KP Unpacked

Choose Your Team, Not Just Your Tools

• KP Reddy

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 52:29

What if the next five years of your career isn't defined by which AI you use, but by who you're working with?

In this episode of KP Unpacked, KP Reddy and Nick unpack the quiet revolution happening in management consulting. OpenAI just launched a deployment company and acquired a consulting firm. Anthropic is backing enterprise AI consultancies. PE firms are partnering with AI-enabled consultants and offering equity instead of hourly fees. The result? Three tiers of value capture emerging: billable hours (worst talent), risk-based fees (middle tier), and equity models (where the best people go). If you're still getting paid by the hour to do AI transformation work, you're in the bottom tier.

But the deeper insight is about career trajectory. KP argues the next five years aren't defined by how good your Claude skills are. They're defined by who you're sitting next to. Are you in a firm where Opus 4.8 launching makes everyone's Slack light up with memes and excitement? Or are you somewhere people still think AI is a threat? The gap between those two environments is the gap between relevance and obsolescence. The conversation also unpacks skills files as potentially employee-owned IP (not company-owned), why structural engineers still double-check software calculations in Excel despite working for billion-dollar firms, and why Zero's training program spends two-thirds of its time on mental models and thinking frameworks, not AI mechanics.

Key questions answered:

  • Why are OpenAI and Anthropic launching consulting practices and partnering with PE firms?
  • What are the three tiers of value capture in AI consulting (billable hours, risk fees, equity)?
  • Where does the best consulting talent go: hourly billing or equity models?
  • Do you own your skills files, or does your company?
  • Should companies make employees sign IP agreements for marketing coordinators building AI workflows?
  • Why do structural engineers still double-check software calculations in Excel?
  • What's Zero's training curriculum focused on: AI tools or thinking frameworks?
  • Why does ambition and optimism matter more than technical AI skill?
  • How should you choose between working at a forward-leaning AI firm versus a traditional one?
  • What happens when Opus 4.8 launches: does your team's Slack light up or stay silent?
  • Why would you sell a $250M/year AI consulting firm when you're banking $50M annually?
  • What's Ramp tracking now: token spend by industry?

If you're deciding between firms based on AI adoption, wondering whether your skills files are actually your IP, or trying to figure out whether billable hours still work in an AI-enabled consulting world, this episode will make you realize the technology matters less than the ambition and optimism of the people around you.

Listen now.

Allergies And A Dusty Studio

SPEAKER_00

Hey Nick. KP. Where you been, man? I was gonna ask you the same thing. I think you had the sniffles last week and had to cancel.

SPEAKER_01

Yeah, I did have the sniffle. So I did, it's kind of a uh dumb thing, right? Like I, and you will find this as you get older. My uh resistance to allergens has gone way down. It goes down every year, by the way, right? When I was a kid, I you know, growing up in Atlanta, I could run around in the yard and eat the pollen, and I was fine. By the time I was 25, I couldn't even like be around someone cutting grass. And then uh now it's like even worse. So normally, like funny thing about moving to the Bay Area, no allergies. I'm not allergic to anything they have out here. I land in Atlanta, I'm done, right? But I still have massive dust allergies. And so I did this radio interview last Monday, and it was in an actual radio station. Nice, that was fun with like wood-paneled, like whatever, something out of the 70s, right? So good, kind of WKRP in Cincinnati for those of you that want to age yourself. But I don't think they vacuumed in there since the 70s.

SPEAKER_00

Wow. So you that was kind of gross. That's kind of gross, more or less got sick from being in the studio.

SPEAKER_01

I got sick from being in that studio.

SPEAKER_00

Wow.

SPEAKER_01

Yeah, it just it just knocked me out.

SPEAKER_00

That's brutal.

SPEAKER_01

Yeah.

SPEAKER_00

So it's also yeah, that's kind of kind of crazy that that dust is doing is doing that to you. Yeah, I'm sensitive.

SPEAKER_01

I'm I am a sensitive flower. I mean, that's that's how everyone describes me.

SPEAKER_00

Well, when you when you when you started telling me this story, I thought you were gonna say moving to California actually affected your allergies. And I was like, what? Because you know, this the South, I mean what we have, yeah, like what I have in Kentucky and what you know what you had in Atlanta, it's it's really bad. The allergies are like it's a reason you actually should not live in the South if you have bad allergies. That's one reason. Yeah, yeah. Not not the only, but yeah, but I'm yeah, moving to California, I expected it to be a lot better, and obviously it is, yeah.

SPEAKER_01

Yeah, but it's it's funny because you know, I we have one of these hypoallergenic dogs, which we didn't buy. A friend of ours moved to Japan and we ended up taking this dog. I am allergic to our dog, like unless she's groomed and bathed regularly. I I get I get the snuffles around her.

SPEAKER_00

Yeah.

SPEAKER_01

So yeah.

SPEAKER_00

All right, so you're back feeling better after the had the rate, had the radio segment go. Was it similar to a podcast? Was it different?

