Where the Real Money in AI Actually Is

Once your team knows how to use AI, what do you point it at? That’s the question at the heart of this conversation.

In this special episode of the Growth Elevated podcast, Julian Castelli is joined by Rick Lindquist to walk through how to turn AI from scattered individual wins into company-sponsored projects that drive real business value.

The real leverage lives in core, cross-functional processes across revenue acceleration, cost reduction, and capacity creation. Julian explains why companies that skip straight to big projects often end up with isolated pilots that burn money without a clear return.

He lays out a practical four-step framework. Run a top-down value scan with your leadership team, repeat it bottoms-up by department, identify the winners on an impact-versus-ease matrix, and decide whether to build, buy, or contract.

He also covers the discipline most leaders overlook, owning your data and governance when working with contractors, and reporting every AI initiative to the board as a tested hypothesis rather than a vague promise.

For CEOs and operators, the message is clear. The value is real, but it comes from structure, ownership, and honest measurement, not from chasing headlines.

Key Takeaways

  • The real leverage is in core processes, not individual productivity tasks
  • Score AI initiatives on impact versus ease, then commit to two or three
  • Decide deliberately between building, buying, or contracting each project
  • Keep ownership of your data, documentation, and governance, never the vendor
  •  Report AI bets to your board using a clear hypothesis and the scientific method

Visit us at: https://growthelevated.com/

Follow us on LinkedIn for daily leadership inspiration and updates: https://www.linkedin.com/company/growth-elevated/

Transcript

Rick (00:04.062)
Hi, I’m Rick Lindquist, and I’m hosting the Growth Elevated Leadership Podcast, where we talk with inspirational entrepreneurs and leaders in the tech industry. This episode is brought to you by Growth Elevated, a community of tech founders, CEOs, and CXOs who are committed to working together to share best practices and learnings in an effort to help all of us become better leaders. We do that through educational programs like this podcast, as well as our blog, and of course our annual Ski and Tech Summit.

Julian Castelli (00:05.673)
and I’m hosting the Verde Elevated Learn Podcast where we talk about information on monitoring the tech industry. This episode is brought to you by Bird Elevated, a community of tech founders, CJ Hosted, CXOs, who are committed to working together to share best practices and learnings and an effort to help people of us become better leaders. We do that through an educational program like this podcast, as well as our blog and of course our tool and tech content, the be brings tech figures to critical custody Utah.

Rick (00:31.944)
Where we bring tech leaders to beautiful Park City, Utah to enjoy camaraderie and collaboration in the beautiful mountains. So if you enjoy skiing and the outdoors and networking with other tech leaders, check us out at growthelevated.com and please subscribe to this podcast wherever you listen to podcasts. My guest today is Julian Costelli from Growth Scaling Partners. Julian works with multiple tech and growth companies as a board chair, board director, and operating partner.

Julian Castelli (00:35.413)
Enjoying collaboration in beautiful mountains. So if enjoyed speaking in the outdoors and networking with other technology, check this out at growthalvated.com and please subscribe to this podcast wherever you listen to podcasts. My guest today is Emily Castelli from Growth Scanning Partners. Julian works at multiple tech companies, the four-tier board director, and operating partners. Today we’re going to talk to Julian about a state, a three-stage playbook. He has started appeared on how to drive.

Rick (00:58.546)
Today we’re gonna talk to Julian about a sta a three stage playbook he has started shared on how to drive value in the enterprise with AI. today we’re gonna cover stage two. Julian, welcome to the podcast.

Julian Castelli (01:05.133)
Hey, thanks, Rick. Excited to be here.

Rick (01:13.754)
Awesome. So if you’re if you’re a joining ed kind of mid series context for you is that this is part two of a three part series on growth scaling’s AI transformation playbook. it’s a three stage framework that Julian has designed to help growth companies transform from pre AI operating methods to becoming AI native. is there anything you’d like to add, Julian, to that from a context perspective?

Julian Castelli (01:15.186)
If you’re if you’re putting it in connected to your context where you go.

Julian Castelli (01:38.361)
Yeah, there are three three stages. Today we’re gonna go through stage two. Stage one is really just getting your employees up to speed. So it’s enabling your knowledge workers with the tools of AI and and basic frameworks. And we we had a lot of

success when we presented that earlier this summer. stage two, what we’re gonna talk about today, is starting to tackle projects within the company. So company sponsored projects and how to how to choose them, how to, how to execute them and how to track them. And and you know, there’s a lot of lot of stuff in the news about that today. So I’m sure we’ll have fun talking about that. And then stage three, which we’re gonna be giving today at a at a workshop this afternoon.

is if you is how do you put it into your product and what are future products going to look like, right? And and the the the the genesis of this is the news is filled with these AI native companies that are kind of racing to these incredible valuations and it just seems like kind of a mystery as as to how they’re doing this. And it’s it it it kind of is because if you’re a a legacy company that was built with legacy systems, people, processes

And then you see some company operating with five people at at 20 million dollars, you’re like, how does that work? it’s really hard to get there when you when you when you’ve architected around the traditional business architecture. And so this is a a program to to to to focus on, you know, if you’re an existing company, you know, how do you start start and then move down the journey to becoming as close to AI native as possible? And how do you compete?

Rick (03:15.804)
Awesome, awesome. And so what what what is the part that we’re gonna be walking through today?

Julian Castelli (03:21.393)
Today is process transformation. So this is after stage one where you’ve you’ve you’ve helped your team your team start to become educated and fluent in AI today is like now you’re starting to sponsor company projects to automate core processes in your business and make them more more efficient, faster, more effective, and and reducing costs.

Rick (03:47.711)
Yeah, definitely recommend watch listening to our previous ep episode on this, the first stage. I can’t get the riding the bicycle analogy out of my head. I keep using it at at work in my in my at windfall and

Julian Castelli (04:00.27)
I’m glad to hear that.

Rick (04:01.415)
Yeah. And so I I we’re still very much in the bicycle, learning to ride the bicycle stage. and I think the point of that stage is you’ve got to get some level of shared understanding of the capabilities and the know how on how to leverage AI from a tooling perspective before you can really drive process operating tran operation transformation. Is that is that accurate?

