(87) What World-Class AI in GTM Looks Like | Kyle Norton, CRO @ Owner.com - YouTube https://www.youtube.com/watch?v=-1w3tRUX0JI

Transcript: (00:00) Kyle Norton is the CRO at owner.com, where he's running what might be the most AI-enabled revenue team out there. In today's episode, he shares how the right tooling can completely shift the economics of your sales team. >> BDRs now talk to 20 decision-makers a day and are booking those at like a 14 to 16% rate, now the economics of a BDR are completely different. (00:24) The closed-won ARR divided by the BDR comp is like well over 10x now. But despite being heavily focused on AI, Kyle pushes back against the popular advice of simply buying your team AI subscriptions. And he shares his experience of how centralized implementations >> [music] >> have been the most powerful way to create results. (00:43) The production quality between what this applied AI leader builds [music] and what like a very AI-savvy AE would build is not like 50%, it's not 100%, it's like an order of magnitude. Plus we cover the specific area of your funnel where you should be focusing 80% of your AI efforts today. And Kyle has a reality check for leaders who claim they don't have the head count or budget to innovate with AI. (01:05) There are no excuses anymore. Your CFO, penny-pinching is not an excuse. Your CIO giving you friction about whatever is not an excuse. Like you just have to Welcome to Topline. Hello, welcome to Topline listeners. Today we have a special guest. Yes, you all know Kyle Norton of the Revenue Leadership Podcast. And of course, I'm joined by my esteemed co-host Sam Jacobs, CEO of Pavilion. (01:31) Osman Zaman, CEO of SalesTalent Agency. Welcome. Welcome, guys. How we doing this morning? Oh, this is good. I like this energy, AJ. I'm doing okay. I've got a board call in 22 minutes, so I'll be leaving a little early, but otherwise good. I got my first OpenClaw agent installed in my Slack and I talked to it. (01:51) And it's very similar to talking to my executive assistant. And I don't know where Pat stands about either. I had Claude code this morning. I was copying tokens from GitHub and I put it right into the chat. And then at the end it was like, "You need to refresh your token. Don't ever do that again. (02:12) Don't put that in the chat." Because apparently that just streams out to the world. So later in the conversation it's like, "Hey, can you give me your token again?" And I was like, "Wait a second, are you tricking me?" And I I said, "No, no, I'm not going to fall for that." And it was like, "Haha, got you." And so it's so funny, the personality Yeah. (02:31) >> [laughter] >> And then I was messing around and I was like, "Hey, Claude code, can you change the font to Wingdings?" And it just said, "No." That was its response. And I was like, "No, you're telling me no?" And it was like, "If you change the code base to Wingdings, it's completely gibberish trash. (02:47) " I was like, "Not the code, the font. Come on." I like that it has a personality. It's cracking jokes with you. Yeah. Here we are. >> You're supposed to use a .env file for tokens. Yeah, you have to do some things that I had no idea what I was doing, which is open up another terminal window and like create a URL and then secretly do that. (03:08) And I was like, "What the heck?" >> random places that novices like the four of us are putting our credit cards so that we can like spin up these instances. It's like I've got a DevOps thing with Alestio. That's where my .env file is, Kyle. And you know, somehow I gave my credit card to AWS at some point. (03:29) It is just like I was like, "I might need something from AWS." So I just I- it's funny. I'm waiting to see all the charges before I contest them. Yeah. Kyle, what are you up to these days? Trying to grow owner.com. That's that is the focus number one. And the podcast is like a fun side thing, so I get to talk to interesting people every week and learn about what's going on. (03:48) I'm a bad podcast businesser because we're poorly monetized and that that it's mostly an education project for me, but Osman's trying to make make that get a little sharper. Thinking about just things you're doing right now in terms of AI, like what's your what's top of mind for you right now? You know, what's working? Like you're you're probably running one of the most AI-enabled go-to-market teams in tech. (04:14) Like on the SMB side, I can't imagine there's another company that's like right there where you are. And so you're at the edge, which means you're figuring out what's working and also what's not working. So let's start there. Like what's actually working at this point for you? Like of all the things you've tried, which ones are giving results and which ones are like, "Nah, that sounded cool, but didn't really go anywhere. (04:38) " I'll start from the the foundations because what we're doing today and getting value from is is built on a bunch of foundation that you need as a baseline. The recommendation I give to everybody is start with data. You have to start with good first-party and third-party data. Without that, nothing really works because you can't give the models what they need to make smart decisions or or give you good outputs. (05:04) Uh do you mean like data on your ICP? So like, "Hey, who we're reaching out to?" Like contact data? Or do you mean like broader than that? Like more data than that? So there's two parts of data foundations. Third-party data and first-party data. Third-party data is having a map of your entire market to start. (05:22) So what is your what does your entire TAM look like? What is your SAM within that? And then being able to go account by account into those SAM accounts. So your serviceable obtainable market, that the companies that you think you can get on a call with and win today. Not your like aspirational fundraising TAM, but the actual customers you want to work with today. (05:45) And being able to enrich every single one of those accounts with the information that you need to decide who to talk to now and how to prioritize that market. And so that's going to be different for every business. But some of that is off-the-shelf data you could find from ZoomInfo. We sell we sell to local economy, like little restaurants, so