Jimmy Bijlani, AI Momentum Partners | Why Most AI Spending Never Reaches the P&L

Jimmy Bijlani is the founder and CEO of AI Momentum Partners, a senior-led firm that puts a dollar figure on the AI opportunity before a PE-backed company spends a cent, then embeds operators and engineers to capture it. He spent close to a decade at Google running professional services for Gemini and built the tech strategy practice at BCG. In this conversation he lays out why most AI spend never reaches the P&L, and what the companies getting real margin out of AI are doing differently.

You'll Learn:

- Why "AI adoption" is a feeling, not a number, and the Tuesday-to-Wednesday test that exposes it

- How the bar in private equity shifted from the rule of 40 to the rule of 60, and what that demands of an AI program

- Why handing AI to IT or a Chief AI Officer usually stalls, and who actually needs to own it

- The case for starting with the P&L and letting the tooling decision fall out of the strategy

- How a 90% license-adoption number hid a structural zero in EBITDA impact

- How one automation on software the client already owned produced about $700K in NPV

- What "sticking around to deliver" looks like: forward-deployed engineers and an embedded senior operator

- The three questions a CFO should ask on Monday to tell a real AI program from activity

Chapters:

00:00 - Activity vs. a real AI program

00:27 - Who Jimmy is and who AMP serves

01:21 - The one takeaway: tie AI to the P&L

03:08 - What AI Momentum Partners actually does

05:24 - Where the AI-to-profit gap first showed up (inside Google)

07:52 - Who should own AI? The Chief AI Officer problem

11:26 - Where to start: the P&L before the platform

14:31 - Case study: a PE-backed vertical SaaS company

17:29 - Why the CEO brought them in

18:14 - What the diligence found: 90% licensed, 60% legacy

22:00 - Telling people the hard truth without losing them

25:34 - Getting managers onside when headcount is at stake

28:42 - Why they built on tools the client already owned

32:00 - The first automation: pro-serve docs and $700K NPV

34:08 - The after-state and the real return

36:49 - What "sticking around to deliver" looks like

40:16 - Why PE targets don't connect to AI work

42:28 - What boards will demand in a few years

45:28 - Close

Listen on Spotify: https://open.spotify.com/show/3OlTNvh2FGlE4VJyCrMJVE

More Expert Interviews: https://www.checkpluris.com/expert-interviews

Proof of Work is the podcast of Pluris, a platform connecting investors and operators with the world's leading applied AI experts. Each episode features the builders, operators, and investors who've actually put AI into production, turning it from buzzword to bottom line through sharp case studies and practical conversations. We explore how AI is used to grow revenue, expand margins, improve operations, and create measurable value inside real businesses.

Learn more at checkpluris.com

Explore more interviews and connect with experts at https://www.checkpluris.com

Subscribe to our newsletter at https://proof-of-work-ai.beehiiv.com/

Transcript

SPEAKER_01 0:01

I it's not a margin play or growth play, it's it's honestly both. And right now most companies aren't really doing either of them well, uh, because they're not really connecting their AI work to the PL in the first place. I just want to be really careful here because this is the thing that everyone gets wrong. Walking to 10 keyback companies ask each, you know, what is AI spend returning in dollars? We're gonna get 10 different versions of what our developers love X, Y, and Z. That's not really the answer.

SPEAKER_00 0:27

Today's guest is Jimmy Vijelani, founder and CEO of AI Momentum Partners. He works almost entirely with PE-backed tech and services companies, and his model is unusual. He puts a dollar figure on the AI opportunity before anyone spends a cent, then sticks around to actually deliver it. Before AMP, Jimmy was at IBM, built the tech strategy practice at BCG, and spent close to a decade at Google, most recently running professional services for Gemini. We kept watching the same movie, companies pouring money into AI while their margins didn't move an inch. That's what AMP was built to fix. Today he gets into why so much AI spend never reaches the PL, why 600 co pilot licenses is a meaningless number, and what actually changes that math. Jimmy, welcome to Just Curious.

SPEAKER_02 1:17

Great to be here, Stu. Thanks for having me.

SPEAKER_00 1:20

What's one thing you want listeners to take away from this conversation?

SPEAKER_01 1:26

You know, I think if there's one thing, let it be this. AI, it's not a margin play or growth play, it's it's honestly both. And right now, most companies aren't really doing either of them well uh because they're not really connecting their AI work to the PNL in the first place. So I just want to be really careful here because this is the thing that everyone gets wrong. You know, when I say ties of the PL, a lot of folks nod that it's obvious, but it's not happening. You know, walk into 10 G Back companies, ask each, you know, what is their AI spend returning in dollars? And you're gonna get 10 different versions of, well, we are developers love X, Y, and Z. That's not really an answer, that's more of a feeling. So here's the frame that I believe actually matters. You know, in VE, the bar is shifting quickly from the rule of 40. Uh for the folks who aren't aware, it's growth rate plus even margin equal to or above uh to the rule of 60. So same equation, much higher bar. I think what the GPs are really saying is they're done picking between growth and bar jib and they want both. You know, AI is the level they're betting on to on both sides at the same time. So the takeaway here really is yeah, in the short term, you should be using AI to take real hard recruiting costs out of the business. You know, the the kind that shows up in next quarter's actuals. At the same time, I think that same program should be building the capability of the DIA foundation, uh, the operating model that uh turns a company to an AI native business, say two or three years out. Um, you know, if you spend 18 months and a lot of money and you have a beautiful slide deck to show for it. I think the companies actually winning this are, you know, they're running the same program and the same financial discipline at the same time. Everyone else is just busy.

