Ep # 179: The Digital Sales Room Playbook: Running Complex Deals in One Shared Space ft. Carl Carell (Co-Founder, GetAccept)

Carl Carell 00:05 If you don't find value very quickly, that's the strange part, like AI has brought kind of an exploratory inspiration, authority to anyone in any organization. But if you can't translate that to something that will impact core objectives and key results, you scrap it as quickly. Intro 00:25 Welcome to Across the Funnel, where we dig into concrete Go-To-Market moves across sales, customer success, and account management so you can build revenue that lasts. Brought to you by Hyperengage and Dextego. Adil Saleh 00:41 Hey greetings everybody, this is Adil, Across the Funnel again. This year we are trying to make sure that we bring these unique stories and we try to not just talk about post-sales. We try to not just talk about product and technologies that are technology leaders. It's always about like customer acquisition. I know this has been harder than ever. Building has been easier than ever compared to distribution and acquiring customers and all. And equally important is the productivity across the sales organization and how these sales leaders, especially the account management or revenue ops, are working smarter, not harder, and doing more with less. And there's so much competition when it comes to outbound cold calling, email outreach. There's like a handful of mediums that are left to acquire customers and make sure marketing gets the product qualified leads and sales gets the qualification pipeline and opportunities up and going. So that's why we're talking to one of the co-founders at GetAccept, one of the fastest growing products helping sales, revenue, account management with the productivity, and with helping them closing more deals and getting them everything that they need across their life cycle. So thank you very much Carl for taking the time today and helping us get to know more about GetAccept. Carl Carell 02:10 Thank you, Adil. Nice to be here. Looking forward to the conversation. Adil Saleh 02:14 I love that. So, Carl, I know that this has been a pretty hot market when a lot of these AI SDRs, a lot of these AI native, what they call like AI first or AI powered platforms coming in heavily and been funded in the beginning of or mid of 2023, early 2024. And now it's even getting closer and closer. So how do you see this, I would say in this category, this huge noise. I know this has been now pretty less, a lot of investors scratching their heads, having invested in this tooling not getting the returns. But how do you see it? I know that you've been there almost a decade now and one of the, I would say one of the mature products in this category. So as a founder, like what's the vibe right now? Carl Carell 03:05 I think maybe for some context, we launched GetAccept almost exactly 10 years ago. We went through the winter batch of '16 at Y Combinator. We had demo day on the 23rd of March, 2016. And we had launched the platform in December of 2015 and that was our first customers. But I think at the time, obviously traditional Silicon Valley Y Combinator company growing fast, going through kind of seed funding and Series A and Series B, but we early on tried to kind of build AI in our product. We were very early on employing data scientists already there in 2016, 2017. We actually got one of our investors early on who was an applied AI investor only back in '16. But we couldn't really, The technology wasn't there, you know, there were no LLMs at the time. So I think what's happened now, you see a lot of AI being applied where it's easy to start with. And I think you mentioned AI SDRs, for example. They started popping up two, three years ago as kind of the first build on top of OpenAI, for example, originally. And I think a lot of these ones came out strong with a lot of big promises, but they really didn't get there. And I think particularly if you're working with B2B sales, that is a little bit more mid-market or complex, they couldn't really handle a more value-add kind of sales process or deal generation. Where maybe I think the AI SDRs as an example, are maybe more useful for something selling very low contract values and are very heavily reliant on kind of a self-service to kind of drive traffic. But we're starting to see now, of course, a lot of verticalized and much more specialized sales and marketing tools that are getting applied. GetAccept has spent a long time to not become one of these kind of AI wrapper incumbent SaaS companies. So we had a transition for us over the past two years that is going from an incumbent SaaS to an AI native platform, which we formally said we made that transition fully in Q4 of last year. Adil Saleh 05:12 Right. I know so much has changed, and I love the fact that you kept it. Like, it's so hard sometimes in this era. I know that you started 10 years back and then five years into your journey there was a huge stream of AI tuning in around, and it's so tempting as well as... It was so new. Like, I'm just trying to get in the shoes of a founder building a sales tool, in a sales function and with this AI trend and all of these tooling building wrappers and creating a lot of noise, especially in Y Combinator. We have like more than 60% of our startups on this show that are associated with Y Combinator, previously funded by Y Combinator, you know, a lot of them, right? So it's so hard to, you know, keep the focus, have a laser focus, and keep everything aligned to your product vision from day one in this category. So I gotta appreciate that. A lot of these sales organizations, we