that data is not in ZoomInfo. We use DataLane for it. (06:09) And then some of it we've built custom scrapers and enrichment flows. The prompt I give to folks, and I just had this conversation yesterday with somebody building the infrastructure, is like, "Go sit with your very best reps and ask them, when you sit down for a demo, when do you know you're like, 'I'm going to close this one. (06:28) Like this is going to be awesome.' Like what are what are those little triggers?" It's like, "Well, when I go to their website or I look on LinkedIn and I see these three things or if they have this role that they're hiring for, I know that there it's going to be like very very timely conversation." And you basically extract all of that out of your your like PMM materials, what is your ICP? And then what the reps says they want as like their perfect customer. (06:56) If every demo could look like this, what is that? And then you have to figure out what's the digital footprint of that. It's like, "Oh, I want companies that are rapidly scaling their engineering org." Okay, well that's an easy one. You can you can look at how many open roles there are on on their careers page for engineers. (07:15) Or it could be something that you're backing into a little more circuitously. So like we want to know how much volume are you getting on the third-party delivery apps? That's not readily accessible, but you can see how many orders how many reviews did that customer have? On month one, how many reviews did they have month two? That delta gives you the number of new reviews and then I can say, "All right, well 10% of people leave reviews, so I I'll take that review number divided by . (07:45) 1 and that's the number of orders that are getting placed." And so you have to build that system yourself and whatever this looks like for your business. With AI, you can you can make those decisions. And then you need the the perfect mapping of who all the contacts are in those in those places. Mobile phone numbers is like absolutely essential in today's world. (08:05) And then some detail to tell you for each of those people what is important to them in in regards to the problems that you solve. Feels like an everyday iteration, I would assume, Kyle. Yeah, and and not just everyday enriching and updating the information, but actually everyday making that scaffolding better. (08:24) So our our the first model that we built to predict how big or small this customer would be, we just completely overhauled it. We hired a new senior data scientist and this was her first big project. Update that whole model and make it better, bring in more data. And then we're we're creating new models for lead scoring. (08:43) We built a model that can predict if a lead is likely to pick up or not. Our head of applied AI built a score that tells you what leads are likely to pick up or not. And if you call high e-connect leads, they pick up at like a 2.3x rate like a normal lead, which is insane. And And TitanX does this as a product. Can you frame like the net effect of this on your go-to-market? Like once you got this data foundation in place and you've been able to do some of these things, what was the actual like numerical benefit you guys had? Well, the original BDR BDR (09:19) team that we hired would not have been able to ever do cold outbound. You know, our we our ACV is like 10 to 12K, high velocity, but our original call to decision-maker connect rate was like 3 to 4%. So if you made 100 calls, you're only talking to four decision-makers, maybe you book one opportunity a day, maybe. (09:41) And then, you know, your close rates from your ACV is like a BDR motion, just the economics won't pencil out. But if your BDRs now talk to 20 decision-makers a day and are booking those at like a 14 to 16% rate, now the economics of of are completely different. The closed one ARR divided by the BDR comp is like well over 10x now. (10:06) This idea of AI being a use case for anyone for everything, it seems like you have and your team, or you AI ops, whomever, have been able to build the AI foundational levels and then you're giving your team the BDRs, whomever, the very specific use cases of how to use it versus giving them AI and giving them open book. (10:28) Is that Is that fair to say? Cuz you have to have this like predictability in revenue. So, giving them the use cases you So, you know exactly what you expect to see and outcomes you see from those specific AI use cases. Is that Cuz we we've struggled with this where it's like, "Oh, we'll build something that's AI enabled. (10:47) " And we're like, "We want to go give this to our team." And uh Ryan, our CRO, is like, "No, you have to have really specific use cases to the team. You can't just hand this over to them." Yeah. So, so and this is the the newsletter that I wrote on my Substack then we republished through Pavilion through the top line newsletter. Everybody's interested in this. (11:05) I'm getting text messages. I got a random text message from somebody saying, "Hey, what do you know about decentralized AI?" What up? I was like, "Who is ready for me to ask?" We're shaping the conversation, guys. We're shaping the conversation. Yeah, this is by far the most popular Substack I've ever written and and I like got so many messages about it people wanting to talk. (11:27) And if you're listening and I've haven't responded, I just don't have time. Talking out. This is why This is the point of the podcast. >> Yeah. So, this was an emergent property for us. This wasn't a pre-planned strategy, but it it just sort of like came out of how we approached it. And so, there I I see there's two camps in how people are trying to bring AI into go-to-market. (11:50) One is a very decentralized approach, which is like give everybody Claude accounts, tell them to build cool stuff, encourage them to to, you know, be on the cutting edge and learn these skills themselves. Everybody in the entire company needs to be AI native. And that seems really good on paper and I was trying to push AI adoption a lot maybe 12 months ago, but what we really found in practice is that all of the best things, all of the most impactful implementations were all highly centralized. So, it was basically our VP (12:23) of BizOps, me, and our