SPEAKER_02 3:03

I think one in the way of saying that you gotta do both.

SPEAKER_00 3:08

Yeah, yeah. And I want to walk through why it's so difficult often to do both and how you help organizations connect the two. For people who don't know you, what is AI Momentum Partners? And uh I mentioned a little bit about who you work with, but maybe a little bit more color around who you work with and how you work. Yeah, absolutely.

SPEAKER_01 3:31

Jimmy Majlani, founder of A Momentum Partners, also known as AMP. We lunched the company back in 2024. So the market gap that led us to start AMP was kind of stark. You know, on one end you have the MEBs and the big four, really solid strategy. Uh and no offense to anyone. I come from this world, but it's expensive, it's a bit slow, and it's frankly not really built from the mid-market. And you got the other end here. We've got technical agencies, really strongly on the build, not really that great about thinking about enterprise-wide value or ROI. So in short, I think the existing options were either a little bit too academic, too slow, too narrow, and none of them really were designed around the timelines that PE cares about, which was yesterday. So we built and to sit in the middle there. Today we're about 25 folks across full-time and part-time uh senior-led team, mix of MUB backgrounds, big tech, and startup engineers, and of course PE operators. Then what we do is focused on the the operation side of the house. Uh enterprise-wide AI strategy and cost takeout, technology solution development engineering. And where it makes sense, we also embed operators inside the company to run the execution. Uh we call it the 100-day AI plan. They stand at the PMO, you know, the whole thing. We try to own all this end-to-end. We've also started doing work on the AI diligence work for funds, but probably a different conversation. And the last thing I'll say here is that uh we focus on services and tech businesses, typically between 50 million to 500 million top-line. We feel like these are the areas where AI is having the most direct PL impact now. We are starting to see much more interest in verticals like finance, healthcare, and what we called uh the physical world too, manufacturing, logistics, all that good stuff. And uh technically our clients are usually CFOs, CEOs, P operating partners, you know, the people that have been numbered to hit at a clock that's ticking pretty fast.

SPEAKER_00 5:24

Yeah. Let's go back to the connection or lack thereof between investments in AI and uh profitability. When did you first notice that pattern? Why is it so difficult? And how does one go about connecting the two?

SPEAKER_01 5:43

Honestly, uh, I think the first time I saw this was in real time from inside Google. You know, my enrolled in customer engineering across ad Zen Cloud, I was working with some of the most sophisticated enterprises on earth, no shortage of smart people, and certainly no shortage on budget. I think the results were quietly underwhelming as it related to AI. Nobody was getting fired over it. Uh there were no big public uh flameouts. It was kind of subtle. The spend kept going up, the margin didn't really move, and nobody seemed to be in a particular area and asked why. I think the pattern was firmly consistent across companies and industries and also deal sizes. They buy licenses, they'd run a pilot or two, they'd brief the board on AI adoption metrics, which, you know, for beyond this is a phrase that you make every CFO twitch. And then six months later you could take a look at the P ⁇ L and find virtually no evidence any of it happened. So people weren't less busy in a way that translated to fewer people. Customers weren't being served with a lower cost, and engineering wasn't really shipping faster uh in a way that you can measure the calendar. All that's in the activities real, but the outcome was uh kind of a rude. Um and the reason was almost embarrassingly simple here. Nobody was underanated. There's no financial model behind the work. There's no operator accountable for captain value. There's no connection between an AI program and the actual operating decisions of the business. A lot of the AI, what was happening in parallel universe to the cost structure. Now you could shut down the entire program on a Tuesday, and that same company would have looked the exact same on a Wednesday. And that's also when the P angle really started to click here, too. Towards the end of my time in Google journey to work directly on Gemini, and it's giving me a clear view of what was coming. So yeah, at the same time, also watching PE back companies get pushed hard to adopt AI without any of the existing market options that I just mentioned being built for how PE actually works. You know, PE is hold periods, they didn't find out timelines, they don't have a spare 18 months for some digital transformation agenda. So that's really the gap here. You know, it's the reason that AND exists. Someone needs to come in, put a real number of the opportunity, and actually be in the building when the cost came out. So in 2024, we did.

SPEAKER_00 7:51

Yeah. Let's talk about accountability on the client side. Companies these days, many are handing their AI efforts to IT. It's owned by the CTO or the CIO. There's obviously a trend to hire chief AI officers. Who should own the AI initiative and how do you well let's yeah, let's start there. Who should own the AI initiative?