get to speak with CROs. We get to speak with teams highly focused on the sales function, like sales heavy kind of products and big sales teams, SDRs, BDRs, account executives, and all. They are so much reliant on sales intelligence, tooling like tech stack like Gong and many that you know, and a lot of these industry verticals that you mentioned, like healthcare, off the record, they're heavily reliant on sales enablement and training and all of that. So how do you fit in, how do you get the right positioning that will, If you could speak more on the positioning part, amongst the tools that have like, not on the product side, but they have a bigger cap in the market that they've captured in the sales function and how to strategically position the platform or how big of a challenge that was. Carl Carell 07:06 Yeah. I think positioning is extremely important, right? And that's it. We iterate on that at GetAccept over the years and it's changing all the time. I think that's kind of a living organism. But you're mentioning some of kind of the giants in the sales engagement, sales acceleration, et cetera. The sales tools overall. So like the Gongs, the Saleslofts, now Apollo for example, you have a lot of those ones who started out more in conversational intelligence or kind of sequencing tool. Outreach is also another one. They obviously, since they hit a certain, 100 million, 20 million ARR, they need to kind of broaden their scope and you've seen acquisitions or they're going into each other. I do think for example now it's much more commoditized with transcripts. Transcripts are, I think, the key to run a really effective internal AI-first organization, because that's exactly what customers are saying. That's where you find the data, right? So that's extremely important. But pretty much every tool that you find today, it's much more built for internal use. So if you think about the CRM, the UI is for the AE, for the sales leaders. The same thing with Gong. Yeah, sure, they wanna send some automated emails and sequences, but you know, it's run on emails and, you know, at the top, some PDFs, et cetera. Our vision was that there's so many different tools, and this is what we pitched to Y Combinator back then, that focus on creating deal generation and CRM systems, It's a very mature market, but you're building very little for helping account executives and account managers to approach the buying committee and work together in one space where everything you need to kind of go from discovery all the way to signed deal and renewal and upsell. And that's what we wanted to build. We wanted to build a UI for collaborating between buyers and sellers in one space, because that doesn't exist even today. It's extremely fragmented. If you think about it, people are sitting in their own LLMs. If the company doesn't buy them ChatGPT, they may use a free version or buy it themselves. You have LinkedIn for doing stakeholder research. You have PDFs. If you're a little bit advanced, maybe you have some kind of content management where you share links, et cetera. But if you think about a little bit more mid-market complex sales cycle over a couple of months, many meetings, maybe 5, 10, 20, 30 stakeholders. How do you project manage that as a seller, to ensure that you provide an easy buying process? So it's easy to buy from you and feel like you're actually buying, not getting sold to. That is kind of the positioning we wanted to do. So how we position GetAccept is that we are an AI digital sales room that brings everything you need to work smarter, close faster, and win more together with your clients in one experience, one link, one microsite for that opportunity or that partnership that you're exploring. So we have positioned ourselves at the perfect partner to conversational intelligence, to CRMs, where we actually bridge the layer of all these data internally to translate that to executive-ready content like a discovery, case studies, and hyper-personalized content that address every single pain point that you have, data in the CRM or what exactly did the CFO say about the investment and what's important for the company and how they're evaluating. We're trying to take that, so instead of spending hours on follow-up to really craft this as the best enterprise sellers do, anyone can do that in SMB mid-market to enterprise in the matter of seconds after every call and interaction and keep this updated. And it doesn't matter if you are a person who's not hitting quota or you're a top performer crushing quota or anywhere in between. Everybody will produce the ideal kind of buying experience and sales process externally towards your customers. So that's our positioning when it comes to that. And our AI, it is positioned to really understand the whole picture. And I think that's really important where we understand what's happening in the CRM. We understand the customer you're selling to based on AI intelligence, but we also have every single interaction and stakeholder engagement and what they say in those transcripts to kind of build that. And then we train our AI knowledge on one of our customers to really understand how does their internal sales process look like? What are their best customers? If they use MEDDPICC, what are the M ones? And we triangulate all this to kind of create a very repeatable process. So it doesn't matter if you're 5 or 10 people using it, or 500 or 5,000, all those sellers will produce that ideal buying experience for every single client no matter if they know how to do it or not. So that's in the essence what we do. Adil Saleh 11:44 Interesting. And