VP of RevOps for for a long time were by far the most interested and on the cutting edge of of learning this stuff and we would just come up with these ideas and and this is really like our VP of BizOps and data who is who did most of the building. And we would build something centrally and then just deploy it to the team in the experiences that they already use. (12:46) So, we we didn't like vibe code a pre-call like prep app cuz because I don't want additional surfaces. There's already too many surfaces that our reps lives in. We would just give them that information in Salesforce, in SalesLoft. Like now we've got the supply AI lead who builds all this stuff. (13:05) The production quality between what YG builds and what like a very AI savvy AI uh AE would build is not like 50%, it's not 100%, it's like an order of magnitude. What YG builds is 10, 20 times better than what the really savvy AE might build. And so, you just get a completely different result out of it and you can properly enable your whole org to adopt it and you don't have this like soup of custom gems and custom GPTs and and all of this stuff. (13:37) And we have found that that has driven like far better business results. You see this like enthusiasm. So, if you're lucky, you have people that understand that they have to play with AI and they're trying to wrestle with it. And that's fantastic. Like you have the right people. But if that's all you're doing is this decentralized thing, what you end up getting are the benefits of like a really pro-consumer. (14:02) Like it's the consumer benefits of AI that you really feel inside your businesses. Your emails get written, you know, maybe if the person has good taste, they're sending better emails, maybe they're sending worse. Like you get like the meeting notes off. You get like decks being built a bit faster. You get like certain things happening sometimes faster, sometimes better, sometimes a little bit of both. (14:25) But you don't get like real business use cases, like the ones that you can say, "Ah, this moved the needle for the business in a really substantial manner and is ready for prime time." Like that ready for prime time thing, you just don't get it. I totally agree with you. And I'm sure what Kyle said is correct. But also, one of the things I'm seeing inside my organization is you do see speeds changing. (14:50) And it's and it's it is getting strange when you have certain people who are, you know, you can call them Claude native or co-work native, who are able to process output, just regular work output, like what's the outcome >> really good example of this, by the way. Yeah. And and like the the And and it's weird because when you have the disconnect of some people that haven't they haven't adopted it personally, they are now slowing down the people that have. (15:18) And so, there's a group of people that read I forwarded the Clervo article AJ about, you know, if you can't do something in a day, you're not going to make it. And you still you feel this tendency for certain people to want to defer, you know, they think things should take weeks instead of days. And and even though I do agree with Kyle like there should be centralization so that you can build high-quality >> High-leverage roles where you're making decisions that have a broad impact. (15:44) So, if you're in RevOps enablement, if you're in in marketing, you have to become AI native just to do your job at a different speed. But when it comes building the core systems to accelerate workflow, that should be centralized. And so, I don't really need my account executives and BDRs spending a bunch of time in in Claude code and building stuff because they they are they are a role that has like a multiplicity, like there are 60 of those people in that role or 40 of the people in that role. Those are core systems (16:18) that need to be built centrally. If you are the sales enablement manager or director, you need to be able to figure out how to make your organization AI native. And it's interesting like almost everything that gets produced now, prototypes, mock-ups, like like the initial versions of things like are all Claude code artifacts. (16:41) Like in in our organization, you know, we wanted to cut a bunch of data to to make a smarter decision about how we're managing the launch organization. And so, the RevOps person pulled everything together and the output was a Claude code artifact that you you could like move the you could like pull the dragger, you could change things and it would just update. (17:01) So, it's like a better version of a spreadsheet now, a way more usable version of of like a massive spreadsheet. And that does need to become the default if you are in a role of of any leverage. So, what should you build centrally, what could be on the edges and just like, you know, vibe code and throwaway apps or artifacts? And and this extends to build versus buy. (17:23) I think like the core intelligence of the organization needs to be centrally built and managed. And I don't think you want to buy those solutions. And so, [clears throat] I was at the Clay um CRO summit on Friday and this was a big topic of conversation of like what are people building versus buying, what should I have internally versus not. (17:46) I think you have to own the intelligence, and you have to build the intelligence internally and buy things that are like workflows and user experience. And so, the the example on the sales engagement side, I haven't seen any of the sales engagement providers build a build an AI product that does any of the intelligence stuff. (18:10) Like, "Oh, this is the next best action or you should do this these things with these deals." I'm yet to see those companies produce something that is like all that viable. Curious if you guys have seen anything different. But they're a good place to inject the outputs of the intelligence layer that you built internally. (18:31) You know, the the deal health score or your estimated win rate that that prioritizes leads. Like you inject those things. Like you do the intelligence, you print that somewhere on a lead record, and then, you know, the sales engagement platforms, those are good experiences. I wouldn't try to recreate an experience because you need rock-solid