SPEAKER_01

You know, I've done it's kind of interesting. I've done pre-podcasting. I used to do a lot of radio shows. It's it's similar, you know, other than the commercial break, right? The little because it was live radio. Yeah. And then they do like, I think 30 minutes of live radio, and then it goes into their podcast format. So you have to like stop for the commercial breaks, which is kind of you know, it's kind of actually not bad, right? I feel like if we had a commercial break, we might adjust what we're gonna talk about next. It gives you a break to do that, which that that's not bad. But the conversation was super interesting. I mean, this these guys they uh they run like one of these what's it called, think tanks around studying consciousness and machines developing consciousness. And so, like Stephen Wolfram's part of it, it's it's it's a pretty like heady group of people, right? So had some really cool conversations about how we think about kind of uh we went down this rabbit hole of skills files. And I don't know if you want to explain to people what skills files are, but yeah,

Skills Files As AI Instruction Manuals

SPEAKER_01

we've talked about it a bit.

SPEAKER_00

I'm sure the audience is getting familiar, but to refresh, a skills file is basically giving AI, giving whatever your LLM, LLM of choices, an instruction manual on how to do a specific set of work. So you could have a skills file for architectural drawing, hypothetically. You have a skills file for marketing copywriting, you could have a skills file for business development, outreach, cold outreach, basically like any functional human area, you can create a set of instructions that are hyper-specific. And maybe, you know, I think one of the one of the exciting aspects of skills files is it can be somewhat proprietary. Like if you have a really unique set of knowledge, like you can give you can give that skills file a lot of tokens and instruction on how to conduct and execute work. And you know, that that that might not be transferable to any large language model, more or less. Like the average person is not gonna know to instruct the model to do this. Like that's that like there's an advantage of experience.

SPEAKER_01

So so that was a great, that was a great description.

SPEAKER_00

Oh, thanks.

SPEAKER_01

So what we talked about is if you go work for a company, you sign off your rights, right? You work at this company, it's the company's IP. But many of us are developing very deep skills files, right? That to that are very proprietary to us, right? For me, it's not a big deal, but for someone else, it might be a big deal. So it used to be, you know, when we talk about these PIA, like you get your people to sign away their IP and all that stuff, right? It's generally been highly focused on software engineering or

When A Company Claims Your IP

SPEAKER_01

research into like technical people that build stuff, right? Not as much. I mean, I think a lot of folks like we do it as part of like routine, right? It's just the next step on the hiring process. But you know, are you getting your marketing coordinator to sign those agreements? Right? Maybe, maybe not. You might think, like, why do I care? Right. But it turns out like your marketing coordinator might be building a pretty solid set of skills files that are theirs that they can take to the next firm over that maybe is a competitor. And so we're just talking about like, so you know, in in that world, could you also like sell your skills files if it's really good as a piece of IP? And like one of the things we started asking people as part of the hiring process is to develop skills files, right? I have an intern with me in the half meme office, and I gave them a task, and I got them all Claude trained through our certification process, and then I was like, hey, go do these tasks. I let him go do them for an hour. I'm like, hey, do you have a mental model of what you need of the tasks that need to be done? And he was like, Yeah, I'm like, great, get Claude to build a skills file, get Claude to do the rest, right? And he was like, Oh, like he thought he was gonna be sitting here all day, like doing internet searching. Oh no, sir, we don't we don't operate though. I'll give you the next thing to do. But but you know, on this radio show, it was a pretty interesting debate about like this idea of like how do we think about IP, and you know, we're we're all generating a lot of IP, and of course, you know, is all of it defensible? Probably not. Is some of it more defensible? Maybe, you know, I don't know. But I do think what is super cool about skills files, and I believe this about the AEC industry specifically, is we can't help ourselves, right? The engineering brain, if you give me a set of standards, what's the first thing I do? Change them.

SPEAKER_00

Yeah, sure. Yeah, you're gonna want to you're gonna want to deconstruct first principles exactly.

SPEAKER_01

Right.

SPEAKER_00

What's in the yeah, what's mechanically involved? We got to go back to square one, basically.

SPEAKER_01

Right, right.

SPEAKER_00

And this is how I do things.

SPEAKER_01

It's funny, I was talking to someone, this is years ago, maybe like only like 10 years ago, to a structural engineer that used was using a very well-known structural and engineering analysis software. And he was like, Well, I also like double check at Excel. I'm like, are you out of your mind? Like, what, like, why would you do that? That makes that makes no sense, right? I mean, it's like when the big multi-billion, like, this is not some little company, multi-billion dollar company, right?

SPEAKER_00

I I gotta know for myself.

SPEAKER_01

I got it, like, I'm putting my stamp on it, like I need to know. Yeah, and I I can't deal with this black box, and I think it's just an absolutely fascinating thing, right? Like when you look at you know how different construction companies do estimating, they all have their system, right? But then if you talk to individual estimators, they're like, Yeah, yeah, yeah. So like we use the on-screen, whatever, and we have standard, and then I export to Excel and do it my way. I'm like, you work for a $10 billion GC. This is like insane, right? But you know what? It is what it is, right? That's just what it is, right? You're not gonna change people.

AEC Skepticism Of Black Boxes

SPEAKER_01

So it's pretty interesting when you look at some of this AI stuff, and a lot of the resistance to change, people think it's like purely like, oh, AI is gonna take my job. I don't think most construction people believe this. Like, it's not gonna take my job. That's not how that works, right? But I think what they do struggle with is you know, I put something into the black box and the answer comes out. How did it do that?

