you need it. Yeah, you go. Sure. Yep.

Jim (31:49)

Thank

Alon (31:50)

Empire State Building size elevator?

Yeah, I’d say, you know, start by imagining some future world that you want to live in, that you know, that you want your customer, your company to be part of, you know, some future view of the world and imagine it then going backwards from from there. Right. So

Because the aspirations have to be big in this use case for it to totally be worth it, I think, for ⁓ major shifts. Now, you could do lots of smaller things. But if I’m talking to senior leadership and I’m saying, OK, where do you want to be three five years from now, let’s say, and how ambitious are you being about those goals? And then work from there to say, how does technology fit into that?

Shanti Greene (32:36)

I might start with how are you currently measuring the success of your business? What are these metrics that you care about? And what would it take to start improving them? AI is a great enabler for some of them, not a good enabler for others, but let’s move the levers we can move using the tools that we have with the understanding that it’s just going to keep getting better in certain areas.

Jim (32:58)

So I’ve said this before, ⁓ there are a lot of people who sort of view AI as sort of some magic wand right now. We’ve got to, you we forget how, we think about this all the time, we talk about it all the time. We forget sort of where some of our clients are, the potential clients that we talk to about their journey and their maturity in this topic. ⁓ I think a way to frame it for them is,

AI gets work done. It may be work that you’re already doing. That’s fine.

It may be work that you wish you could get to. Let’s think about what you would love to be doing more of that AI may be able to help you scale. Or it may allow you to introduce new products or services to your customers that you simply are not able to today, again, because you can’t scale in a way to do that work. Let’s talk about those domains and present it in terms of the business. What work do you want to get done more of?

What work would you like to be able to do to deliver a new product or capability to your customers? Let’s start from that, the dialogue. And that’s going to, some things are going to fall out of that very quickly that will deliver ROI to you. And let’s not get hung up on sort of all of the technology of it or what models what or sort of what’s going on right now. But let’s think about it from a work getting done standpoint.

Pete (34:26)

All right. And so since I am hosting, I get the last word. So here you go. Here’s my, here’s my take. I wrote down one, one set the example. Like your team is not going to embrace this stuff unless you’re embracing it. And I mean, literally you should be digging into, know, starting to use these tools and really sort of take them to places where you’re not quite even sure you could go. I say two, prepare to.

Alon (34:32)

Peat steak.

Pete (34:50)

really reorganize your company around it. If you let these things sort of continue to take place in their silos, you’re not going to get the benefits ⁓ that are really possible. ⁓ And I guess the final thing I would say is you need to be thinking well beyond what you probably currently believe is even possible.