stability. (18:53) You don't want to vibe code the thing that your reps use every day even though it could be a little better because you need rock-solid stability, you need the support ecosystem, you want something that's familiar to them. So, I think like it diminishes the value of the sales engagement platforms cuz it's just like a, you know, a dummy UX now. (19:11) But I think that is the way that companies should think about this this interplay. Like what do I build versus buy? You want to build internally at the the core intelligence. Yeah, the sales engagement's a really interesting point given that their data all lives inside the CRM. So, they don't really have data as a moat cuz they're not really doing the enrichment aspect of it. (19:31) But the workflow distribution part of it, I think is what is being still very undervalued in the market in terms of what SaaS really means in 2026. And it's it's this is just a a really fascinating conversation in the sense that uh for QuotaPath with and similar to to how you all are thinking about it. (19:54) You have this like central core intelligence engine and then you basically bring everything in with the MCP, which is APIs for agents. It's just API. You bring it in. And so now we have dust frames for the pre-sales demo, dust frames for the handoff net notes, dust frames for the um executive I'm going to talk to one of our customers and I have the total timeline there. (20:21) And so you kind of templatized the data and all of the use cases very specifically for the organization, but the workflow management tools we'll even call Gong as one of those. Those remain because that's that's still a really important and central part of of the workflow. And so I I completely agree there. What's interesting here is that these are three very different sorts of businesses. (20:43) So Carl, you guys have raised a lot of money, you have lots of revenue and so you have resources to invest into this. And so sometimes people will listen to you and be like, yeah, you can do it. But what about us, right? Like at an earlier stage in the life cycle. But Agent's business is, you know, at a different stage in its life cycle where they're earlier on in their revenue journey and they've been able to make a lot of headway. (21:08) So Agent, can you talk? And then there's Ours, which is Agent, you have venture funding that you can invest into this with an eye on the future. STA doesn't have that, but we have all profit that enables us to also have some sort of a centralized view on this. And so we've also stood up a centralized operation to think about the intelligence layer being a central component and building something that is actually modeled a bit off of what you've done, Carl, like a a bootstrap profitable business version of that is what we're trying to build and it's (21:40) really interesting. It's like you can do a lot more than you think and the appropriate thing for your stage. So Agent, can you speak about you built that AI ops role early on. You've done all these things. You've invested different to Carl from a magnitude perspective, but you've gotten a lot of ROI. (22:03) What has it taken? Like give give us a flavor of what it took to get what you've got. A lot of existential crisis by a founder at night. Honestly, I mean, I'm serious where I say like every day I'm having to reinvent and rethink and relearn everything that I'm doing and so is my team. And that's really an important part of it because at the end of the day, I don't think QuotaPath as a product today will be the main revenue driver for the organization in 12 months. (22:29) Um I think the product that we're we're going to release in the next 2 months will be the thing and I have to I have to communicate to my team that this innovator's dilemma is not just real, it will happen. And that's what's ultimately the most challenging part of this whole entire thing is how do I tell a team of 70 people including individual contributors of like, hey, we're going to die. (22:51) Just sorry, like that's just the reality of the situation if we don't innovate. I mean, that's all startups, but it today that's all roles. That's all responsibilities. Every company is going has to go through that. >> for STA to think that the way that we do things today, 2 years from now, 3 years from now, that's the job. That's where the imagine in in our world of all the things we do, where is the value accrual? Like what are people actually paying us for? That thing will completely change. (23:17) And so if we don't change, then we'll miss the boat and it's happening in previous platform shifts. 10 years from now, you know, I want to build a company that sustains past my career. So 10 years from now, do you think STA is making money doing what it does today? Highly unlikely. Highly unlikely. It's really strange time, right? Like strange things to think about. (23:38) I sort of want to challenge the premise of like the previous question a little bit. Like, oh well, Owners raised a lot of money, so it's different or we're bootstrapped or or whatever because Agent's got it right. If you do not If you do not become AI native in terms of the product you offer and how you operate as an organization, you will just get swallowed up. (23:58) Because [clears throat] either it's competitor like AI native competitors come and surpass you or competitors that make this pivot or incumbents that will just that will that can build the feature that is your company so easily. And I think you need to become the intelligence layer as much as as much as you can. (24:15) And I think you know, for QuotaPath differently than the sales engagement tools, there is a differentiation opportunity in terms of the intelligence that you have across all of these customers to provide unique unique insight and unique intelligence that is not just a UX. If you are just a UX for comp planning to drag make drag and drop easier and see pretty visuals, yeah, that that's not the way of the future. (24:41) But if you [clears throat] can >> Carl, the thing that you said that's so interesting what earlier on like I haven't seen a sales engagement platform that does the insights well. Now for QuotaPath we'll release Atlas and it will ultimately say, these are the three things that you