Julian Castelli (04:24.892)
I it it it one hundred percent is. I I think the the the the the core message behind there is bringing everyone up to speed is gonna unlock more leaders, more contributors, more ideas, more fluency, more capabilities. And it’s no different the way you and I are learning, right? I know you and I have had several conversations where we’ve just we’ve compared notes on on on our personal AI journeys and and every one of those

Rick (04:45.117)
Mm-hmm.

Julian Castelli (04:51.492)
things where like you share something you’ve done or I share something I’ve done, I’m learning a little bit more. And it’s about the responsibility that leaders have if they want an A and I AI enabled work t workforce, they have to set the conditions to make that possible. Right. And that doesn’t naturally happen. In fact it’s probably not happening more than than happening. Even though everyone wants to do AI, it’s not happening. It’s not happening because people are busy. And that’s the that’s the bicycle thing. So

Rick (05:06.61)
Mm-hmm.

Julian Castelli (05:18.0)
The the the idea that you you know that you just referenced is if there’s a man running in the desert with a bicycle and he’s he’s tired and he’s stressed and he’s sweating and and the the stranger asks you know on the side of the road, says, My friend, why are you running? And he go, you know, why don’t you just ride your bicycle? And he goes, I’m already behind. I I don’t have time to learn. And and that’s that’s that’s what AI represents to some of the CEOs I work with, where they know they have this bicycle that should be able to transform their business.

But they still have to raise their funding round. They still have to close their their their sales quota. They still have to operate and fulfill to meet their their their operational responsibilities. And those things don’t go away. Now, the difference between running with a bicycle on your back and riding a bicycle is huge, but you can’t see that when you’re running with the bicycle on your back, right? So so that’s that is that’s why it’s such a great analogy.

Rick (06:03.742)
Yeah, exactly. Exactly.

Awesome. Awesome. We’ll take us into it.

Julian Castelli (06:13.101)
Terrific. Okay, so I’ve got a presentation here, which I’ll share. so if if if you’re going along this process stage by stage, and you’re coming out of stage one where you’ve invested in your people, you’ve put a charter together, you’ve enabled people with designated tools, and most importantly, you’re doing these working groups, so everybody is is sharing AI and learning at an accelerated pace, you might start hearing some different questions instead of like,

I don’t have time to do AI. You might you might start hearing like, hey, couldn’t we automate this workflow? And you know what? What if AI handled this? And that’s what you want. That’s the value of the organic bubbling up of ideas. That that’s a signal that you might be heading towards stage two. for example, you know, in sales, you might say, Why aren’t we automatically updating Salesforce? This is one of the biggest pains of every Salesforce, every salesperson is, you know, they of course they want to be productive, but they want to be talking to salespeople, they want to be helping customers, they don’t want to be

Going through fields on a spreadsheet and translating notes on an airplane into the CRM. But if they’ve gone through stage one, they might say, couldn’t this be done automatically? And the beautiful thing is it can be, right? Same thing in finance. Like, why are we pulling data from three systems manually? Right? Well, the the person who’s who’s running with the bicycle just keeps doing that and and and you know, and the bicep you know, but you know, automating that is one more chore that they don’t have time to do, right?

riding a bicycle is saying, okay, why don’t we just pull these things and have AI do it and have them to do the first draft and we just edit the first draft and and and and and and really we focus on the MDNA because that’s the human element, right? so if you start hearing those things, those are signals that stage one was working. And just to go through again.

Rick (07:55.863)
J just to to summarize that point, I think it’s an important point. Don’t you don’t want to exit stage one before you’re ready to for stage two. And the signal for stage one is there are multi step processes that are being identified as sort of the old way that that that could be transformed by AI, but it requires maybe more than one person or multiple teams to collaborate.

Julian Castelli (08:19.263)
100%. It’s probably cross-functional and it probably gonna cost some money or or or commitment of resources, right? In phase one, we’re talking about the individual productivity things, right? It’s summarized this document, all the things that we’ve all been excited to learn by using Chat GPT, and now that you know I’m starting to see my household use it. I know that the you know the world is is is is adopting it like the way the world adopts internet.

Rick (08:23.443)
Mm.

Julian Castelli (08:43.837)
Those things are easy because it’s the in it’s the knowledge worker doing their everyday job faster, right? But now you’re pointing it to a process. We close the books every month, every month in the f and the finance department. Boy, there’s a lot of manual processes. Every time I go to a sales call, I love talking to customers. I can’t stand the idea of updating Salesforce, right? And you’re starting to say, wait a minute, some of these processes, if you do the math, there’s a hundred people in the Salesforce, every and this everyone’s having to update HubSpot or whatever it is.

How many hours of work that is, could could we automate that? Now that’s not someone that’s one person’s gonna just tinker and just do on a rogue rogue way that because it’s a company system, right? You’re touching company data that it it requires sponsorship by the company to, you know, for permission, for security, for compliance, for cost. There’s all sorts of things, but but that’s when you really start unlocking great value.

Rick (09:37.864)
Makes sense.

Julian Castelli (09:39.94)
Okay, so so this is stage two. So stage one, knowledge worker enablement, AI process transformation, that’s where we are. And then stage three, which we’ll talk about in a few weeks, is putting the AI into the product that you’re offering your customers, which is the the whole theme of this this kind of race and using the bicycle is it is a race. You and I were talking about earlier that these AI native companies are just mind boggingly boggingly fast and out front. They are. And if if you’re a company trying to compete with them,

You better get on a bicycle and and and and get in world class shape. So it it is a race. Hell however, I do think there’s value in going in this this order for all the reasons we’re about to discuss.

Okay, so stage one, just to quickly review the the steps were you you designate an AI ch a champion, you publish an AI charter, you provision officially sanctioned company tools. Those are kind of standard. The the thing that’s a little unique here is you build working groups where everyone’s doing effectively show and tell for either two hours a week or one hour a week, depending on the size of your company and and and your appetite.