thinking about these sales rooms, I know that context is super important across the function and when it comes to revenue ops, it is even more critical when, you know, having, hitting and understanding those external signals that are more outside of the conversations. You know, maybe their budgets, their hiring, raising funds, their growth, so many things that are coming outside of the organization. So I know having a 360 view for rev ops is super important and a contextual window. So you mentioned the AI part. Today everybody's trying to build the AI for their customers and have them tune it better, where it's taking around 30 to 60 days in the first, you know, helping them grab some content, context and knowledge base and based off of their intelligence. So, in terms of cycle, because this has been the biggest challenge, like a lot of these AI copilots, or I'm talking about the copilots that are built within platforms like yourself. I'm familiar with GetAccept AI you're mentioning, and it must be super cool. So now thinking about for a customer from day one to, in one industry vertical for that customer, how long does it take for your AI to be a hundred percent accurately specialized? Capabilities in terms of context, to make it repeatable, to make it not miss it in terms of the responses and so many other things you're mentioning, the presentation or decks or success plans or executive summaries. And there are so many things that an account manager can do. So how long is that process? Could you also talk more about that process? Carl Carell 13:28 Yeah, and I think you're touching upon something that's very important. How time to value is extremely important in a day like this, and particularly in the world of AI. I think previously before we built our native AI platform to kind of backup the entirety of our traditional SaaS platform, it took quite long to kind of navigate and set up the content library to be able to move fast through our product. The interesting part now is we've spent a lot and I hired a sales enablement director who's very skilled at what we do, and we really worked to make the kind of the baseline of our AI knowledge base to understand if a company uses MEDDPICC, okay, great, then you adapt like this. So we're not reliant on a client to sit and train our AI in any way. We have out-of-the-box maybe 90% efficiency from the get-go. What we need is the internal kind of sales process documents. We can scrape G2 Crowd for reviews, for example, and triangulate them into sales content. But I would say it takes a couple of hours to kind of get all the content into the knowledge base. We have configured GetAccept to understand the knowledge base and we know best practice on how to deliver it. So you can actually within a couple of hours, I can outperform follow-up for any AE at our customers with my platform versus them not using it because our AI can grasp a much broader picture as long as I get the transcript. So what we do quite a lot of times in our sales process, we ask for a sales transcript from a discovery call from someone. Ask them to send on how they followed up that one and how fast they were able to kind of build an executive summary or a case study to kind of move that forward. We can quite quickly, I mean in a matter of hours after we're set up, I can produce a better one in the call together with a CRO. I can tell him, hey, look what I was able to create after the same call that your rep has. This is how he or she followed up, and this is how I can follow up. And by the way, I know nothing about who you are and what you do. Our AI has figured that out for you. Imagine what happens if we spend a little bit more time configuring this together. So for us, time to value is extremely important because I also judge platforms in the same way. I sit and experiment a lot. I sign up for a lot of free trials, and I have my own custom GPT. I sit in Claude, et cetera, to kind of try things. And if you don't find value very quickly, that's the strange part. Like AI has brought kind of an exploratory inspiration, authority to anyone in any organization. But if you can't translate that to something that will impact core objectives and key results, you scrap it as quickly. And we talked about before we jumped on here, the importance of post-sales, right? Like what you guys do in your platform. And that becomes even more important. We've seen a trend, and I think you're going to see a significant continued trend with first-year churn for people signing one-year contracts as AI companies, because a lot of people are willing to take the bet because people are signing budget for AI investments, but they're much quicker than the historically traditional B2B SaaS to cut them right. So I think there's something you have to think about. How do you adapt your maybe sales process that previously was allowed to run faster and onboarding? You probably have to cut that by 80% to get to what the customer expects today. Adil Saleh 16:56 You hit the nail. Absolutely. So this is where it's going. Like a lot of these AI investments and to your point of qualitative data sources, this one part I wanna serve as well. All these qualitative data sources and qualitative conversations, like meeting notes, like a lot of these platforms that we use, we are also doing our own note taker because it's so hard to adopt people to, you know, or get them to upload their meeting transcripts and everything. So we built our own note taker so they can go inside their calls from sales to, you know, every call like QBRs and all, because we are serving post-sales as well. So are you also thinking