SPEAKER_00

Sure. And I and I think one interesting thing about AI is that it does not use the same process to find the answer that we use, right? Like it's not like the process for it to, you know, to execute a math formula is entirely different. And that's actually one of the critiques, is like, you know, it's much better now, but early on, if you asked AI to you know do a really simple math equation, like a lot of times it would get it wrong because it's not able to like, you know, process-wise, like it's search, it's searching for it's search, it's searching, it's searching through text, right? And pattern matching, probabilistically giving you back an answer based on what it's seen. It's not actually just doing four plus four.

SPEAKER_01

Yeah. So I so I was at this job site, right? And this superintendent's there, and I was like, hey man, what do you do all day? Tell me about what you do all day, right? And he's like, Man, the drawings from the designer suck. So I'm always having to deal with that. And subcontractors act like they don't know how to read drawings. So I'm always like working on drawings. And so I was like, What do you think about this idea that everybody looks at something differently? And he shows me like the same set of drawings, and he was like, Yeah, like some people might estimate this, like square footage from center of column, other people might estimate it from edge of column. And both are right in a way, like you can get to the generally the right answer, right? And he's like, But you know, people are just weird that way. That's just how they, you know, there's some guy that does center of column because it's quick and easy, and then they discount the square footage by X percent because they have some rule of thumb in their head, yeah. And to discount out the this the size of the column or whatever. And I was like, this is wild, right? So you start to realize like, no matter how much process you push on this industry, right? We're still designing and building a custom project every time with the set of actors that changes every time with no real standards, yeah. Yeah, guardrails, yeah.

SPEAKER_00

I think the I mean, one, I think what's interesting about our industry and different than other industries is the level of risk involved, right? So, like, why is that structural engineer double checking the math that a computer is going to do more accurately? Like, statistically, the computer is going to be more accurate than him. At the end of the day, he's still signing off on the structural integrity of the building that he just designed. Right. So, like he still has to be comfortable with that idea. And I think just yeah, like in general, the level of risk that whether you're a contractor, whether you're an engineer or designer, like there's gonna be some, there's gonna be an instrument to not to never like fully trust someone else's methodology or process, despite right how efficient standardizing might be. Yeah. And you know, I listened to Thomas Lafont from speak recently. He's a founder of the city.

SPEAKER_01

Yeah, we're just talking, I was actually talking to our friend who's very close to him. Nice. And just yesterday.

SPEAKER_00

Oh, cool. Yeah, yeah. So he he when he was speaking, he said for every large deal that we do, so basically any deal that he's involved with that like circles up to his deal.

SPEAKER_01

By the way, Kotu, if you don't know who KOTU is.

SPEAKER_00

Oh, yeah, very large investment firm. They, you know, have had have had multiple different types of investment strategies over the years, but you know, multi-billion dollar investment platform. Now they're buying land. Yep, speculative land, yeah, yeah. Yeah. Actually, I'm like excited about that thesis, but that's that's a sidebar. But what he said was I I build a financial model myself for every deal that I do. Yeah. Like he just has to see the math himself, right? Of course. Obviously, he's got an army of analysts that have built models, like they've all probably built their own, you know, models and maybe even more sophisticated than what he's building. But he's like, I gotta see it with my own eyes and I gotta think through it, and that's how I think. And I thought I still I found that fascinating because like I don't know, I don't think you know Bill Ackman and every well-known public investor these days is running their own models for every deal they do.

SPEAKER_01

Yeah, I don't know if I told you when we back in the day, you know, I used to do tons of acquisitions, and I would create the find the company we're buying would send us a financial model, and I would always start from scratch. It was it was the weirdest process, but I don't know, maybe not that weird. I would have printouts of their financial models on you know one of one of these kind of things, right? I have a flip flip through them and I would read through it and then create a financial model from scratch, like literally file new Excel, right? Yeah, which is like wild. Like, why would you do that? But that's how I that's how I gained an understanding of the business. Yeah, you know, I don't I don't think I would do it now. I I will say I still monkey around in Excel to create rules of right, like I will do like I'll sit there and play around with sensitivity, right? Because I like to play around with like, well, you know, the marketing spend should only be five to seven percent of the overall revenue, like stuff like that. Like I like to play around with those kind of key ratio stuff, and I still do a lot of that monkey around myself. I would never definitely not build an entire federal, but I still look, I think there's a little bit of I think people like to chalk it up to like, oh, they don't trust AI. It's like, but no, actually, we all still want to learn, we're all still trying to gain an understanding. And I think like a lot of times AI, we use AI to gain an understanding of things that we're just not intellectually that interested in.

SPEAKER_00

I saw like a great, a great pithy tweet on this. It was you can outsource your thinking to AI, but you can't outsource your understanding, right? I think it's like there's like a lot of wisdom in that. I I think I think the incentive is going to be more and more to outsource a lot of your basic tedious thinking that you don't want to be doing, but that deep first principles level understanding, especially for all the engineers out there listening to this, like no one's asking you to do that, and you and you shouldn't, like you should always deeply understand.