This the key here is if the CEO doesn’t lead this, it’s not gonna work. It has to be delivered by the CEO. He has to be working in in a in a in the working groups and he has to be contributing to the the the the innovation list. Okay. Then you launch the working groups and then you have a quarterly showcase where you celebrate wins. It sounds simple, but it’s it’s very hard to get get companies to do this because it means stopping, putting the bicycle down, and having everyone stop and start riding the bicycle. It’s difficult. But

I I I do think it’s worth it and the companies I see that are leaning into it, I’m seeing faster transformation.

Rick (11:23.134)
Yeah, the I mean if if you can if you can do this, even if it’s I mean what we’ve I we have not I’ve been not been able to get my organizations to commit like you’re you’re talking about here. but the if you just dedicate one hour a week to start, like you start to see the benefits. Like it is a huge like two weeks later, people are are automating their job.

Julian Castelli (11:40.535)
A hundred percent.

Yeah. Right. And it’s it that’s the beauty. When you start to see it happening organically, I I’m a board board chair and I work with you know, one company and and the finance reports used to always come out a day or two late. They always were a little bit lacking. And all of a sudden I got this report like, you know, three days early, and it was so crisp and so published, and I’m like, What is going on here? And I started reading it and then I could I I started, you know, I could see some telltale sides, right? A AI leaves some exhaust, right?

And I could tell this this was probably done by AI. And I said, you know what? I’m really glad it was done with AI because it’s it’s it’s it’s it it was delivered faster. I know that the team member work, you know, was able to devote resources to something else. And frankly, it was better. Even though it had a little bit of, you know, of exhaust in terms of AI, it had some other it’s not this, but this, you know, you could tell.

Rick (12:33.232)
Mm-hmm. Mm-hmm. Yep.

Julian Castelli (12:35.512)
But I I was like, this is great. This is what I’ve been pushing the team to do and and good for them. Now they there’s they just bought themselves three extra days in the close cycle. How how valuable is that?

Rick (12:44.382)
Huge.

Julian Castelli (12:45.74)
Yeah. Okay. So we talked about the stages. Let’s let’s get on to stage two, okay? you know the the reason you need sta there’s differences is individual wins are great, and that means everyone’s working better in their in in their you know knowledge worker environment, but but the real the real leverage comes from big processes. I had I had Kathy Slavinski, who runs an AI first company on on on the podcast earlier.

And she said there’s a pyramid of suck. And at the bottom of the pyramid pyramid are the things that you really hate to do, but they they’re but they’re big and wide because there’s a lot of them, right? And and I just gonna use updating HubSpot or that your CRM as an example, right? How many I mean, every single salesperson I know can’t stand that. And I I can barely do it for my business. It’s just like I I don’t want to do it, you know, but but you know, I I don’t mind talking to someone and saying, hey, go go go go update for me.

And that used to like, okay, get me a sales assistant or someone in RevOps to do that. Now, if you can have AI do that, think about the productivity, right? So it’s productivity for revenue acceleration, cost reduction if you do have assistants doing it, and capacity creation. That’s the most important thing. A salesperson’s, you know, time on task with customers is directly proportionate to the productivity they can have in terms of the what matters, which is closing sales. Time on task updating the the

The CRM doesn’t correlate that, but but having a good CRM and a process does. So it’s kind of a catch 22, right? But this is where the real money is, is what I’m trying to say. And I think that what I’ve seen and the people I’ve talked to is that every function, every process has opportunities. you know, some of the earlier ones I’ve been in customer support, and and and engineering with a with the coding, but I think every, you know, we could go through and find

examples in every one of these areas where processes that are enabling best practices can be enabled with AI. And that’s a key thing. Sometimes you have you don’t have best practice because the processes are too difficult or complex, right? And so now everyone can have best practices because you theoretically can can do, you know, if it’s a 10 step process for sales, if that’s best practice, if you don’t have the resources, you might just get away with three steps.

Julian Castelli (15:08.978)
And you know, you’re probably getting away with three tenths the benefit, right? Now you can do the steps, but you still have to, you know, you still, you know, what you don’t want to do is you don’t want to automate the three step process that you’ve been doing because you don’t have resources. You want to actually figure out what the best 10 step process and automate that. So it’s not easy, but it’s possible. Does that make sense?

Rick (15:26.974)
Totally. And if you go I if you go back to the last slide, I just wanna if if you’re listening and not able to see the slides, I think this is a really powerful slide. it’s it says stage two is where the real money is. and it basically breaks down three key outcomes to measure what stage two can do for you. one is making you more money, revenue acceleration, another is cost reduction, saving ti saving money. and then the third is like efficiency gains. And I I find it very interesting to see these three buckets because

Julian Castelli (15:41.639)
What stage two can do for you. One is taking anything, but acceleration, another cost reduction, saving time saving money. the third is like efficiency gain. I find it very interesting to T D three because you think about the SAS wave the the three values the three common value propositions of SAS as what these three buckets are. So basically we’re one hundred percent.

Rick (15:55.295)
If you think about the SaaS wave and the the three value the three common value propositions of SAS, that is what these three buckets are. So basically we’re we’re talking about we’re talking about building our own SAS custom to our org or our own AI enabled service that that that basically sells us more money, more time or b or or more efficiency.

Julian Castelli (16:11.582)
Yeah I think.

Julian Castelli (16:19.196)
I love that point, Rick. That that’s exactly what software has done for the enterprise. And they they talked about how software’s gonna eat the the world and and and automate every function. And let’s f let’s face it, it it has. There’s probably three software solutions for every major function, but now the software is capable of doing more. Right as whereas and that’s gonna be the the the stage three discussion. Whereas SAS used to do I have a s a seven seven stage framework for automation.

Rick (16:42.206)
Mm.

Julian Castelli (16:48.805)
And s SaaS is right now does typically to level two or level three. And AI is gonna allow it to be level to level six or seven autonomy. And that’s why a lot of the conversation about the SaaS pocalypse are out there because like, my god, instead of competing for your technology budget, it actually now can do the job of the entire department, right? And that that has excitement for SaaS companies, it has you know fear for employment, all the

Headline. Headlines that that unleashes. I think it’s real, but I think there’s you know exaggerations on both sides. But the the point is the AI is gonna allow those companies to do much more of the job to be done than than they historically have been doing. And therefore they’re gonna deliver more of these three key value propositions. And you know, as someone who sells a lot of SaaS, someone who evaluates funding SaaS, you always ask that question.