SPEAKER_01 8:19

Yeah, I think you know, companies usually hand that AI uh program to IT or you know, even a new chief AI officer who in many cases doesn't really have budget or the power to change how other teams work. And I think, you know, in these cases, you've kind of created a role that has all the responsibility but none of the authority and a very weird, like untangable LinkedIn title. Trying to figure out where the CIAO fits into the picture uh as it relates to the CTO, the CIO, the CFO, CXO, whatever. It's a whole different thing. And also take a look and think about what we're asking the CIO to do. You know, we're asking them to drive measurable change across functions they don't control, the budget they don't own, against the resistance of functional leaders who have very good reasons to be quiet about it, and then we're gonna evaluate them and outcomes that depend entirely on whether those functional leaders cooperate. It's a setup that would be funny if it wasn't expensive. I think the CIAO problem, it's structural and not really personal here. Now they're excellent people in these roles radion, they're trying their hardest, the architectural job, in my opinion, is the big thing here. AI that creates real margin impact doesn't sit inside one function. It's going to cut across pro serve, support, engineering, finance, back office. No single function has had the authority to make those changes across the org. And if we think about the CIAO, who's theoretically in charge of it. They usually have the authority to make the changes in exactly zero of them. You know, they have uh convening power, they have influence, but they don't really have decision rights. And any organization that's ever existed, influence without uh decision rights is a playwright sending well. Good luck. I think the monthly I haven't really covered here is rent it. I think that's seeing that announcement plays out when AI gets handed to IT. IT, they can deploy tools, they can manage platforms. Can I redesign how support handles tickets or how pro series folks engagements or how finance closes the books? Not a knock on IT by any means, but that's not really what IT is structured to do. So you end up with uh you know a beautifully deployed set of tools and zero workflow changes downstream between them, which you know, as we established is the dominant failure mode in this whole category. So what actually works, it's structurally different. I apologize, too, this is a very plunged answer, but uh you need an executive sponsor, typically the in many cases the CFO, perhaps even the CEO or the CEO, someone who's genuinely on the hook for the financial outcome. You need functional leaders who are going to own the change inside of their function, with very explicit accountability for value capture and outdoors adoption, and you also need someone with real operational authority. Sometimes it's an internal operator, sometimes it's someone embedded from the outside. But you need that governance so it connects the AI work to the financial plan. So it's actually part of the operating cadence, and it's uh it's not operating in parallel with its own dash core reviews. So when you do that, we have seen that you're not putting a single person in an impossible role, you're building an operating model around the work, and that's the big difference. You know, the company is winning at this, they don't have a better CIAO, they have better structure around the entire program. The title is a bit evaluated in my opinion.

SPEAKER_00 11:25

Yeah. And so let's say you've got that like ownership and accountability. Where should companies start? Presumably they're already doing some lightweight experimentation. Maybe they have kind of bought the tools and rolled them out and are hoping to see results. Probably aren't what's the right approach to begin to see like value creation from investments in AI?

SPEAKER_02 11:53

Yeah, it's a really good one. I think uh think about that for a second. Let me answer this one from a different lens.

SPEAKER_01 12:03

I think many customers they often buy the tech, whether that's a tool, a platform, without really thinking about what is the business problem that they're actually solving for. And it's funny, like literally in no other category of capital investment would you ever do this. Nobody walks into a board meeting and says, well, we should buy some manufacturing equipment and figure out what we're making later. It's in a simple role, we'll decide how we're selling that next quarter. But somehow in AI, this has kind of become the default. Like we back into this whole tooling tech platform first and then reverse engineer strategy from there. I think I know this part of that the reason it happens is that AI has become uh a board-level demand item without a board level definition. And as a CEO is being asked at every meeting what the company's doing about AI, and I think there's an enormous pressure to have an answer here. So if you know when you buy a platform that kind of produces this immediate, visible, defensible answer, like we signed an enterprise agreement with X, uh, rolling out Cokel across the organization, definitely one of my favorites. I think the fact that it doesn't actually answer the strategic question is a problem for an exporter's court meeting, which is you know, exporter's problem. So to answer your question, uh, I think the right sequence is almost embarrassingly obvious once you say it out loud. Start with the PL, you know, where's your calls concentrated? Where are the workflows uh with real leverage? Where is the data plain enough to actually do something useful? And also, you know, what's the financial impact of when after each of those pizzeria underwritten? So I think that's that's how you go up with your portfolio, and then and only then you start to ask which tools, which platforms, which build versus buy decisions support that portfolio. I think the platform decision should be the output of the strategy, not the input. I think there's also the longer-term layer here as well, you know, back to this rule of sixty that we keep talking about. Um, you know, asking which of these initiatives are also building capability that compounds. You know, if you're spending troll months, auditing support, are you also building the knowledge magic and foundation, dig infrastructure, the AI fluency that makes the next troll initiatives much cheaper to deploy? Because that's really where we see the Mi'kmaq trend sitting there. And not in any particular use case, they're all valuable. It really comes from the compounding effect of doing this well over multiple cycles. And again, most companies are doing this exactly backwards. They pick the tool, they deploy it broadly, they hope for the best, and then they wonder by the numbers to move. You know, we've cleaned up many of these situations over the last couple of years, and we know that's a very expensive way to solve this problem. And the most painful part in the lesson is yeah, it's free if you just sequencing is the correct way.