SPEAKER_01

Yeah, I mean, I think people get tired of me talking about first principles. Like, I feel like I say it like 10 times a day, and it sounds like I'm trying to be like this hyper intellectual person. I'm just saying, like, what a great time to rethink everything. Like, stop building off the basis of all the things you know and all your experience and all that, right? Like I said something the other day that

First Principles Versus Playbooks

SPEAKER_01

got my team mad at me, and I said worksheets, workflows, best practices, playbooks, those are all paths to mediocrity. And they were like, but how do we function without a workflow or a process or a work, you know, uh a playbook? Like, what are you talking about? And actually, and it was funny, like I sent them a copy, a chapter from my book, which was Deming will be turning in his grave, right? Um because of AI. And I think if you look at Deming's principles, it was always about building process and standardization. Standardize everything, right? Everything has to be standardized. And when you standardize everything, everything's a checklist. And and this was going on in Slack, right? And then I ended up writing a blog post about it saying, you know, in the blog post I wrote on Substack was McDonald's is the best process engineering engine. That doesn't mean they make a great cheeseburger. You know what I mean? It's and that's what I was trying to get through to folks. Like some of you know, to me, the AI unlock is that everything doesn't have to be a highly repeatable process, right? And we actually get to think about things individually about different things, how we approach things, right? And you don't have to have a playbook for everything. The playbook can change, right? It's kind of like what I was doing in robotics with mass customization, right? It's the same idea, right? And I think the people, it's so ingrained in them. I need to go build a playbook, right? Versus saying, no, I don't need to build a playbook, right? Like, no, like let me get back to first principles, right? Of what it is we're trying to do. Yeah. And I think we've seen that in venture a little bit, right? The way people have been hyper-thesis oriented versus like, here's my thesis. It's like, great, your thesis, you're gonna run right off a cliff with that thesis.

SPEAKER_00

Yeah, I think I think anyone who's doing AI well right now is doing exactly what you're talking about, which is either they've they've they've augmented their own, like I'm thinking of like, yeah, I mean we can talk about a couple different approaches, but like an engineer, they've augmented their own specialized approach, like back to this to bring this back to the skills and d files. They've they've operationalized their own, like they've operationalized AI with their own proprietary knowledge and process that's not unique to you, that's not unique to me, that's not unique to their engineering team. They can apply it to a broader set of people if they want. But I think what's interesting about AI is like if that guy is a great systems engineer, he has the he has maybe the best process in the world for systems engineering. There might be a guy that is the best cybersecurity engineer that has a completely different process and is going to run counter to like what what the heuristics would be for the systems engineer. Probably should, right? And so, like as you move down and are solving different engineering problems, like you do not have to have the same process and protocol, and you shouldn't. And to your point, like we're in this now unique time where that hyper-specialization can occur at scale for the first time, and we're not limited by you know the resource constraint of like have having having to hire you know armies of all these specialized people. I think this point relates a lot to our industry, right? You talked a lot about specialization and the spectrum to you know, specialization versus you know, the master builder who was you know a generalist that can move between all these different scopes of work from start to end of the project. And you know, now like we can do that without like prohibitive cost again. And that's the that's like that's the big idea. And so yeah, no, I I it feels like that that's a just a common thread we keep bringing up, but that's that's I think always the goal.

SPEAKER_02

Yeah.

SPEAKER_01

No, but I think that's the point, right? Like everybody's trying to take like these standard processes and say, okay, let me you know, AI is really good at standard processes. And I'm like, is that true? Like, is that is that really what it's good at? Because I would argue a spreadsheet and you know an ERP system is pretty good at standard process.

SPEAKER_00

Software software 2.0 was was perfect for standardized processes, right?

SPEAKER_01

Right, right. And I think if you look at I I think we were talking about what was it we're talking about the other day about craft ventures, how they had SaaS so dialed in. Oh yeah, yeah, yeah. Right, metrics, ratio, like just so dialed in. And like now what? Right, like now what, right? And it was and they're they've been an engine, right? I don't I don't think David Sachs cares one way or the other. He's probably doing well with his SpaceX stock and everything else he's done. But but I do think there's something about like the end of over-specialization. I think you and I have been debating this a lot as we think about, you know, you know, new funds and new fund thesis and all that is like, do we have to be do we have to think about things in such a narrow way? Because what you know, we have to really think is like, do our first principles line up with our LPs?

SPEAKER_00

Absolutely.

SPEAKER_01

And it might be that our LPs are like, hey, like I don't need liquidity in 10 years. I I'm good. Like, I just want to like invest in great companies, right? Yeah, but that might be their the alignment might be is like, hey, just invest in great founders that are building really hard things, yeah, and we're in, right?

SPEAKER_00

That's what we can adjacently help with sometimes. It doesn't have to be every single deal. Yeah. For instance, we are very rigid on hey, we have to be able to like get in the weeds on every single deal, and it has to like we have to have you know 10 customers.