You know, what’s the payback or the ROI for the customer? You know, and and and to your point, they they they measure that I’m gonna get this much revenue, so I’m gonna spend a hundred thousand dollars on the software, but I’m gonna get five hundred thousand dollars of of gross margin off of revenue, that’s a five to one payback. Or I’m gonna be able to reduce cost by five hundred you know, five hundred thousand dollars, therefore this is a five to one payback. I’m happy to do it. Or I’m gonna create more capacity to create value. And and so that payback math is you know how.

Customers evaluate investment decisions and how how innovators have to sell their product.

Rick (18:23.378)
Makes sense. Thanks for going back there.

Julian Castelli (18:26.372)
Yeah, thank you for for calling it out. So anyway, the next point, I think this exists in every department, but the question is how the heck do you get it? Right? And and and we talked about earlier, you know, do you go back to stage one or you just jump in here? A lot of people I see just jump in here. And I’ve been guilty of myself because I could just see I can just see the customer support opportunity, right? And I’m like, just go do that. Right. And I’ve had some companies that have done it successfully and I’ve had some companies that have struggled because

They just dive in and and do some of this without going back and and doing the the the the stage one and they end up having these kind of isolated pilots. and I I I think I’ve read in the news they talk about these pilots that are kind of just hanging chads. And and you know, the the only thing you’re for sure is that you’ve spent money, you don’t know really what you get you get from it.

Rick (19:12.254)
I got a I’ve got a perfect example of this. we’ve we I are you familiar with like an out outreach, which is a sales cadencing tool. It’s like sales loft or gong engage. they they sold us really well and on our as part of our renewal a cheap pilot of their AI, you know,

Julian Castelli (19:21.039)
Yes, yes.

Julian Castelli (19:28.728)
Yeah. Like sort of showing all the prices context you have.

Rick (19:30.156)
tooling, which you know, we’ll we’ll give you professional service, it’ll be a piece of cake. We get on that first call with the professional services person, we’re like, so show me all the organizational context you have around your ICP, your key messaging, and how we’re gonna point the AI. And we were just like, it lives in our heads. and and so, you know, and and

Julian Castelli (19:43.573)
Yeah.

Classic, yeah. That’s exactly what we talked about that at that web at sub seminar last summer, right? It’s like it’s like just documenting the stuff. It’s back, it’s it’s it’s it’s the ETL monkey thing. Like you have to get the data first. You have to have good structured data. You need structure if you’re gonna automate a process, you need a good process.

Rick (19:53.902)
Exa exactly. And so

Rick (20:07.46)
Exactly. And so literally we had to go, can we please pause and come back to this? So anyway, we the the the success story here is we got the people who needed to understand this stuff more access to AI tooling, got their you know, we got their their buy-in on why context is important. because once you start using AI and you ride in the bicycle, you realize how important context is. You don’t appreciate it until you understand. Correct. and so

Julian Castelli (20:30.163)
A hundred percent. ‘Cause it it it it’s only as smart as the context you give it. And and what was the what was the quote I heard this just the other day, it’s like AI is a you know is a hundred percent s you know, faster, but it could be a hundred percent faster good or a hundred front faster bad, depending on what you give it.

Rick (20:46.174)
Ex exactly. So the the success story here is because of that, we kind of we had a hanging Chad, as you as you mentioned, and then we went back to the team and said, Hey, this is why. And then we it took a long time to get it, but we went we basically came back and got a pilot and we’re booking like ten X to twenty X the meetings per rep on outbound using that pilot now. Yep.

Julian Castelli (20:58.723)
Long time to get that. But we won’t be basically We’re looking at the rat on around it. That’s awesome. And that’s and that’s the goal of that software that you’re talking about there. So that that’s that’s fantastic. I mean, look, I don’t buy the Doomerism when they I I it it’s like two things can be true. A lot of people can be stumbling and not getting ROI off their first projects. That could be true. And

There, you know, it it could be a really, really good idea to do those projects because there’s 10x right behind the corner. And that could be true too. But it’s just it’s just it’s it’s bridging the gap between the two. Yeah. that’s a great example. Thank you. All right. So then how do you do this, right? So th this isn’t about how to use an AI tool or how AI works. This is organizational design. How do you actually, if you’re running a company, you know, practically, you know, with this as simple as possible. I’m always about the KISS KISS principle. How do we do this? So here’s four steps, right?

Rick (21:33.501)
Mm-hmm.

Exactly.

Julian Castelli (21:56.798)
We’re gonna you do a top-down value scan, which we’ll go through. Then you do a bottoms-up value audit by department. So you kind of you you mimic the process and do it in every one of those departments we just talked about. Okay, then you identify winners, and then you evaluate are we going to buy, build, or contract? Okay. And then you go and you re you you execute and you report using the scientific method. And it’s as simple as that.

and I’m gonna quickly close this window because like it’s getting loud just a minute.

Rick (22:34.558)
Yeah, and if if the the slide here basically is a four-step framework on you know basically how you get started here. I I think we’re gonna go through each of these, right, Julian? All good.

Julian Castelli (22:45.833)
Yes, sorry about that. Had a lawnmower starting out in my my my neighbor’s house.

Rick (22:50.45)
That’s that’s a great time to cut grass, is right in the morning.

Julian Castelli (22:54.217)
Yep. But yeah, we’re gonna go through each each of these four steps. Okay. So first thing, and this is the you know, this is the message for the CEO. You gotta you have to model it by doing it yourself. And and and the fastest thing you can do is get your executive team together. Could be a two hour session, it could be a half day, but it’s your ELT, it’s your stage one AI champion, it’s homework beforehand, and you’re you’re gonna just evaluate your top ten. Get your top ten.