SPEAKER_00 14:31

Yeah. Let's let's maybe can we unpack that sequencing? You mentioned the first step is kind of taking a look at the PL. Walk me through how you would help a company identify where identify and I guess prioritize like where the opportunities are by starting with the PL.

SPEAKER_01 14:51

Yes. Should we dive into the case study actually? I think that might be.

SPEAKER_00 14:55

Yeah, that's a great idea. Why don't we use the case study as a example of that? And so yeah, set the scene. Who came to you? What was the problem, and where'd you go from there? Yeah, totally.

SPEAKER_01 15:06

Virtual SaaS company, Prime Dequity Death, like many of the other companies we work with. I don't want to give too much specifics away, let's say between 100 to 150 top line revenue. Uh Dell Margins, fairly healthy, let's say close to like 40%. So, you know, ulti-business, not a turnaround by any means, and non-fixer wrapper. It's the kind of company that's already doing a lot of things and is already trying to figure out what the next year looks like. So I think the AI question on the table when we walked in was uh a pretty good one. CEO, the CFO, they already decided that AI was going to be a mean level in the value creation plan. And everyone knows that the BCP is uh could be a little bit of BS at times, but I'm not gonna talk about that one. Anyway, they've been investing, yeah, developer tooling was rolled out, the CTO was engaged, and there was a lot of enthusiasm. The board was aligned, uh, financial expectation that AI was going to contribute meaningfully to the marginal improvement. So the conditions were worrying about changing this market uh genuinely promising. Hypothesis that they walked in with, and I'm gonna be careful because yeah, it's a hypothesis here conceivable is already doing AI. We just need to organize it and scale it. You know, in their minds, it had to already happen. Tools were deployed after it was visible, engineering is using Copilot, uh support team, you can launch a chatbot, an error internal that is basically, you know, we're in great shape. We need someone to help us tie it all together and turn to that next wave. So that's the framing. That was also the frame that's well only about 90% of the time, but you don't really find that out until you've actually done the work, which is where you know we pick things up. And I think what was at stake was fairly concrete, you know. Explicit cost takeout targets in the value creation plan. The board expects AI to be a real contributor to the margin in the uh current fiscal year. And I think the part that gets you know really understated is the the opportunity cost plot running an epic quarter of delay. And if you e back company, you don't really spend 10 cats or value that isn't recoverable at exit. I think that's the actual urgency, but uh you season was true.

SPEAKER_00 17:07

No, no, no. I just was saying uh I agree. I guess like a question for you maybe they felt like they were doing the right thing. Who came to you then with the problem? Was it the board was like, hey, I see all these investments, I hear what you're saying, but I'm not seeing a change in numbers. Is that what kind of catalyzed your involvement?

SPEAKER_01 17:29

So this came directly from the CEO, and for us, that is one of the best screen vibes we can see. Uh if the CEO views this as an opportunity, also an existential threat, which you know, it's asset services. I mean, their entire business model, pricing, services, products, it's all being disrupted simultaneously. You know, the CEO came to us and he was the one who brought us this problem. Now, like in many other PAC businesses, the CEO and the CFO, pretty attacked. They shared a very similar mantra ethos philosophy, and that's when we knew that this project was going to be much more than just caping the ladling, which is obviously very important, but that cost and that margin focus really became front and center uh after a couple of meetings.

SPEAKER_00 18:14

Great. So got this set up. Investing in AI, maybe aren't yet seeing kind of the returns you come in. Um where do you go from there?

SPEAKER_01 18:26

Yeah, so let me just share a little bit in terms of uh you know what it would actually unpack me once we got a little bit deeper. I think what's interesting is the shape of the gap more than the actual size of it. You know, we did the actual work inducted stakeholder intermuties from the exec team down to frontline operators. You can just see a lot of voice agents these days, but that's another topic. We took a look at about 75 plus enterprise artifacts, ranking financials, SOP, system mass, budgets, historical project data. And then of course we had many, many working sessions were modeled the appro independencies at the workflow level. I think once you go to that depth, the real picture starts to emerge pretty quickly and didn't match the internal narrative at all. You know, in this particular case, some of the more interesting things are that uh engineering, you know, leadership is very proud of them. And you know, in most cases, engineering is typically the first mover and being a bit more tech-sanded at other functions. You know, over here, roughly 90% of developers are licensed on uh cursor and copilot, you know, on paper, adoption number. CTO, and understandably, was pointing to a major win. And then when we got deeper, uh we found that you know, in the actual workflows, about 60% of the code base was running a legacy language and proprietary ID that sent you a zero iteration with any of the AI tools. You know, developers literally copy and pizza code and out of the AI environment by hand. You know, the productivity game was real for the modern parts of the code base, but the modern uh part of the code base weren't really where the cost was. Cost was actually the legacy stack and AHO and wouldn't touch it. So the adoption metric was a real real one. The E the Dynamite was structurally immediately. Now, if I take a look at customer support, it tells us a very different story. And the reason I'm uh really anchoring on those two things too is because in most businesses, especially SaaS, the big opportunity here is really around engineering support and probe serve. Um so in this particular case, uh support had a chatbot deployed that was a live progress, but the knowledge-based DM was only about 20% usable. I'm not exaggerating by any means, that's uh literally what we had found here. And you know, the automation existed, but the deflection didn't. Chatbot was technically deployed and operationally useless, and nobody was really tracking the gap between those two things. Uh back office almost in test by AI. I think the bigger problem is the process of the needs were really ready for it a year. I distinctly remember uh uh you know one of the leaders in this company saying how hard in 2026 should it be uh to produce an org chart. We have to do mainly every time. Sorry, you know, spawn a form of process death. I think the worst you can do with process that is throw AI on top of it makes a display dysfunction faster. A whole lot of governance here. Uh which sounds gonna die, but obviously it's where a lot of these programs die. There was technically an AI policy, and the functional leaders told us flat out that employees never see it. The actual government was verbal guys to use it lightly, which is like uh the corporate equivalent of be careful. The owner to the AI agenda as I mentioned kind of fragmented between the CTO, CIO, CFO, COOL, the chief staff. And uh the plastic situation where simultaneously said everyone's job and also no one's job. So let me just kind of closely clip on this for them. You might say that the assumption was wrong, the assumption that license counts and pilots meant that they would join AI volunteer to fail. Well, it really makes years of foundation and read the visible activity, it just wasn't in place. It had the options of an EAP program, but it didn't really operate reality of one. I think the difference between the two of these things is 100% more valuable, dear.