need to change with your comp to change the behaviors of your team today, right now. (25:01) Quarter to quarter, what we're seeing with and I'm going to write the top line newsletter on this, AI native companies and commissions are changing their comp every quarter. There is a 250% higher chance that they will change their comp every quarter because their pricing models change every quarter. And that is the thing that is dynamically the world is changing. (25:23) >> it, Agent? Like what what variables or things are they playing around with the most? The usage and credits and outcomes based pricing is what's changing the most in those organizations. They're still trying to figure out margins in terms of what makes their business uh actually viable. >> the same way? So is the structure of the comp plan the same, but like the percentages of what they're playing around with a little bit? >> Yeah, there's there's like basically what we're seeing is that a lot of these AI native companies are going to (25:52) enterprise a lot faster and and with a CFO at an enterprise, you have to have much more more prediction around what you're going to be paying and they don't like usage based as much. >> We've also another thing we're finding is that there are a basket of AI companies that are just growing so absolutely quickly that they don't have to offer commissions at all to their sales people. (26:15) >> 71% of companies enter the year without quotas. 71% of all all SaaS companies. The the reality is AI native companies sometimes don't even have quotas. Like there and I and I've had this conversation with Matt Brailey as well where you have sales cycles that are like going from 12 months to 3 months because the the buying in the budgetary is opening immediately and they're not even being able to set quotas. (26:44) So they're setting rates and percentages of it, but not in the same >> that's the variable that's affecting whether you pay commissions or not. Whether you pay commissions or not is just dependent on one thing. Is the value of your stock options. So if I can give you stock options that it's growing at an astronomical rate. (26:59) Imagine Anthropic, right? We know people that Anthropic that have just made like the sort of money that a CRO would make off of like the one of the largest exits ever. You know, and they'll make that as an AE. Like, you know, that that's the sort of equity that they monday.com monday.com infamously did this very first. (27:17) They didn't pay commissions even as a public company. But they That was like Yeah, yeah, but their equity actually wasn't valid enough for that. So were you able to attract the best people and retain the best people? If you were monday.com with that stance, you won't. Like it would it didn't work for them as well as it does right now if you're Anthropic or OpenAI. (27:36) So there is a basket of these AI companies that have the sort of growth where they're like, listen, you'll have a really solid base, we'll give you equity and that equity is just going to grow at astronomical rates. You will be rich. Chill. And that that can work. If you are like anything below the fastest growing AI companies in the world, you can't do that. (27:54) It won't >> history. Yeah, in history. You have to be one of them. Otherwise, it doesn't work. You're going to see talent density drop. And so everybody else has variable components, but now how do you structure it? And then that's where that conversation happens. So you need to you need to reinvent yourself as a business that is a prerequisite, but you have to make these investments into the AI capabilities in order to do that. (28:19) And so people will hear me talk oh well, I can't afford to get higher like a fancy applied AI leader. And we've got, I don't know, three or four people in applied AI now. We've got a ton of open roles for that team. We're just stuffing people into that. Okay, that might be a little unique to the Owners and Vantas of of the world, but you can have one person. (28:41) You you can you can let you can find head count elsewhere. And you have to force yourself to do this. Go and and do the stack rank of your work organization and figure out who are the people that if they walked out the door today, you'd be like, oh, okay, we'll be fine. And and you're just going to have to use those folks who >> STA is able to do it. (29:04) Like if STA can do it, which is we don't have venture, we are a business that has to that has no recurring revenue. Every quarter starts at zero. Like we can find it. We were able to find it and what's the walking approach to this? And then I once you start seeing momentum with it, then you can build around. Now you're like, oh, I can add a second now because like the first one's the most important investment. (29:28) After that, it kind of like just convinces you on its own. It's so obvious. Like you you you When when you look at the projects that that team is shipping for us, it is so obvious we we will just hire people as fast as we can into those seats, no matter what. There will be no budget conversation. (29:46) However many people we can bring in to that have that type of skill set. And and let's talk about what skill sets to be in that team. But like I'm going to channel my inner Jason Lemkin on this. Like there are no excuses anymore. We've been doing this for coming up on >> die on the wise. Like that that is Yeah. Like you have to go carve out the budget and people are like "Oh, well, you know, my CFO's tight with the purse strings or blah blah blah blah. (30:09) " It's like I don't care. It's like go go find a way. Like make you know, make room in your org chart, move some people out that are your bottom performers, and spend the money on somebody who is truly world-class at this stuff. It will be the the best returning investment that you make. By by bar none. With there's this interesting question of centralized versus decentralized. (30:33) But then department versus company, right? So like do you centralize it company-wide? Do you centralize it department-wide? Like how do you think about the intelligence layer between those two components? Where do you centralize It's got to be company, Kyle, right? I don't know. But product product and go-to-market are so