I I recommend that you know you use a framework that is is pretty simple. It’s a two by two met matrix. What’s the impact potential and what’s the ease of implementation, right? we’ve all seen these, call it the BCG box, I think. But the quick wins, this has high potential and it’s easy to implement, right? Those are gonna be great. there’s gonna be strategic bets that are high and hard, and they’re gonna be nice to have that are low and easy.

And then there’s going to be the hard ones that are low impact that you just defer for later. But ultimately, this is the rubric which we which you should be scoring all the innovation so that coming up from stage one. And now you’re you’re you’re you know you can use that list if you have stage one, come up with your top 10 and score them. And then the second element of this is once you’ve scored them, it’s you it’s easy to look at the quick wins, but then you have to do it. I recommend a data assessment.

Right. And we talked about it earlier, making sure you have the right data. You pointed out making sure you have a good process, you know, all the things that you need to do. But take a look at it. And I’d put in one more screen, like the the okay, the data for this exercise is is is perfect, or we need to do a little bit of work, or hey, maybe it’s red and we have to put this lower on the list because it’s a great idea, but we we need to work on our data hygiene before we can really utilize the tool.

And so going through this, having a session or two to to identify the top 10, and then you know, really focusing on the top one, two, or three, I’d say three max, and say, okay, these are gonna be corporate priorities, that’s really important in as a leadership team. And that’s where the pilots come from. And and and and then those we’ll talk about how to exer execute those pilots.

Rick (25:09.64)
So ha so one challenge I I I expect a lot of leadership teams to have is like where do you do this? Is this in the weekly e staff meeting? Is this an off site? Is this a w is this a one time thing or a recurring cadence? Like do you have any thoughts there?

Julian Castelli (25:21.978)
I do. I th I think it should be dedicated, specific meeting. Th this is a strategic priority, kinda like a strat planning meeting. I think it should be clearly communicated with the AI champion and the CEO, both owning it. And I think everyone sh should be given homework with the framework and you know, come with your ideas. We’re gonna debate them. We’re gonna you know, we’re gonna disagree and then we’re gonna agree on the one or the two projects or the three projects that we’re gonna do. And we’re gonna, you know, we’re gonna

publicly share it with a company so we can all learn together. We’re gonna put our money behind it. We’re gonna share it with the board. These are gonna be company sponsored initiatives, just like the one that you just described, you know, with sales outreach. You know what you you you clearly the potential must have been really big and, you know, the ease, you know, may have been a little harder. You kind of learn that go by doing it, but it was it was clearly an important bet that you guys have made and and you you know you’re learning from it, right?

And that’s the key that the everyone gets in together and we’re gonna learn and and you’re not gonna automate everything and not everything’s gonna work. Right. And I think that’s where the headlines come up. They they come up with all the examples that don’t work. But the i you know, it the the the ability to make these things work is also hard to argue with. It’s it it i i it it’s obviously possible, it just is not easy.

Rick (26:41.062)
How so so typically, you know, if you’re starting a process like this, it’s probably happening mid mid plan year. You’ve got a strategic plan in place.

Julian Castelli (26:46.852)
Yeah, yeah. This is meant to be able to be done as a as a SWAT. Go do it. You know, I think, yeah, that we could do a separate episode on how to how to work this into annual planning. And they and they probably should be some priorities that, you know, start coming up in Q three and Q four when you’re doing your strat planning for the next year. But I recommend that everyone do this, you know, as soon as you’re kind of you’ve you’ve enabled your your first team. They this is a sequence you can layer over any time in the the annual pl annual cycle, which is like enable your people.

Rick (26:57.661)
Mm-hmm.

Rick (27:04.062)
Mm.

Julian Castelli (27:16.713)
Once you’ve got that, you’re gonna have people wanting to try and do these projects themselves. And my recommendation is, you know, let the individual experiment and go, you know, start and organically work on individual productivity tools. But when you’re touching something as expensive as your Salesforce and your sales process or your your financial you know to integrity, they’re probably need to be company sponsored. They’re probably gonna cost some money, they’re gonna probably have some requirements in terms of security and and guardrails.

Therefore it’s something that companies should sponsor and and and should should make it a priority and that should be worked into the, you know, your current goals this year or, you know, some should be deferred to next year based on your planning cycles.

Rick (27:57.791)
Yeah, I like that you said that. It’s kinda like the the the one prioritization lens here, in addition to the ease versus impact, is how aligned is it to what we wanted what we said we were gonna do this year?

Julian Castelli (28:09.1)
Yeah, yeah, I I that’s a great great ad, right? Okay, so so the the the purpose of the first meeting is to get a consensus, get everyone’s opinion on on on the the the areas of best opportunity. The second per purpose is to teach every one of your leaders this process and then send them to their departments and have them do the exact same thing. So if if as an ELT you get 10 and you’ve got five department leaders.

Rick (28:11.805)
Mm-hmm.

Julian Castelli (28:38.637)
Then you’re gonna have 10 plus 50, you’re gonna have 60 ideas. Of course, there’ll be some overlap. Maybe you have 30 at the end of the day. But having 30 value add added ideas that are green or blue is pretty good, right? So that that’s the goal. And then you wanna screw you wanna apply the filters so that you can sort and rank them. And and and you know, then look, if if we if we do nothing other than that, a if a if a CEO goes from a man, I I don’t know about riding a bicycle and I I feel like I’m behind an AI to like

Rick (28:50.398)
Hundred percent.

Julian Castelli (29:08.245)
Okay, I’ve got I’ve got a a an educated workforce that now has has stack rank twenty I value creation ideas. That’s a that’s pretty good, right? Of j just getting there. I I think most people would say I, you know, feel a lot more comfortable of where they are in the FOMO AI race if they have that than where they most people are today. Wouldn’t you agree?

Rick (29:27.896)
Mm. A hundred percent. Yeah, it’s great.

Julian Castelli (29:31.584)
So that that’s what it gives you. It gives you a list. It also gets it also educates your entire company on the ETL framework, the data framework, how to think about ROI. These are all business disciplines that are valuable for your entire team and their teams to understand. Right. And one of those examples is, you know, there’s a lot of best practices. It’s not just how many hours does this save. You have to think about it in terms of opportunity unlocked, revenue potential.

real business impact, you know, the hours saved are the obvious things that get the headlines, but quite often it’s the productivity and the ability to to be more effective, like what you described in that that sales process. Yeah, you probably save some money and some time, but the value is being able to have more at bats because you have a close rate that you can you can measure it against, right? I I’m sure that dwarfs the cost savings, right? Yeah. So that’s an important thing. But guess what?