about building your own note taker that can enable those people? Because this could be a turnkey, let's say an account manager can simply go in, sign up for free, you know, get like Fathom or Otter. They just go in and use your dashboard for all these post-call nudges and summaries in just no time. This is just my thought on this. Carl Carell 18:00 You know, man, I think you're right. I think we looked at it a bit differently. We want to be the best partner to kind of sales AI, conversational intelligence and revenue. So think the Gongs, the Saleslofts, Fathom. We have Glyphic, another partner. We have local ones in different markets, but because we've got international, we integrate all of the ones I've mentioned. And I know for a fact that the vast majority of companies are building on another AI SDK that comes out of Y Combinator called Recall, right? So as I'm assuming a lot of these ones are using the same thing. For us, it doesn't really matter as long as we can get the transcripts. And the interesting part, we also integrate directly to Zoom, like we're on right now recording. We integrate directly to Google Meet for the transcripts from Gemini. And we also do it directly to Copilot in Microsoft. So it doesn't really matter if you purchase it, but I do think there's a different use case for those conversational intelligence tools that we don't want to build for, that you still should purchase them. And I do think the sales-specific ones are much better than the non-sales, the generic ones that a lot of people buy, right? So I'm much more in pro, for example, of platforms like Gong, Glyphic, Garbo, a local one here, or any one of Salesloft for example, because I think they're built to be more sales-specific and I think that's better for the end user. Not everyone has a large revenue operations organization to kind of manage your platforms and the more they're purpose-built for something, I think that's beneficial. And that's the cool part with GetAccept. You don't need to understand AI. You don't have to train your reps on AI to use GetAccept. We have built it out of the box to be tailored for AEs and account managers to just work. No prompt engineering needed for the rep. And I think that's a key thing that I'm coming back to with a lot of how you build AI. I think today you have that, like we talked about before, kind of that authority to explore in the AI world. I don't want my reps to sit in different AI platforms where the potential of a human error and how they prompt, what knowledge base that pulls from GetAccept, we're trying to deliver something where reps don't need to prompt and you control as a sales leader and a rev ops organization or enablement organization, you control repeatable outcomes to your client because I don't want one of my 50 sellers to communicate differently from the second to the third, to the fourth to the fifth. So if I get an inbound, it's essentially a 50-person round robin and they can get 50 different experiences depending on who they talk to. So I think that's a key thing for solving if you're a revenue leader or Go-To-Market leader is, with all these AI tools, how do you ensure consistency according to your methodology and what you know wins or loses deals, rather than kind of giving this big opportunity of the vast universe of AI with as many different outcomes. I don't believe that's how you grow. I think that is a giant risk. A lot of revenue leaders are coming to the realization, oh damn, I need to tackle this because currently it's a little bit of you're spinning the wheel and hoping for the best with all the AI tools people have implemented and allow people to use. Adil Saleh 21:12 Yeah, I mean this becomes really interesting, of making sure that how you can trust AI in mission-critical environments such as healthcare, like a lot of these engagements that are economic engagements, like financial analysts are banging their heads a lot as well, because they were told more than a year back that their job is going to be automated. A lot of these law firms are struggling to make sure that how they can build systems to have the right check, a human loop that knows better than AI to get the best responses or review. So now on this topic, I know that we have done almost 150 surveys with GTM leaders. That means these companies from a scale of Slack, Salesloft, Gong, to these Y Combinator startups from 2020 to 2021 and beyond. What we have found is more than 35% of these customers, they churn between the onboarding to adoption stage. You know, there's so much of customer education involved, a lot of I would say lack of success metric definition. Like they have not figured out what is what they call a successful adoption. Like what kind of steps or what kind of usage metrics, what kind of other milestones that they have set so to ensure they're adopted, that is going to drive the retention. So this has been late 2025 or last quarter 2025. And this survey, this has become so horrifying and it's going to be even more now because there's so much noise. People are trying to solve the problem first, but they are not thinking so much about the experience side of things like customer experience, user experience. They're just enabling AI to the fullest potential to solve the problem. So how do you see this problem and how do you mitigate it internally for GetAccept? I know that you have a pretty decent-sized post-sales team and you know about data and you're doing it for sales organizations. So how do you internally measure it to a point that you can increase the retention part and of course the net revenue retention is the biggest