SPEAKER_01

we can introduce to the you know them two right away to have some sort of unfair advantage yeah yeah and I think we debate about ownership right we debate about that all the time yeah yeah yeah great debate it's a great you know but do we have to decide like do we have to decide like I don't know would you rather have like would you rather have a half a percent in SpaceX or 10% in uh a pre-con estimating tool spacex has an unlimited TAM we're gonna see how how it how that uh that that pitch lands at the public market it's gonna be interesting uh I I haven't dug into the s1 too much yet I've glanced through it but yeah I want to I want to spend like a full day on it and yeah and just I've always it seems it seems quite compelling. Yeah it's pretty funny like you know I I've generally opted out of like mainstream media news stuff I've kind of been opting out of podcasts lately I'm just getting really bored with it like what's happening is whatever the topic is they're all doing it right whether it's the New York Times whether it's Keris like whoever it is right across all types of stuff right SpaceX IPO yeah right and I'm like I don't I'm just gonna go read the S1 like forget all you people right so I've kind of been opting out of podcasts because I think there's just like kind of pig piling on whatever the topic du jour is regardless of like what the format of the podcast is supposed to be.

SPEAKER_00

Totally yeah everybody's talking about the same stuff it's like uh yeah not that exciting anymore so maybe we'll be the first to talk about this on a AEC oriented podcast but um that's good good transition here to yeah one of the things we wanted to talk about today which is so AI so AI companies are partnering with PE firms and management consulting groups yeah and so a you know ai companies being openai anthropic all of all of the large uh

OpenAI Moves Into Enterprise Consulting

SPEAKER_00

models are are actively trying to figure out how to apply ai to to enterprise and there's so yeah like I I think that like so one one of the more recent announcements in that on that front that I saw was OpenAI law launched the OpenAI deployment company very original name. And as a part of that so the open AI deployment company they are going to partner with large corporations large enterprises to help them dial in and figure out the right architecture for their AI systems. I thought one interesting part of that announcement was they acquired a consulting firm to stand up that initial workforce it's like 150 people ish that are consultants for deployed engineers we talked about the FDE model you send you know more or less a a tech oriented person out to the field that's like that acts as the consultant but they're actively building product while they're assigned to that to that account. So let's talk about this like because I obviously this is topical to to you but I also think in general it's an interesting question for a lot of people inside corporations listening like what's the what's the trade off what's the good and the bad with that model where you see it going right and wrong.

SPEAKER_01

Yeah so a very dear friend of mine is basically the head of one of these top five management consulting firms and he called me and we're having this exact conversation he's like KP like you're in the middle of all this nonsense like you tell me what what's going on and what what do you think we should do and I was like well you're a management consulting firm nobody believes you can code so first your strength of brand is now working against you because people believe like I can use Claude to be a management consultant I don't need you right so one you have to prove that your technical competence right so I think the acquisitions of some of these smaller firms right do we really think Accenture has technical competence wouldn't be my reason for choosing Accenture if I were to consult with them. Right or or McKinsey or Bain or any of them right that's not what we're looking at right we immediately go to Palantir right like this where we go it's like technical yeah that's a it's a good yeah good good to bring up Palantir in this discussion because they've pioneered a different model before the model companies yeah so so one they actually have brand working against them where brand has always helped them the other thing that I thought was really interesting the conversation I was having with them is like look you know we have always you know historically have been you know we bill for time right it's a labor arbitrage we provide resources we bill you for a team per month and we we do our work and there's a set of outcomes that we strive for but we get paid either way it was like you know about you know a decade or ago or longer a lot of these firms because of data science right because of data science actually they started going to risk free models where oh you want to change your turn time on some inventory we're gonna deploy our data scientists uh and actually McKinsey had a huge group of data they were hiring lots of PhDs in data data science and the idea was hey we're gonna come in and you only pay us for the outcomes right but now when we think about value stacking you know value capture in these engagements well billing by the hour least amount of value capture for the for the for the consulting firm if you start going like this risk based contingent based model much more value capture right like I talked to this other company they said that they charged a client five million dollars in an insurance company to to AIFI their underwriting process right and they generated a half a billion dollars in net new revenue not savings net new revenue because it turns out there were a lot of things that they were under underwriting either incorrectly or too slow the come in the edge cases that were hard to underwrite they just never got back to people and they were able to build AI models to underwrite edge cases that it turns out were very profitable because they are somewhat unique and custom and not routine right not a standard workflow playbook thing right so edge cases outside of the playbook turns out they were very lucrative generated half a billion dollars money extra money for these guys right so we charged them five million bucks i we got screwed you know it's kind of his point of view like we should have gotten we should have done more you know risk based right and then so now if you look you know trying to do these risk oriented engagements where you get a piece of the action that's like the second tier of value capture right well the third tier is equity right and I think where the management consulting firms are struggling they don't know how to do the equity piece right they don't know how to value it. So I think what's what you're starting to see is that the private equity firms are coming in alongside with one of these management consulting firms that's AI enabled and they're saying look we're gonna buy this company for a billion dollars we're gonna give you some warrants or options in this deal or you can write a check if you want to and then when this thing when you come do your work and you do an amazing job you get the maximum upside right the that absolute massive value capture versus like a contingency fee. And so I think those are like the three buckets that are evolving but once again like you got to have the right people and I just don't know that these firms have the right people you know I I think they they struggle they they they don't know how to get away from the you know the what we call the butts in chairs model right and I mean and I'm not picking on any one of them it's all of them right if you're accentor it's about how many warm bodies can I put you know putting warm bodies in someone's office does not make them a forward deployed engineer.