Rick (30:19.454)
Exactly. absolutely.

Julian Castelli (30:25.567)
That’s not something that’s obvious to everybody. This is a lesson for everybody as you learn the process. Okay, so once you’ve done that, let’s assume we’ve got two great ideas just for purposes of this discussion. And then you say, okay, well, we’re going to prioritize this initiative. We’re going to put it on our roadmap as a strategic priority. You have to decide: is this something you’re going to build internally? Is it something you’re going to buy off the shelf? Or is it something you want to contract with someone to custom build? And

This is a a a a key step. you know, this bait is based on you know, do you have resources internally that can build it? You know, what I will say is, you know, not all companies do, in fact, many companies don’t, but but so some software companies do. I I’d say s doing stage one dramatically increases the chances that you might have some people because you’ve identified those stars that are AI forward that might be able to build some things pretty some pretty cool things internally.

If nothing else, you’re gonna have more AI fluency that allows you to make this decision and evaluate vendors, right? The buy situation really becomes down to, you know, where what type of function is it? Right. So you described something in the sales outreach process. That’s a a huge value added process, a hugely complex process where there are a lot of vendors, you know, innovating in that space. And so it doesn’t surprise me that you found

you know, some good off the shelf solutions. And you know, the value of an offself solution is that it’s it should be vetted at scale for many use cases. you’ve got a whole team thinking about that product domain and and and working on updating it. There’s a lot of value of that, having that. And and there’s you know probably going to be multiple vendors that you can compare. So if even if you’re considering building it, evaluating what the market is offering and what the price is is a valuable part of this exercise. And you can decide which whichever way you want to go.

The interesting thing that I’ve seen more and more, and this is what’s new, Rick. The idea that you might just contract and get and build custom off the shelf. I don’t know that, you know, custom custom software kind of was, you know, in my experience was back before SaaS, you know, because SAS became so fast and quick and and efficient. You know, that then that it kind of eliminated the need to, you know, do custom software unless you’re really trying to do something that was really unique. But now

Julian Castelli (32:43.26)
The I I saw a a demo yesterday of some custom software that was in the green category because it was a cu was it was a it was a it was a PE back company and the the leader of the PE innovation group just built this thing over like two months. I’m like, this is enterprise software. You you built this? And it was custom for the company and it did so many different things and it it united so many different workflows and I was just blown away. It was

It’s a it’s it’s a whole new world of what you could build if you have resources or what you can contract for for relatively inexpensive for the you maybe for one and a half, two X the cost of an A of an ARR subscription, you can build a cut a a a a custom piece of software. There’s there’s all kinds of caveats that we’ll go talk about, but you do have three choices now.

Rick (33:29.702)
Yeah, I I totally agree with these three choices. and you know, I just want to recap real quick of where we are. So stage one is leadership top down sort of prioritization exercise, then a bottoms up sort of like what’s what let’s rank these and get into the details. We pick the we we pick the the one or or few that we wanna push on and then we need to decide how. And and that’s what you’re talking about right now, which is build, buy or contract. I I

Julian Castelli (33:50.769)
That’s right.

Rick (33:54.865)
What I found, I don’t know if this is true, but with the with the Meteor items, you might do all three. you might build you might have to build the context internally. you might buy a tool that you want to implement and you might bring a third party in to help with the implementation.

Julian Castelli (33:55.428)
I found that I don’t know if this is true of the with the with the easier item I do all three. I might build yeah and develop the context my highest and then migrate a third party all of the implementation. Yeah, that that that’s a great point, right? Because when you get into it, you realize just like your your example perfectly you may not have the process perfectly created. You may not have the data.

arranged and and and pipeline the way you need to. You may not have the the pro you may have to change the processes by which people do work. And this is what’s so hard is to get back to the bicycle. You know, you know what software doesn’t change people’s work behavior. That’s that’s change management, right? So yeah, you you may may have to use a combination of all three, but I I just want to make it clear that you know you have these options and the discipline on how to use each three is probably a muscle that you want to create because over the next 10 years

Rick (34:39.24)
Totally.

Julian Castelli (34:51.705)
You’re gonna you know, the companies that succeed are gonna be doing all three of these things and figuring out what works and what they like.

Rick (34:57.342)
And the beautiful thing about contracting with this type of stuff is you really can bring someone on and have them automate themselves out of a job and not have a HR issue. it’s a beautiful thing.

Julian Castelli (35:08.086)
I you know, I I agree. There’s the and there’s a lot of these companies coming up, you know, coming up now. In fact, you know, at at our event we had Kyle Richless from Zygo AI and and he gave that presentation. And and you know, the the the jaws were on the floor because it just you know the value of that is he’s working, he’s probably doing fifty different projects for different companies. He’s like, Hey, you want you want you here are three things we’re doing for sales companies, here’s three things we’re doing for recruiting, here’s things we’re doing for for finance. And everyone’s like, my god, that makes so much sense. We should do that, right?

But then you get you run into okay, and you then you go go in and and you’re you start carrying the bicycle again because it’s like this idea you had at a workshop. And then you know, work work intercedes, the real life intercedes. But it’s so valuable to see that. And we you know, we put his his his his his presentation on the Growth Elevated website so people can take a look at that. and and obviously we can connect people with with Kyle and other resources that we we we have as partners at Growth Elevated. But it’s an exciting time to have access to those kind of exp resources. It’s exciting, exciting time to be

Rick (35:37.374)
Totally.

Julian Castelli (36:05.855)
in one of those companies helping companies innovate because we’ve just gotten this this golden key that unlocks all this new opportunity.

Rick (36:12.958)
Totally.