different. Yeah. (30:54) Okay, that's that's that's separate. Yes, that's 100% separate. I think like the the AI you're applying to how your product team works and the AI you build into your product that is a separate endeavor. And basically it's the point like if you're not an totally AI pilled engineer now, like you're you're going to lose your job imminently because everybody else around you is going to shoot >> still writing code like manually, like if let's say if 20% or more of your code is handwritten still, you are very behind the market. Like (31:26) that's where you are. Like you're kind of at that point where if you look at the 10% of the best engineers in the world, they're not writing code manually anymore. Like they they're just not. Yeah, I don't know what the specific benchmark is, but It's literally that. Like it is that. If you're doing more than 20%, you it's a problem. (31:45) You shouldn't have to. Yeah. And so let's let's just say like okay, product is over there and then and then everything market-facing from marketing all the way through to customer support is is a separate basket. I do think that that needs to be quite centralized and you need central data management and intelligence building that can serve a bunch of these purposes. (32:10) One of the other reasons that I don't want my intelligence to be in these other products. I don't want the intelligence engine to be in the sales engagement tool or be in my CS platform is I want to manage those centrally and I want everything to be able to come back to the same sets of skills or context files or prompts that that then get shared uh centrally. (32:34) So that you have >> you the variable on this in a services business? So like contrasting product versus services businesses. I think I've thought about this a lot. If you're a product business, yours ages. I think you have to separate these two things. So I think you centralize by core department of the business. (32:54) And so you say product is its own thing, this is its own thing, you're going to centralize it and operate from that. And you want to give people the chance to experiment on their own because you're going to get interesting ideas from like the edge of where people are experimenting. Your your most AI pilled whatever is going to bring ideas to the table, but you want to centralize things that go to production. (33:12) If you're a services business, you probably need to centralize centralize a lot more because you're kind of your product is the people that are going into the market and solving that problem. So if you're Accenture, if you are a law firm, like your product is the humans and it's like intertwined between product and go-to-market. (33:32) Like in the services business, those two things are highly intertwined. And so you centralize even further over that. And so that's an interesting distinction on how like you would effectively build it here versus that. How are you picking the problems that you decide to focus the resources into, right? Like you've got At this point, you probably have all these ideas of like really impactful stuff that could happen. (34:01) Um and so how do you pick between like A versus B? This is what we don't do often. Is that just taste? So the framework really needs to be identifying the most important business problems to go solve. Like people will randomly be like "Oh, like this cool thing on X or I saw I got this idea from a buddy um and or like they just thought of this idea on their own like I could use AI for that. (34:27) " And so you end up doing a lot of these quality of life like little itty-bitty things that are not all that useful versus forcing yourself to think about okay, what are the most important strategic priorities for the business? And then what are the ways to solve those business problems? And then where can I apply AI to that? And so I have this like five P's framework. (34:49) So first you want to map all the possibilities. Like what are the problems you want to solve? And what are the possible possible ways to solve that problem? What's the payoff? So if I choose this possibility versus that one, what's the If that works, what's the payoff of it? Is it a million dollars in more ARR? Is it saving 200k? What is the probability that that works out? And if you take payoff times probability, now you get expected value. (35:14) A lot of this is like a riff on Annie Duke's stuff from uh How to Decide and Thinking in Bets. Um and then you have to think about the perspiration. Like now what is the effort required to make that uh make that initiative work? And then you can then you can pick. Then you can choose what you want to do cuz you obviously want high payoff, high probability bets, possibilities that have low perspiration. (35:41) And so that you can [clears throat] either use as an actual framework or as just a heuristic. And you keep coming back to yourself and you think and you when you're thinking through problems, okay, like you know, these are the options and and like well like what's the outcome if we get that right? Well, it saves every rep like 15 minutes a day. Okay, like no. (36:00) That's not that's not that compelling. Versus and I think 80% of your AI efforts in your early innings should be pipeline focused. Pipeline is the You talk to any CRO, any founder other than the fastest growing companies in history, what's their problem? Like if they had more pipeline, things would be better. And AI is a really good pipeline >> [clears throat] >> pipeline tool. (36:25) Um and so once you get your first-party and third-party data foundations in the right place, now what are the options to go do that? Is it about picking the absolute right accounts and knowing what time to reach out to them or knowing what to say to them on a cold call or in an email or using AI to generate like compelling artifacts to send to those customers. (36:48) Like we have this AI website creator that's that's like the world's best lead magnet. Um that [clears throat] that's an example of building AI as a pipeline driver. It's really helpful. >> point, your AI initiatives as it relates to pipeline, like how often are you looking at the different channels? And are you measuring then cost of acquisition to that? Are you adjusting them on a weekly