SPEAKER_00

Well yeah I mean I I it's just called staph log right it's just staph log very I mean very I'm very sympathetic to how their model potentially gets disrupted with AI embedded in these core systems because yeah I mean one the the the linear labor model just yeah butt butts and seats and the billable hour model like it's just different and that's the standard operating model that every consulting and PE firm is has run with and profited you know profited from for for for decades basically throughout human history right when when have we really broken that labor so the other thing that I thought was interesting I was talking to another friend of mine it's great to have friends and he saw the president of one of the big LLMs give a talk and ask this question like what is systems integration again like what like I don't understand like and he wasn't trying to be a jerk.

SPEAKER_01

It was just like I don't understand. Yeah doesn't everything just connect like what what are we doing right and then you know you look at companies I was commiserating with one of our another one of our friends that just deployed SAP I'm like I'm sure you're very happy because everybody's very happy about their deployment of SAP right or any ERP system for that matter. But if you think about the Accentures, Deloits, people like that that put butts in chairs to do systems integration or or infotech or tech Mahindra like all these Indian companies cognizant they live on the butts and chairs business but it's like yeah I'm gonna deploy 50 people to integrate SAP into your supply chain management system or God knows what, right? It's wild and to the president of one of these LLMs like I don't understand it. Like I don't get it. Like that's not a thing like it it can't be this hard. Like they're they clearly are not using AI or something right which is probably fair right like I got to keep people busy. So I think if you look at the three tiers of these relationships one being bill for time pure labor arbitrage then the next one being kind of these risk product risk models and then the equity models those are like the tiers of of value capture that can happen. So what does that mean? I think that means that the top talent will only go to the highest value capture models. So if you're hiring an AI consultant that's building you by the hour to help you do some transformation you're probably getting the worst talent right and then if on your particular account you're saying so like the management the management consulting group is sending the worst talent to you if you're getting a late if you're getting a flavor model. Yeah yeah they're sending you butts and chairs warm bodies right yep on the risk model you're getting the next tier of talent but you're I mean if you're

Billing Hours Risk Fees Equity Upside

SPEAKER_01

one of these big firms where is their best talent going to the ones the equity models when you say the equity models to clarify this for me so you're saying through an acquisition the the the acquiring company is giving equity to whatever whatever firm is being bought like who's giving who's transferring the equity to who yeah so just I'll just use real names because it gets weird this was no there are no facts behind this. I'm just using the names right so Blackstone partners with McKinsey Blackstone says I'm buying this pet food manufacturing company for a billion dollars hey McKinsey you're not gonna charge me but here's what the current model is I'm gonna give you a bunch of warrants and those warrants vest if you hit these milestones. I'm not gonna pay you in cash and the the elasticity of those milestones i normally operate with you know we generate on a billion dollar revenue we're generating a hundred million in EBITA so anything over a hundred million like I'll I'll convert that to some kind of equity right so that's how they're partnering because the management consulting firms can't mark the market so to speak right they're not gonna buy these firms they don't you know they can't come into a firm alone and set a value i mean they could it's it's kind of hard though right walk into a construction company and say hey what what's what are you worth they kind of don't know yeah right so I so I'm so my point is where are you gonna put your best talent where are the best firms going to go they're gonna go to these equity models not to the hourly rate models and so talk talk about how the different consulting offerings from the different groups so there's now new players in the consulting mix.

SPEAKER_00

Yep OpenAI anthropic you know maybe XAI does some consulting and then you have and then you also have Palantir which is in a different bucket how are their models different from the big from the big five well I think Palantir one they've they've focused on like a lot of government and industries right so they they can't you know you don't get to take equity in the Department of War right or the Social Security administration so they're they're clearly just a very expensive you know consulting model I don't I don't think they have a choice there based on their customer segment you know our friends run a company called Percepta they're doing a lot in healthcare and they're all ex palantier folks and their investors are AWS and Anthropic right someone by the way none of these LLMs are making a single bet on the consulting business right they're making multiple bets right they're investors in starting whatever because at the end of the day I mean what business are those guys in tokens man tokens yeah right that's all they got they're they're guaranteeing they're guaranteeing some sort of token token buying from all the firms that end up embedding their their LLM architecture into their systems.

SPEAKER_01

Yeah it's not even that different like during the telecom days during the early internet days right like we made a lot of money because we were able to drive more content over pipes and when you can clog up the pipes people would need to buy bigger pipes right and so the ATTs of the world were happy to partner with us because we were throwing all kinds of content down the pipes right yeah yeah it's it's no different right drive more consulting buy more tokens you know that's that's all they care about.

SPEAKER_00

So in that respect consulting could be a loss leader for them just to yeah acquire more tokens from that enterprise over time.

SPEAKER_01

Yeah locking into open usage tokens are forever actually it's kind of funny we're on ramp ramp should sponsor this podcast but we're on ramp and I got this email notification about our token spend and they're I guess they're they're tracking token spend now and I'm sure ramp does a good job of this people feel weird about it. I don't mind it they're probably going to publish a quarterly like average token spend amongst their customers or something which which will be super fascinating right token spend by industry and that kind of thing. So then people will feel some level of FOMO that they're not spending enough tokens I like it.