Julian Castelli (36:14.613)
Okay, so so this is the bybuilder contract. a couple things for contracting. I think we’re gonna there’ll be a next wave of of newspaper articles, you know, alarmist saying, my god, this company, you know, contracted for this and had this disaster. So let’s talk about some of the best practices if you’re out contracting, right? I it’s not a good idea to be completely dependent on a contractor. you know, look, most of these companies didn’t exist two, three years ago. They’re startups themselves.

And they’re innovating that everyone’s innovating. No one tells you like they’re rock solid, like you know, there’s no you know, McKinsey or or or price waterhouse of of of this industry. It just it does hasn’t been around long enough, right? And even if there are those those companies you can’t count on for this stuff anyway, you have to take ownership. This is the whole point of all AI is AI gets you 70 to 80 percent of there. You as the owner have to take ownership, whether it’s you’re using a tool for individual productivity.

Or you’re building a tool for enterprise efficiency, you have to take ownership. So if you’re not a software development company, you may not have some of those core things in mind around security and sustainability and data integrity and all those different things. So tr make sure someone on your team, and this is why stage one’s helpful. If you have if you’ve done stage one, you have some some budding leaders and and some high poke folks out there that you say, okay, you’re gonna be in charge of this project after the consulting firm leaves. Therefore, you better learn.

Where the data is stored. You better learn how to ch stop it and turn it off. You better learn how to stop it and do do some audit tracing to figure out why it did XYZ. Where are the vulnerabilities, right? So you’ve got to build that harness of governance in your team, not through a vendor. Otherwise, you’ll have what’s it called a recurring software, it won’t be any cheaper than SAS, right? you gotta make sure you have full documentation so that you know if if whoever’s doing it leaves, you can you can have sustainability. I I recommend that you own the data.

you know, in unless there’s an advantage of getting data from some sort of consortium. I think you know, your data is a key asset, and it will become more obvious the more AI you do, how much data is an asset. define success metrics with the vendor before signing. It shouldn’t just be one of those, let’s just start working together and see what we’d be build. And then, you know, always be thinking about how to replace the vendor because you know, look, there are some vendors you may want to work with for years on many projects. I’m not saying that that that’s the thing, but

Julian Castelli (38:39.859)
Don’t get stuck into the point where you’re you’re you’re really just signing up for hiring more people and built and hiring an internal development team you know on every project. You know, every eventually for the ROI to to come come due, you have to have a fixed cost and then have the the the gains keep keep multiplying.

Rick (38:59.122)
Makes sense to me. This really is a new frontier and I think and and it’s kind of going back in time almost. and and yeah, you’ll you’ll you’ll know it if you’ve done any contracting work around AI transformation to this point, you know exactly what this is. If you don’t, like definitely think about this stuff ahead of time. It’s important.

Julian Castelli (38:59.273)
So those are some initial thoughts when you’re when you’re when you’re working on this.

Julian Castelli (39:16.632)
Yeah. It’s gonna come up. There are gonna be headline stories of it not working. Look, it’s a it’s a brave new world, but I think it’s it, you know, the the benefits are far greater than the potential dangers. And and I’m really enjoying doing doing some of this work with some of my companies right now and seeing what’s possible. And I’m just like I said, I I’m just my my jaw hits the ground when I see like look at this piece of custom software that would have taken three years to develop, would have cost two hundred and fifty thousand dollars a year.

And and and one one person who’s not really a software developer built it in two months. Now, of course, is it perfect? No. But you know, the the the this is something that’s owned by the company and and and and in house and it’s a real asset. And it’s it’s it’s gonna create careers, it’s gonna create champions, it’s gonna create opportunities, and it’s gonna create mistakes and headlines all at the same time. It’s gonna be interesting.

Rick (40:05.79)
That’s awesome. Yeah, t tell me about stage four or step four.

Julian Castelli (40:08.893)
Yeah. Okay. so step four is reporting to the board or reporting to whatever your governance structure is. And this is important. I I sit on lot of boards and every board is asking about AI. Lots of CEOs are anxious and and you know trying to figure out how to do how how how to, you know, how to implement AI number one, but number two, how to how to, you know, convey to shareholders that they are gonna be w one of the winners in this race, right?

And so what I have here is w what a a an outline on whoever your stakeholders are, how to how to how to communicate and keep the the the group that is is is help helping to either fund or just h helping you in terms of governance to to make sure that that you’re keep they’re able to keep you on track and keep you accountable.

And if if you if you own your own company, you should do this for yourself. And you should you you know, I recommend that you create some sort of accountability board so you have that because it’s a it’s a best practice. So, first thing you do is is you know create an outline of the initiatives that are out there. You’d think this is so so so basic, Rick, but I I sit on five boards and I’ve seen very few, even as as good as a simple little little little summary of here, where it’s like, what is the department?

What is the initiative? You know, when is it due? And are we on track or at risk? As as simple as that. You know, but I you know, I think CEOs are a little anxious because they they have huge expectations and this is new. I’m just not seeing it. I’m just like, guys, come on, this is basic one on one stuff. Share what’s going on, you know, particularly if if we’re spending money on it, company resources. Let’s be clear of what we’re trying to achieve.

And and and and is it on track? And some, you know, they don’t have to all be green, right? But that that’s that’s part of the scientific method. But we’re gonna we’re gonna use a scientific method to say, okay, here is our hypothesis. Put your hypothesis onto the impact report. Okay. here’s what the initiative is, here’s the value metric that we’re gonna use to judge it. Here’s the hypothesis. Today it takes four hours. We think by the end of the project, it’s gonna take less than one hour. Okay, what’s the result?

Julian Castelli (42:30.245)
You know, in this first line, guess what? It was actually, you know, our early results are even better. So we’re exceeding the hypothesis, right? The second one is okay, you know, we we we are a little behind on on the on that target, and some of them just aren’t gonna work at all, right? But that’s the scientific method. You learn by stating a hypothesis clearly, doing the experiment.

And then observing and and what did you learn from it? That’s and and and the faster you get into those cycles and the more rigor you have is the way you get better. And no one’s gonna do this and and bat a thousand, right? If you can bat three fifty in the major leagues, you’re you’re making millions of dollars. Right. So, right? but you need the at bats and you need the ball but someone calling balls and strikes. What you can’t just do is like, you know, be be practicing in the backyard and not telling what’s happening and then, you know, hopefully come up with some huge win and and and and expect, you know, you know.