talk of the time. Carl Carell 23:30 Yeah, no, I think, and just to show you some, we're very metric-driven as well. So when we look at the win-loss analysis, is that when people buy us, and one of the main reasons is our native AI, win rates are significantly higher when it's not important. So we've seen a lot of companies burning themselves. So we saw in our latest report internally from all of our thousands of conversations we had throughout Q4 and into January pretty much, and, you know, a little bit into February. That's the data set we looked at. So I'm pretty sure it’s the timeframe, but still statistically significant, is that software companies and IT and tech, the broader category of professional services and healthcare are the ones that we see are still optimistic about AI. There's a lot of more traditional industries that have tried it, got them burnt, like you say, and churned on platforms. And for us, I think it comes down to the time to value discussion that I mentioned earlier, is that you have to allow for much faster value. So how do you design to maybe then implement 80%, 70% of your product that you know will give them value from day one, rather than trying to run these larger traditional B2B SaaS onboarding projects. One part that we do internally, so we have two stories we're trying to tell to ourselves, to investors, and to everyone is that we're a native AI company internally, how we operate, you know, across all of our functions. And we're a native AI software platform as well for B2B. On the other side, like what we're trying to do to kind of be native AI in the handover is we do the same thing there. We look at the actual transcripts. So instead of relying on an AE writing up the handover, we actually draft a new digital sales room for the account management piece based on these details, the same way we would write up a summary for a person making a decision on investing in GetAccept. So we're trying to use our own product, and we patch that together with some Claude and HubSpot CRM data and external data. And just ensuring and we're still doing handover meetings, AE plus AM plus CSM. And the CSM is always part before a customer signs on to actually hear firsthand and present on how it’s gonna onboard for several reasons, you build trust. I think still AI products should help sellers to be more personable, build better trust, and seem like a trusted partner, and I think that helps doing that. The other part of that as well is if that person actually is part of designing the onboarding process rather than the AE, there's full commitment to that. There's never gonna be that conflict between an account executive and a CSM that quite often happens. Oh, by the way, you promised stuff that we can't do or we shouldn't do, et cetera. And, you know, there's a lot of good comedians in sales that talk about, oh, you have such a fantastic this. Yeah, of course we have that. And of course... No, I think that mitigates that bit. And to be honest, like I've never seen an issue where we've said to a client, no, we can't do that, but this is what we suggest instead, that actually builds trust, particularly when they work with some of our competitors who say, oh, we can do everything and this and this, and we can solve it. I don't think that's trustworthy. So I think that part, like using AI to really triangulate all information to help the CSM present and also ensure that the customer kind of signs off, yes, this is what we bought, this is what we want to see. So if we achieve X, Y, and Z in the onboarding and activation, are you happy? Because you also should run that as a commit in the same way you say, hey, if I can secure you this price and we can fix these things, are you going to sign and become a customer? You should run the same process for the activation process. If your onboarding is, you know, four weeks, eight weeks, twelve weeks, whatever it is, what are those milestones? And I want in writing, checkbox that they say, hey, yes, we're gonna be happy because then we can go back. We've actually delivered on everything that you said means you're happy with the platform, and I see so many companies missing out on this. They run this elaborate onboarding process, but it's never really been asked. The question is is this what you want at the end of the day? And again, how do you take that back to if you built a business case where the CFO signed off for the CEO? Is this helping to solve those critical key results or objectives and challenges or not? And I think it's still as important to the AM and CSM team to continue to kind of engage the C-suite of your buyer. Partly, can we upsell, can we expand? Can we cross-sell? But also just to understand the health of the accounts. Because first-year churn hasn't been an issue as much as it's been in the past, but you're seeing that trend, like you said, going up. And I think people need to invest more into that part of your customer lifecycle journey. Adil Saleh 28:30 Yeah. And this is like, I love the way that, you know, you're keeping your post-sales hand in hand with sales when it comes to going into the conversations, understanding the context, because a lot of this gets siloed later on. Like all the goals that customers share during the sales calls, they're going to never change. Not anytime soon. So to be able to evolve those goals across the journey of the customer has become super important. And then of course, articulating it really fast to deliver the value, that becomes another challenge. So now thinking about all this data and you talk about post-sales and specifically customer success side of