SPEAKER_00

I like it so what's your what's your prediction within this group of consultants who wins who has you know who has the advantage do you think credibility of the large of the big five do they does that you know does that buy does that buy them time to figure out you know their workforce and how to how to shift over like where do you think this ends up yeah so here's what I mean I think look during the dot com we saw this right let's take jump in the wayback machine right there was a whole segment of new and I was one of them of these new consulting companies web consulting companies right because IBM didn't know what they were doing none of the big firms knew what they were doing right so there were companies like Scient and Viant and there were all these new age and most of them got purchased by the IBMs by the Deloits by the Accentures of the world I mean if you think about Accenture has a massive digital marketing practice right so if we want to like you know just you know think about how things worked in the past I think you're gonna get some very very large kind of AI consulting transformation companies and I don't think I think these management consulting companies are going to have to buy like I I don't think they're gonna credibly get there.

SPEAKER_01

I think they're gonna have to buy them I think that's just how that's gonna work the problem though is there's not a lot of incentive to sell right there's just not a lot of incentive to sell because capital there's a lot of capital people are making a lot of money it's like sell and do what like if you're running a 200 person ai consulting company you're probably generating 250 million a year in revenue probably a million ahead and then you're probably banking 30 of that right so you're sitting there you're Nick Durham Enterprises and you're banking 50 million a year and only growing well i gotta I gotta change jobs like what you're gonna sell sell and do what yeah I mean they they they have to they'd have to offer you like several billion dollars right it they'd have to offer you so much money that like they can't you they can't afford you right and then like what and by the way sell and do what go work for Accenture sounds pretty bad. I mean I was just writing this article about like I mean because I think it's been said many ways like the next five years are like defining the next hundred right it's been said in various forms and I was just writing this article about like I think how you spend the next five years of your career is not defined by the technology it's defined by who you're working with. That's it right it's all about who you're working with and if it's everybody that's forward leaning AI

Choose Teammates Not Just Tools

SPEAKER_01

pilled whatever term whatever X term you want to put on it right I think it all comes down to who you're working with is all that matters. So choose wisely right choose wisely because I think I mean I hey like if you're a structural engineer and you're working at a firm where they're like oh I don't know about us all the AI stuff find a different firm right find a different find a different group of people to interact with and work with I mean we just you know our Slack lit up a minute ago because Opus 4.8 came out did it yeah man like our internal Slack lit up all the emojis all the memes all the everything exciting very exciting times right did that happen at a construction company today no no right so I would say like it's even less about like it's like the people man like what are what are the people around you what are they getting excited about what are they being optimistic about because I think that's part of the thing right it's like ambition and optimism is what's going to help you execute it's not gonna be how good your claw code is right I mean yeah sure get good at your claude code get your get good at using the tool like absolutely right but I think it's your ambition and optimism and and that comes from who you're around it's hard to be ambitious and optimistic by yourself.

SPEAKER_00

Yeah I like that frame I like that framing I agree I mean yeah I mean surrounding yourself with I mean we've surrounding yourself with people that might be on both sides of the fence right now I think is a really critical thing to figure out yeah like like there's a there's just a theory out there that if you're surrounding yourself with people that are AI pilled or robotics pilled or whatever pick pick your pick technology it doesn't really like nothing else really matters. Like it it'll it'll all take care of itself like you're gonna be hit you're directionally going to end up in a good spot versus uh and and and by the way like you might you might be an AI AI doomer being surrounded by AI optimists is going to be a painful existence for you. So if like you want to go build a craft business which like I think there will be some winners in a and you know human craft people that are that are that are you know doing highly specialized work like people are going to want that as well but you should pick I like I think you need to you need you need to pick your team yeah and uh you have to pick pick a direction I think being in the middle is a terrible is gonna like those are the people that are going to be in pain.

SPEAKER_01

No I I agree I mean look I mean you're not gonna get a tattoo by a robotic tattoo artist are you yeah so but you know I have friends that like get lots of I have a good friend of mine hopefully he's listening he got this one tattoo and we're at lunch his first tattoo and I don't have any ink yet because I just can never make make up my mind right but he at lunch all he's doing is talking about his experience with his tattooers like yeah like we've become friends blah blah blah you know spent two hours in the chair with him great dude we'd like a lot like I'm like you went on a date like this is what happened right I was like I guarantee you the next time I see you you're gonna have a full sleeve and the next time I saw he had a full sleeve do not fall in love with your tattoo artist that

SPEAKER_00

That can happen. That can happen.

SPEAKER_01

I think it can happen, right?

SPEAKER_00

It's easy. It's easy to get it's easy to get addicted to the tattoo cycle. I yeah, yeah. I got I've got three. It's easy.

SPEAKER_01

Right. If you like your artist too, because you have to sit there for hours, right? He's got like a full you got sitting there for hours, right? Yeah. And then if you like the person and you have plenty to chat up and time flies by, yeah, you know.