SPEAKER_00 22:00

I want to go, I want to go deeper. And I'd love to understand how you sequenced solving these problems. But maybe before, and maybe this is the first step, how do you communicate to the organization and the leadership, not just the CEO and the CFO who presumably aren't bored with what you're exploring, but everyone else who thinks that they're doing it well, then in fact they are not. How do you go about doing that? Because I can imagine that people are sensitive in Bristol, and you obviously want everyone to be aligned in driving change. And so, yeah, tell me first, like, how do you communicate that? And then how did you sequence um resolving some of these issues, whether it was with respect to the engineering organization, whether it was customer service, you know, um, whether it was something else?

SPEAKER_02 22:52

Yeah, I think uh the first thing that I want to be very clear about is empathy. Empathy because I think unlike many of us, if we're doing something and we believe we're doing it well, don't be empathetic at the end of the day.

SPEAKER_01 23:05

I mean, these are quite literally people, and you want to acknowledge the situation that they're in, what they've done to get there, and really trying to help them understand that. I think that doesn't matter regardless of whether you're talking to the CEO of the company all the way down to a frontline worker. I think it's really important to approach with that type of mindset. And uh uh that goes a long way, you know, for the work that we've done here. And Terry's seat that's saying that's a really interesting one too. You know, as part of the work that we had done being out of the roadmap, ultimately it consisted of about seven different initiatives. We whittled it down from about 18. And I think sequencing is really important too, because we think of this in you know multiple lenses, right? We have all the governance operating model, take you know, all those different things. We call that the foundational layer. And there are obviously some elements that we get to get out of before others. For example, the CEO needs to be clear to the company about how they're proceding AI, what they want AI to do for the business, you know, essentially incorporate into the vision and communicate that across the organization. So I think that is really like the first part that uh many companies miss. You know, part of that as well, there are other elements that we found uh lead to successful AI transformation. One of them is standing out of an AI DMO, doesn't need to be overkilled in the mid-market. You don't want to have too much bureaucracy, but you do need a DMO to make sure that the initiatives are actually on and on track. Uh so that kind of you know to suddenly as to the governance of the operating model part. But if I was to cover the use cases part, you know, we have seven different initiatives. I think it's really important to think about uh you know short-term impact and also long-term uh capability building. Some of the initiatives are going to require much more foundational data work to happen, especially as we think about building intelligence later, which really drives the rule of 60 forward, you know, being able to have that central brain across your company. That's something that needs to start happening now. But when it comes to the other initiatives, you know, this is a PDAP company. Like you cannot wait uh two to three years to show uh impact in the PL. So we were very, very explicit about extracting two to three quick things that we deliver over the course of anywhere between one to six months that would actually show a substantial uh you know PL line impact movement. But in many cases, when we talk about that line impact to head count, we're referring to tools, we're referring to other vendor costs. But uh I hope that answers the question.

SPEAKER_00 25:34

Yeah, and I guess like going back to the change management, but also related to um kind of the return, I I would imagine that like a lot of the lower level department heads were worried about losing staff as a result of this, but you you kind of need them to be on board with respect to the change management and the transformation that you are proposing. How do you get those managers to align with you when losing staff is probably not something that they're super keen about?