basis in the different channels? You're constantly looking at this. (37:16) As Do you feel like it's revved up more as a CRO prior to AI or is it about the same? >> Way more. We have tests that are happening every single week in terms of new ideas. And and I would say 80% of what our AI lead does is is pipe pipe gen focused. We actually had multiple tests happening last week. You know, small things. (37:41) Like what hours of the day should you be calling? And so we like built some tests to figure out okay, how do we control for dial volumes in person and and figure out exactly what times of day should we be calling? How do we structure the team so that then we put meetings like, you know, everybody has to have some meetings. (38:01) We put meetings in the parts of the day that are are low connect rate. And we really make sure that the team is like as revved up as possible for those high connect time periods. And you can just take every little piece of your funnel building motion and figure out how to how to improve that. Like time to first touch. (38:19) When somebody fills out a lead form, how fast you get to them? How do you use AI to make that better? Well, make sure that the rep doesn't have to do any research before picking up the phone and and calling. So AI does the research and fills in the information and and hand delivers to the BDR exactly what they need just the three things. (38:40) Not 20 things that you could say on this cold call. They just need like two or three. And then you can experiment like Okay, let's let's try a different a different pattern interrupt. Let's try a different offer of value. And you can start to get really intelligent about that system. One more thing like you can't do that unless your first-party data is good. (39:00) So we talked about the first-party data piece which the third-party data piece which is understanding your market, the accounts with the call contact information. First-party data is do you understand what's happening in your customer journey in a deep level of granularity so that you can do these tests? And and that's where we use momentum so that it scans every call and fills in all the fields and we can, you know, like write prompts and then backtest and backfill from previous calls to have this level of intelligence of like when people at (39:30) ask these questions, what happened? And when we give these answers or we rolled out new pricing. So we've got all of these prompts that are set up to tell us if the new pricing is working or not. You need that infrastructure as a baseline. And my I like to tell everybody to use momentum if you're like slightly bigger. (39:47) Do you still tell them, by the way? Congrats also, uh, as an advisor. But I've just I'm kind of being facetious and joking about that comment. I'm No, actually we use momentum. It's a great product, of course. I actually think the Salesforce acqui- acquisition is awesome. And and I was like a big supporter of that on both sides of the fence, actually. (40:08) Because it'll just help them get deeper into making Salesforce be like the best system possible. And they'll get a bunch of advantages now over all the other call recording tools that are that's that's going to be a good product. I went from being a bear on Salesforce 24 months ago to being like very bullish on Salesforce now, which I never thought I would say because, you know, they bought Bluebirds, Momentum, Qualified, Informatica. (40:36) They're they are all in on AI. And I think the way the ecosystem is going to move is actually advantages Salesforce as like this pivot point of where all the integrations are, where all the partners are, where governance and roles and permissions all sit. I actually think it's going to work out pretty well for them. (40:53) It's interesting because they have so much revenue. Like to move a $40 billion business, you know, if you want to grow that at 30%, 20%, like that's a lot of revenue you got to find. And so that doesn't mean like if in the public markets they aren't able to show the sort of growth that is going to be amazing in the short term for their stock, it doesn't mean that from a value provided to a CRO and a revenue ops person and hence the sales team that they can't figure out a really, really good integration of all of their stuff and hence product for the team. I think on (41:25) one side you have to be bullish, on the other side like they still probably have a tough time in the market for a bit. But it's founder-led. That's that's the key. Bennyoff can make the call and say we're all in on this. We are agent force. >> that's true for I think so many startups. (41:40) I it's going to be really interesting to see what happens to the PE companies or venture companies that have installed CEOs that are very much the predictable revenue drivers and have done it for decades because that's going to be I don't think it's true. Like I think it's like there are more companies that were founder-led that have failed than succeeded. (42:02) There's more market cap that's been created by non-founder CEOs than founder CEOs. Um, and so I think it's like there is advantages when you're the founder. You're willing to like take certain risks and bets that maybe the other person is going to be more measured about. But I think it's undeniable that on the other side you have CEOs of all sorts that can do that as well. (42:28) Like I think Satya, Sundar, like these guys at different points have shown the founder sort of mentality. So then a founder CEO says, "They're like me." But it's like, "No, you are just a really good CEO and they're a really good CEO." It's not like you're a really good CEO because you're a founder, you've just figured out how to be a really good CEO. (42:46) And that guy is doing the CEO job really well, too. It's kind of that. That's where I think it is. Just like how like the founder mode thing was like complete nonsense. It's like mm and it over-tilted. It went into like this wrong area. Like it started like Ben Horowitz spoke about this. He's like it started happening where you found companies that were actually not hiring the sort of people they needed to hire to be able to go solve the problems they needed to solve. (43:12) So founder mode became an impediment to growth within