SPEAKER_00

I got my I got my first tattoo when I was living in China, short period of my life. And the tattoo studio was a group of Japanese artists that had migrated to China. Interesting. China was not, you know, there's they're not, you know, you're not known, they're not known for having great tattoo artists, but these were like world-class tattoo artists that lived that were from Japan and studied in Japan. You can imagine the artist art artisanal level of what they were doing. And the day I went to get my tattoo, there was someone getting a full body in there, like not a sleeve, like they were getting their whole body. Yeah. And then just like understanding, like we deconstructed like how they got to that experience. But I mean, they they viewed themselves as like an art exhibit, and the and these, you know, tattoo artists, like they're just painting on the canvas. It was quite a fascinating. But my point is if you're a great tattoo artist, that person fell in love with their tattoo artist. I mean, that's just what happens.

SPEAKER_01

That's what happens, right? So I'm I'm like, you know, so my point is like we've hit every topic on this episode, but I think if you're a great tattoo artist, you can opt out of all this AI robotic stuff, and you will have an amazing life and have a great to your point, like absolutely, right? I think you're right, it's the people in the middle. But I think really if specific to the AEC industry, I don't know if anyone's like tattoo artists out there in the AC industry. Maybe there's a side hustle happening. But generally speaking, it all comes down to who you're hanging out with, who are you working with, not even the company you're with, right? It's who you're working with. And I think that's the biggest driver around like what your career looks like for the next 10, 50, or 100 years.

SPEAKER_00

I think for for people on the fence, if you're not sure if you love AI, if you're not sure if you hate it, I would recommend going back and reading old AI research texts, like not white papers, but like like fame, famous physicists, you know, like from the bell, the bell labs era that were like first exploring what artificial intelligence meant and what it what it what it was. And like the similar question was like, can machines think, right? You

Can Machines Think And How We Train

SPEAKER_00

back to your your point on consciousness. I mean, like we've been debating that like like since the computer, since the invention of the computer. And it's like I I've spent some time with some of these texts recently. And I think it's like a helpful thing just for humans to like really think through. Like, do you think me, you know, can do you believe machines can think? What's your position on that? Like, like think through your own personal position, get, you know, get this, get the perspective of science of this of the scientist. And you know, there's been a lot of scientists that have taken both sides of that argument. Like, you don't have to land on the fact that you believe machines think to use potentially the best tools, but I but I do think like I I tend to observe people that are on the fence and in the middle as just not people who have not spent enough time just thinking about their own relationship with machines and how they want their how they want that interaction and that cooperation to exist.

SPEAKER_01

Yeah. I mean, well, I think that's why like I mentioned the word ambition, right? I think it's if you're not that ambitious, it's easy to end up in the middle. I think ambitious people cannot like they cannot sit in the middle for very long. But you know, we I mean it's the reason I was writing this article, by the way, is like we're we've been building as the businesses, you know, both within zero and the companies that we're buying as they're scaling, we're we're just learning that we've got to have a really good training program in place.

SPEAKER_05

Yeah.

SPEAKER_01

And so I've been spending a lot of time on that, and knowing that on our teams and spending a lot with really thinking through like what does training look like? And what's funny is we're probably spending two-thirds of our time that has nothing to do with AI on the training side. It has more to do with how do you think, how do you approach, kind of you know, back to like kind of first principle stuff, right? How do you think about like if I tell you, like, here's how you should think about building a mental model? When do people get asked that in their day job every day? Right? Like rarely. It's a big part of this stuff, right? It's a it's a new way of thinking about things beyond just like, hey, I'm really good at using Claude, right? And so I would say, even like two-thirds of the training curriculum that we're putting together in kind of a certification program, it's it's not, it's not everybody, you know. I think people are looking and go, Oh, like I have to be a coder. It's like, no, that's not that's not what this is, right? Um and then really, even like you know, word choice and framing and com and communicating to clients like how do you think about this stuff, right? So anyway, but it that's why I was writing that article because I I do think it's you just need to be around people that are like thinking about stuff, you know?

SPEAKER_00

Yeah, yeah, yeah. Wanted to give the audience a preview that we've talked about having guests on this podcast. Yeah. We mentioned it in previous episodes, and it keeps hey, it's us two again in every single, every single episode. We, but we actually are gonna act on this. We're gonna have some guests. Um, a few from a few initial guests are gonna come from the shadow ventures portfolio. A few founders from our portfolio are gonna come and hang out, and you know, we'll be a typical podcast episode, but we'll we'll tailor the conversation to the the business area they're focused

Guest Preview And Closing

SPEAKER_00

on. Some of some will be applied AI companies, some will be robotics companies, some will be robotics companies that are acting as operators in the field. So you're gonna you're gonna get some exposure to a lot of different different technologies, a lot of people, you know, thinking through architectures on the AI front and how to how to interact with systems like SAP, right? Like uh I think it's we're in a fascinating time where like we're we're having to transition across technology stacks for multiple generations. And so I think we're gonna get like even robotics to some extent that that's that's happening too. So um, we're gonna get some cool perspectives, and I think that'll start. I think the first guess is June 14th. The second, it's the second week of of June. So look out for that. Should be should be really fun, and hopefully it'll spice up uh the conversation if you're getting bored of me and KP. Yeah.

SPEAKER_01

I mean, how often, how long can you live? Stare at Nick's hair. I mean, I mean, all day. All day. All day. All right, man.

SPEAKER_00

Yes, uh sounds great. Appreciate appreciate everyone listening and the continued support. We'll we'll uh work on getting ramp as a sponsor for the next podcast. All right, we'll see ya. See ya.