SPEAKER_01 26:09

Yeah. No, it's a very understandable uh you know predicament that you know many folks find themselves in retail. I think uh, you know, if you walk into a working session and you asked uh the director of customer support how much of your team's work could be automated. The answer that you're gonna get is somewhere between Holmes Denim and a very efficient thank you. Uh it's not necessarily because it's a mine, but it's kind of a very self-informating question. No rational human is going to answer a self-informating question honestly when there's no upside. So I don't necessarily think the trickier is doing them to confess. The trick is maybe relative. You know you map the workflow at a level of detail where the tree becomes visible, regardless of whether anyone volunteers or not. And you send with the team, you walk through how tickets actually get handled. You ask what happens with each step, who touches it, how long it takes, all those different things. And I think once you take a look at this, uh themselves usually start thinking that, yeah, this part is kind of ridiculous. I think the other ticket also from Hall is too, already the more important one uh to answer your question, is really changing the conversation from your team is going to string to your team is going to do higher leverage work. It's not corporate speak, it's the actual reality that's plays out when it's done, right? You know, the support T doesn't disappear, it gets to be structured. Tier one work that used to chew up seniors in your time, it's nailed differently. The senior people now spend their time in the order problems and cut through retention, you know, on the work that actually becomes value. And the function gets smaller in place and stronger in others. I think uh, you know, that framing matters not just because it's more pleasant and also banets directly back to this rule of 60 points. Now, if you're only thinking about cost ticket exercise, the manager is maybe defensive, and you're literally there to you know take people out of the function. If you're thinking about is both cost ticket and capability building, mean the function gets cleaner and better at the what the actual dread growth. I think there's something here for them too. The T becomes higher leverage, which I mean more strategic the conversation comes from uh goes from how much me cut to how your redesign's function for what the next weaver is going to look like. Same workflow, same data, a very different conversation and dramatically more honest. I don't want Lieutenant assault sunshine, though, you know, their bandages will be stone ball regardless. I think that's where executive care cover really matters. Yeah, the CEO and the CFO do clear this is happening with the gold rail, definitely target rail, and put you saying productivity uh is the task release resistance. Most people figure out which way to make this one pretty quickly, and you know, for the ones who don't, you kind of learn to work around. But the vast majority, when you make it safe to be honest, and you give them a version of the story they can actually insult to their team, they're good to engage. Uh before they affect.

SPEAKER_00 28:42

Yeah. My understanding is you uh ran this whole process on tools and software they already had. Why did that matter? And I guess what was the first thing that you automated with the existing kind of um set of solutions?

SPEAKER_01 29:05

Yeah, I I think the big reason, and this is something that we very quickly learned where we're be back companies, is speech value and paramount. You know, the moment you introduce a new platform into the equation, you've added lots of procurement potentially. Integration, security review, every negotiation, change in management on top of whatever else you were trying to achieve. It's not to say that you know it doesn't always make sense to buy a new platform. I mean, look at how many of us who have done to claw over the last six months. Some things are very compelling, business reasons to do so. But it also means that you know the program no longer pays back at the current fiscal year when you're introducing a new tool at a platform, which means the financial story of the board gets weaker, and it means that uh the political momentum that AI programs need to actually capture um yeah, structure of so it's a very default. It's a principle, not a one-off choice, to use whatever the client already owns whenever it's good enough. So in this case, uh, the Microsoft stack is already deployed, go pilot it, Pallad automate, SharePoint, the whole environment. It's all sitting there. License is okay, but most of it underutilized. So we built the uh the entire panel inside that stack. No procurement cycle over integration project spinning up uh in parallel. Just a very delivered use of the tools that they've already paying for. The CEO, uh, CFO definitely appreciate this more than I can convey because the method of value attack was almost pure margin. The panel itself, uh you know, the first thing we automated, you know, built in them very clear about is we don't do strategy work, but we are doing strategy work, we need to automate something in parallel. I say that because we have spent way too much time seeing all uh the dusty data set sit out on a shelf after delivery. I know I've seen them while that's a BCG. And uh we try to also automate and show early fiber SPS in the AIM what we can. So we did a pilot, and the first thing we automated was uh professional services documentation drafting. I come back as an example often because it's a very clear illustration of the pattern. You know, consultants on their approach team spending real doable time manually writing uh requirements analysis, BRDs, uh statements of work, meeting summaries, you know, really highly structured and repetitive and templated documents that follow very predictable patterns. You know, senior consultant might earn, let's say, about uh two weeks on a fig map analysis, uh, one to two weeks rather, uh depending on the complexity. You multiply that by every engagement of a consultant every week, and you're looking at a very meaningful chart and catastrophe going to documentation as opposed to claim uh playing decent work. I think the sales uh equivalent to this example is uh you know, most sellers spend most of the time, uh a big chunk of their time not actually selling through between 25 to 30%, I believe. You know, when it comes to this particular pilot, uh this one out of a hundred possible candidates. We run a structured uh three-stage filter and a stage one financial materiality is the cost being spent to your action material to the PL. Stage two is feasible and towards spring. Can we build a little test and dominate this for roughly three or four weeks to exit to the stack? And stage three is about change capacity. So does the function actually have advanced right now to absorb the change and is the leader actually motivated to operationalize it? I think the last one is the most underrated filter. You can have a perfect financial case in a function, and you know the leader to know interest to change out of the team works, tell us for the diet production no matter how they detect this.

SPEAKER_02 32:21

So the pro server use case pass all three filters.

SPEAKER_01 32:27

Because they were able to see their best consultants breaking down the documentation instead of doing the high value work. We built this one. Uh we used toolpal studio, all the inputs from the transcript discovery notes because they're hand-offs, the historical shareable document library, around the station workflow together, and the output was a first draft document with citations and how to talk consistency. The consultant reviews or finds and ships that you know that one to two weeks becomes one to two hours. And the once that the game uh you know needs to be drug, it becomes very editorial. And the quality, uh the quality went up here because the drafts were a bit more consistent than what the individual consultants were producing at a deadline pressure. So to bring it to the end act here, we're willing to get a 700k uh in gross NTB from that single workflow. Sorry, MTV, not gross MDB. Uh on infra they already uh again, they already owned it in a software bot.