the Indrecent, uh, portfolio. And they had to push really hard to kind of break that mentality within their portfolio. That was really interesting. I still think founder mode is is is like really important. The Anything in the extreme can be detrimental. >> You're going to trigger Austin on that one, Kyle, because I know he's going to jump on this. (43:34) I think if you look at founder mode as like you want a founder who has a certain like I don't know, a certain energy to them and certain sort of proactiveness to them, sure, great. But at the same time the theory was framed. So it's like the framing of founder mode, is that effective or ineffective? I think it's like one of the most ineffective things that they they published because it's it's hard to understand and implement it correctly. (44:00) It's one of those theories that always gets you overextend when you implement it. So if if its implementation is traditionally overextended, it's not a good theory. It didn't get like understood. Yeah, that was fair. What do you think as cuz you're at the edge of this, you're noticing like as model improvement has happened, new use cases have become available. (44:23) What are you excited about in the next 12 months, um, beyond the buy fine stuff? Where you're like this is not doable today, but I think it'll be possible at some point. Co-work is a really interesting first step in this where but, you know, the last 12 months you could build a bunch of the stuff we're building now, it just was way trickier and you had to like I had to learn how to use terminal and set all this like other stuff up and and there was a pretty big learning curve. (44:52) And I think [clears throat] co-work and what Claude announced yesterday, which is sort of like their their uh, open claw competitive uh, push, like dispatch, I think it is, where you where you can have like mobile access. I think we're just going to we're going to take I don't know is here, it's just not evenly distributed yet. (45:14) I I think we already live in this where there's a lot of people that are completely living in the future and are just doing like things that are that are were once unimaginable. Like my VP of RevOps was like he's like I yesterday I accomplished what I would have done in two weeks. Like two years ago. And he sent me the list of I was like, "Dude, tell me more. (45:35) Like what did you do?" And he sent me the list of these like six things. I was like, "Wow, that is crazy." And but Steve is living in the future. He's got his open claw instance and like all of the context files and skills and it's API'd into everything. But I think bringing what he people like him and people what my my applied AI leader is doing to like many, many more people. (45:58) And I think the average employee in the next 6 to 12 months will be much more like a Steve. And and we're seeing this in our business where the people that have adopted the tools just jump out of Slack at you. Like, you know, something goes to the enablement team and then they spin up an output that is like super high quality and I'm impressed by and it was three hours later. (46:24) I was like, "This is multiple days of work." And and so those people are just so much more valuable and they do so much more and they're going to rise the ranks, uh, so rapidly that it's not that everybody's going to be like, "You're not AI native, you're fired." It's just going to be like it's so drastic the difference in output between these two people. (46:44) But I think the co-works and dispatches and, you know, somebody building like a easy-to-use open claw will come to more people and and that I think is going to unlock much more productivity and and we'll probably start seeing that in like GDP numbers over the next 12 to 24 months. And it would also expose like another group of people. (47:02) Like, you know, the the ones that that latch onto it, like that's an accelerant for them. And the ones who don't, it becomes like an exposed it exposes that person greatly as well. So like it's going to be a bit scary for a lot of people, too, as a result. I like so. >> Well, Kyle, you made me feel slightly better about what we're doing, uh, at Quarterpath. (47:24) But I still I'll still have the existential crisis tonight. It happens every night. I just go through this and like sit there and I'm like, "Oh my god, what are we doing?" Uh, right now we have Claude code and Claude bot and uh, co-work all running at the same time. And it's just like I That part of it is a mess. (47:41) And we are also trying to figure out how do you reconcile all these different context files, all these different skills. How do you give people access with the right governance and permissions to like what parts of this GitHub repo? That is an unsolved problem. But I think, AJ, your team is a such a good example of like you guys raise venture dollars, but you're not on the like you know, growth at all costs, like spend spend the most trajectory. (48:07) And you have a team that is like AI pilled and building real stuff. You do not need to be owner and spending, you know, the or burning the amount of capital that we are playing our game. You can do it in I mean, both you guys are running organizations like this. And and so I think the message to leave everybody as we wrap is like there are no excuses. (48:29) Your CFO being penny-pinching is not an excuse. Your CIO, you know, uh, giving you friction about whatever is not an excuse. Like you just have to go run through the >> being months and years is not an excuse. Like there are If you're not in a position of power, like just leave that company. (48:48) Like you're in the wrong company then. If you're not in a position of power and you're noticing that sort of like hesitation, leave. And if you are, then push hard. Like if you're in a position of power, expose that as like the thing that's going to be the thing that makes us fail. And if you're the founder CEO, fix your org. (49:06) Like you have to make this work. This has been a great episode of Topline. What I learned from it is that founder mode is very real and Austin >> [laughter] >> Yeah. Thanks Thanks, Kyle. I hope, uh, Sam's board meeting went well as he had to hop. But, uh, thanks for for joining us today. Yeah, appreciate it, guys.