Rick (43:10.866)
Absolutely.

Julian Castelli (43:25.088)
Some people will do that, private companies that can, you know, can’t afford to do that can do that. But if you’re expecting your shareholders or your your your your board to to give you more leash and more funds and more ability to go do more projects, then invest in clear reporting so that they can they can understand the bets that they’re making with the company resources that they’re they’re they’re charged of governing.

Rick (43:49.087)
Makes a lot of sense. I I love the example report you have up on the screen. So if you’re listening on podcast, definitely recommend finding a video link to see this. do you like I see the outcome reporting on the impact report. I’m curious, like how do you think about ROI? Like th there’s a cost to get to this outcome. usually it’s bl it’s it’s someone

Julian Castelli (43:57.768)
You do video.

Julian Castelli (44:01.826)
Like how do you think about ROI? Like there’s a cost to get for this.

Julian Castelli (44:09.826)
You forget. But it’s it’s coming.

Rick (44:12.018)
The cost is someone it’s tooling, right? Like we talked about, the contracts if we have those, but it’s also people’s percentage of people’s time not doing something else. Have you found a good way to to measure that?

Julian Castelli (44:14.71)
Like it talks about the counteractivity and the the auto people find something else. That’s right.

Julian Castelli (44:24.437)
Yeah, that’s a great point. for the for the it’s it’s kind of easy when you’re when you’re buying, because you know, you have a annual budget for for a SaaS software. It’s it’s it’s maybe a little harder when you’re contracting because now you’re you’ve got a consulting fee that’s gonna last for a certain period of time. And to your point, you sometimes you know, you’ll you almost always will allocate internal resources, but do you allocate that as part of the cost? I think that’s a good practice to do that, but it becomes a little bit more gray.

and then there’s probably all sorts of other people get consulted for that. So like the the time it becomes a little bit of a a little more challenge. But I I’d recommend that you try and be a holi as holistic as possible when when doing that. And so you’re honest about the ROI.

Rick (44:54.877)
Mm-hmm.

Rick (45:06.62)
Makes sense to me. This is great. I this is I’m one of my takeaways from this is I I can do a better job personally at at the impact report for the key stakeholders.

Julian Castelli (45:16.81)
Yeah. It’s it’s look, this isn’t rocket science. This isn’t about how to split atoms with AI. This is basic governance. Right. However, when you’re when you know, I’ve been to so many AI presentations where like, let me show you how to, you know, how to prompt Claude Claude and do a skill and do multi things with code. And I’m just like, you I I come out and say, Okay, that’s great, but I I don’t know that’s that’s what I what I do as a business leader. I I I actually need to get the the company to do that. And how do you do that in a structured way? That’s why

Rick (45:22.94)
Mm-hmm.

Julian Castelli (45:45.066)
You know, that what was great about this workshop is I was able to give this kind of governance and and operating execution lens. And then I was able to hand it to Kyle and he was able to talk about which model to use and and all the technical things, which, you know, I’m learning and I love learning, but I’m never gonna be the expert on that. So I that’s why I partner with with with folks like Kyle and and the folks over at at Ascend and other places. It’s great to have, you know, those kind of resources around.

Rick (46:10.622)
It’s awesome. It’s awesome. Well, thanks for sharing this. What else would you I know we’re getting close on time, Julianne? Is there any other key items you wanted to cover before summarizing?

Julian Castelli (46:12.15)
Yeah.

Julian Castelli (46:18.066)
No, we’re we’re we’re right at the end here. So, you know, the the the goal is if if you wanna adopt this, schedule your value scan with your ELT, rank the top ideas on a two by two matrix, you know, then go to the bottoms up, have do the same things, choose two or three initiatives, go through your build buyer hire decision, and then use the scientific method.

to to to whoever you know you you work with to to help drive accountability and progress. And that’s stage two. Once you’ve done this, you know, and and you and you’re you’re on track, come back and check out stage three. we’re gonna be presenting that today at our event at Growth Elevated. And you know stage three is about putting it in your product. all this deck and all the stages are available on growthelevated.com.

And if anyone wants help doing this stuff, you know, I’m available at Gro at Growth Scaling Partners. Awesome. Well thank you for the time. I’m forward to discussing these three. Awesome. Thanks, Rick. This has been fun.

Rick (47:14.558)
Awesome. Well thank you for the time and looking forward to discussing stage three.

Timestamp

Introduction & AI Transformation Framework (00:00:04) Rick welcomes Julian to discuss Stage Two of the three-stage AI transformation playbook.

From AI Adoption to Process Transformation (00:03:21) Why companies must move beyond individual AI productivity to automating core business processes.

Where AI Creates Real Business Value (00:12:45) How AI drives revenue acceleration, cost reduction, and operational efficiency.

The Four-Step AI Transformation Framework (00:21:56) Julian introduces a structured approach to selecting, executing, and measuring AI initiatives.

Prioritizing High-Impact AI Projects (00:22:54) Using impact, ease of implementation, and data readiness to identify the best opportunities.

Build, Buy, or Contract (00:30:25) How leaders decide whether to develop AI solutions internally, purchase software, or hire experts.

Managing AI Implementation Risks (00:36:14) Why ownership, security, documentation, and vendor accountability matter.

Measuring AI ROI & Reporting to the Board (00:40:08) Using clear hypotheses, performance metrics, and reporting to track AI investments.

Closing & Preparing for Stage Three (00:46:18) Julian summarizes Stage Two and previews AI integration into customer-facing products.

Growth Elevated Leadership Podcast
Growth Elevated Leadership Podcast
Where the Real Money in AI Actually Is
Loading
/

be a guest

We look forward to learning from all types of leaders, investors and advisors. We’ve had great discusions with Founders, CEOs,CFOs, CROs, CTOs Board Directors, Venture Capital Investors, PE Investors and Operating Partners, Executive Coaches and all sorts of Advisors. If your work involves helping Tech Companies win, we want to hear from you!

Scroll to Top