SPEAKER_02 33:19

And I think uh I close the loop on this one.

SPEAKER_01 33:24

The more exciting part about this pilot is while you're doing the strategy readout, you're showing the direct impact of what could be. And that's great in a couple of different ways. One, it shows the leadership team that yes, this thing can actually happen, and we should scale it. Uh we've automated all your documentation types, acting with the analysis, including the BRB, sounds all those different things. It also shows uh you know the folks in the ground, especially the consultants, how impactful they I could really be. And you know, for our company, open selfish course, it did be to every opposite self for us. In my opinion, this for the win-win-win. Now it typically comes down to build the capability inferring, so we don't have to be around forever.

SPEAKER_00 34:08

Yeah. When when you reflect on this engagement, you started with a perception that they were doing all the right things, but also a belief, you know, amongst the CEO and the CFO that they just weren't seeing the connection between those things and uh the bottom line. You spoke to one of the returns, $700,000 of NPV. Tell me what the afterstate looked like, um, if you can, um like overall cost cutting or kind of ROI. Um, how do you frame the return and how different like today is versus before you came in?

SPEAKER_02 34:50

Yeah, absolutely.

SPEAKER_01 34:51

So this client we're looking at uh I don't know, five to six percent uh improvement uh year one, which is quite impressive. A lot of folks will see that it's actually kind of small. And I just want to push back in a couple of things there. Number one, this is extremely conservative. It doesn't represent the the full opportunity size. If they go all in, and let's say let's use the knowledge base of customer support as an example, if they're able to remediate it to a point where uh you know we're running from 20% usability to about 40% usability, that's great. We're gonna hit our conservative target. But what if we take it to 60% or 80%? Yeah, that 5% uh EDOM prudence can quickly snow to 10%. Um of course it's nothing to set it up with one use case or opportunity specifically, but if we apply that thing across the entire portfolio, that's really substantial for the multiple. What I will say though is that it's going to be table-staked in the future. When every other company is uh AI augmented, not AI data, but AI augmented and using AI across the board, I think it's really going to have a significant impact on the multiple. So really where I think the big opportunity, at least over the next few years, is going to be it's really around the intelligence layer. I don't think they'll let closed loops at them, you know, anytime we have uh anytime you know information is coming through Slack chats, meetings, all of these different things um where the information can be captured digitally. I don't think we captured that into a knowledge-based repository and brain in which other folks across the company can also tap into this and use it, but also agents as well. To me, getting to this point not only is analyzed what it really means being a needed, but this is what it's going to mean to get to that rule of 60. Again, so I'm not sure if I answered the question, probably a little bit of a tangent there, but you can let me know.

unknown 36:49

Yeah.

SPEAKER_00 36:49

Well, no, that that's that's perfect. I want to ask you about um something I mentioned in the introduction that you um you stick around to actually deliver the um kind of returns, whether that's cost savings or other forms of kind of net present value. What does that look like for a client? What is what does your model look like when you're sticking around to deliver it?

SPEAKER_02 37:15

Yeah. But we're firm believers that strategy, but execution means they're really uh we've all been there before, you know, having worked an MBB from before.

SPEAKER_01 37:29

There's great value in the work that they do. But many of their firms, MBB or MBB as I'll put out the data and strategies agree on labor, but there's no execution behind it. It's something that comes useless because sure educated and the alignments of folks too, but it doesn't actually translating to impact. So we don't want to make that mistake without clients, then that would be assist at all. And there are really two approaches that we have when it comes to extending support out on strategies then. And the two areas where we're really go deep. One of them, I'm sure you would have guessed it, forward deployed engineering, all the retail, something we've been doing for a while. And typically what this looks like is we'll drop an engineer in. Uh they'll typically start with a particular function. If we already know what the opportunities in, uh they'll they'll go to further immediately. But if this is let's say a fresh customer, but we haven't really done a great strategy because not every customer wants that. I totally get it. Uh it's a it's a huge investment commitment. Sometimes they want to start small and you know scale from there. What we'll do is we'll drop in the FDE, spend about a couple of weeks assessing the currency to identify the opportunities, kind of doing a lightweight functional strategy, if you will. They'll spend the next few weeks actually building out a pilot, thing that's production ready, that can be used and scaled across the team just to show what the value looks like there. Now the second thing that we do, it's quite different than the FDE model, is a senior operator. So we started doing this earlier this year when we found that companies really needed help getting their program off the ground. And the senior operator will come in and typically hit a VP of AI transformation level. And what they'll do is a few things. They'll stand up the AI DMO, they'll stand up value capture, so if they already are a portfolio of initiatives, they'll make sure that um the other ownership they're actually moving forward with this. Uh, I think those are really critical elements. What the senior operator will also do is conduct target enablement. And sometimes this looks like a champions program, which probably works really well. So I'll take some of our clients to about 35 to 40% adoption to almost 100%. And in this case, it is they are working directly with functional uh champions. Uh we need this head using the phone to really understand their function and is it the best. We find that champions they're deployed within the functions opposed to relying on extra insolvents is typically the best way to go here. And let's see if I'm missing anything with this uh senior operator. Oh yes. They will also run the Hunt Day AI plan. So um, in many cases, it means helping bring in some until they to lead that going forward. We also help with that too. Really, it's the kick starts of organizations need to figure out where do we start with AI too. Well, we now have a program, it's working well, it's only going to get better, and we don't need to rely on consultants until the end of time to uh keep delivering.

SPEAKER_02 40:16

Yeah.

SPEAKER_00 40:18

With respect to private equity, they are well, the the companies that are owned by private equity have real targets, they've got real deadlines, but most don't seem to connect those targets, those value creation targets with AI work.

SPEAKER_02 40:38

Why is that? It's a good question.

SPEAKER_01 40:43

I think um there's one question that the CFO needs to walk into the room with, and it's a slightly simple one. It's really showing where AI has changed our cost structure. Now we deploy it, not we're using it or any of these nice little any adopter metrics, but showing the actual dollars where AI has actually changed the cost structure of the business. I think the answer is a company specific number. Yeah, it's tied to a specific initiative. Well, great, I think you have a build program and keep going. But if the answer involves a lot of qualifiers, you have references to things like productivity gain and solid benefits or any version of the phrase we're still measuring it. I think you have activity, you don't really have a program, you certainly don't pee not outcome. And so the longer that treatment about says, the more expensive it gets. I think you know, again, rule of six yeah, I thought I could put a record, but you know, showing me what capability we're building that compounds is also really important to ask. Because the cost to get an answer is real. You're not also building out the the data foundation institutional knowledge operating muscle limits and next uh next time easier than the first one. You're also running series of one-time savings. That's a big margin exercise. Time of transformation, you really do need both. Um I think the the third thing you have to be agree is who's accountable for both of those things by name. Not the person or committee, but actually what person? If there isn't a single name attached to the financial outcome, the financial outcome is just not going to happen. That's what every engagement we run has ever taught us. Yeah, the CFO walks in on Monday and asks us three questions in that order, cost attack capability, who owns? It's gonna tell a lot more about the actual study AI program in 20 minutes than they've had in the last, you know, however many board meetings.

SPEAKER_00 42:28

Yeah. And I guess related, um in a few years from now, what will boards expect leaders to prove about their AI spending? And what should companies start doing now, whether that's building things or otherwise, to be ready for that?

SPEAKER_02 42:50

I think in a few years the conversation is going to be extremely different from where it is right now.

SPEAKER_01 42:55

I think boards right now are very much tolerating uh the or making hard designed AIs and the generate update. But that would be instead of close and honestly even by the next quarter. It's going to be treated like any other outrageal investment, which means CFO is going to want hard attribution. You know, because we put in here's what came out, here's the limited concentration, the margin, here's the impact of growth, here's the volatility per employee versus our peers, and all those different things. They're not gonna be optional anymore. Uh they're gonna be table sheets family issue team that must be credible on a board meeting. And I think um, you know, the companies that built AI is a margin lever and a root foundation, they're gonna stay on real operating leverage. Economics, faster delivery, lower cost server, genuine airflow to the daily work, they built to grow that proportionally adding costs. I think that's what AI really means. It's the financial reality that you see in the model. And uh, you know, okay where the bars are building so quickly, these are businesses that are commanding premium multiples at exit. Um, you know, the companies are random activity-based AI program of license counts, deployment rate, there's a whole figure of it. They're gonna have a very economical few cores. Once your companies start building out the running, I think there are a few things. The first is financial discipline. You know, get rigorous about actually underwriting every initiative against the PL, uh, stop counting with solved benefits and solving the adoption metrics, treating like a capital allocation decision. The second thing is the foundational capability work, data architecture, knowledge management, operating model governance. It's not glamorous to spiral. But the companies that do this work now are gonna deploy A initiative in weeks while their competitors are still uh do that procurement. The third one, probably the one that's most underestimated, is the people. AI fluency is not a training program you run once a workforce capability to be built deliberately over many, many cycles. And the companies with change are typically been working on AI in the daily roles, they're gonna look very meaningfully different. So I think uh the last thing I'll say is uh I think every listener myth it's not really about AI. It's about whether you're building a company that can absorb continuous technological dangers and uh a very real and pregnant operating malady. AI is not the last waves currently, but it's gonna be another one after that, another one after that as well. So if you start pulling the operating muscle down, the discipline, the foundation, the fluency, you're gonna keep winning. You know, the companies that uh keep treating each way as a separate transformation of its own committee, these commies are gonna be very busy and very expensive, and they're gonna fall behind. But I own crystal walls too, Tannel Tell, and we're certainly here for it.

SPEAKER_00 45:28

Yeah, I'm here for it. What a good place to end, Jimmy. I really appreciate the conversation, the clarity, uh the details, and I appreciate you coming by. Just curious.

SPEAKER_02 45:40

Yeah, thank you much too. Appreciate being here.