things, you know, from the point of onboarding to of course scaling with them and helping them with, you know, multiple modules and potential of the features. I'm not talking about it so much from a feature-specific angle, but I know that you have different modules. So how you are measuring all of that product usage or interactions to a point where you can call them, hey, three, four months to six months, hey, this is the power user. This account goes on for another year. We are sensing a strong signal of opportunity to expand or maybe get a referral or all of that. Like how is that being systemized internally at GetAccept for CSMs, and how is this articulating to the next best action, which is of course driving communication and organization, which you're doing it for sales organizations like getting the context and all this meeting conversation, analyzing it, and then giving the next course of action to customer success. So how is this internalized, operationalized in a way that is hitting the revenue and expansion part of your business? Carl Carell 30:15 Yeah, no, I think this comes in, we invested very early into rev ops, and I think that's a critical function for any company. I think everyone I talk to always hires rev ops too late, you know, 12, 13, 14 sellers in, you probably should have it at 6, 7, 8, right? Someone who knows your process. But that's a different story, but just a lesson I've learned that you should invest much earlier into rev ops. But for us, we actually, we consolidated. We did kind of an AI tech review of our internal tech stack and we wanted to make sure, do we have a tech stack? And we did that about a year and a half ago. Like, do we have the tools that we think will work in two to three years' time and how do we invest? Do they integrate with each other in a way where we can work effectively? So part of that is we moved from a CS, a customer success platform, actually into HubSpot. So we started using HubSpot for everything when it comes to customer success and so forth. And part of that, what we do is to kind of look at these signals. We feed all the transcripts and all the data into HubSpot. Everything is in that one system of record as kind of a "data lake" of everything. We use Claude integrated to HubSpot Breeze instead of using their internal to then run things. And then we have n8n on the other end to kind of do some of the agentic workflows. So we combine that and of course that's maintained by my rev ops team, my director of rev ops. And that kind of identifies and triangulates all the different data points from external information to search about the customers, which Claude is excellent at doing, I think much better than OpenAI. And it's much more precise I think. One of the things we noticed, first we started with OpenAI and you'd really have to prompt-engineer it to get realistic answers. Adil Saleh 32:05 Absolutely. A thousand percent**.** Carl Carell 32:08 Where I think Claude is much better for sales use cases because it's much more binary in the sense that it gives output. So this kind of changed for us because we got all the data into one place. We could patch things together with our own platform, together with call AI recording, together with HubSpot, together with Claude. And it kind of just made a better world for us. So that's how it works. And then of course it creates actions and et cetera, according to certain frameworks directly in HubSpot for people. But also having everything visible on the customer account that AEs, AMs, and CSMs see the same view. In the past, we had a CS system where only the CS and account managers were in, but not the AEs because we didn't want to pay for them. So we tried to push some of it back to HubSpot. And it didn't become as direct. So now we're much more proactive and of course we integrate every Pendo user data into this data set as well. Anything we get from our own platform, and of course anything that we can get from our support, from Intercom, et cetera. So like you're getting a very good picture. And instead of having decision-making done by reps, we make the decision with AI and feed them the critical information to then refine that decision on how to act essentially. Adil Saleh 33:25 Yes. I love that. I mean, I was also on the consumer side of things. While you were explaining all of that, I was looking at the vendor side of things a lot. Like how a lot of these sales functions and post-sales functions, including the rev ops that is kind of a conjunction or analog of both, they're working in so close-knit and tooling, let's say CRMs are trying their best to make sure they keep them, but again, at the number one. I was just talking with the C-suite at HubSpot long year back, like Jonathan Corbin. He ended up building his own GTM tooling, Maven AGI. And there was an interesting conversation that how it can do the same thing. I know what HubSpot does and what price it does it. And it doesn't do it. That's why you're using Claude and all these connectors to do it. It is so hard to measure health when it comes to post-sales. And it has become slightly easier because with this AI, with the longer contextual window, as you mentioned, Claude is better at, and the reasoning window is better. It becomes easy, but still people are finding it hard to get one tooling that works with your CRM beat anything and serves for your account executive or your revenue team and your purely customer success teams or account management team that are working across the lifecycle of the customer, helping them with the journey and whatever it takes. And then signals, opportunities, and of course, stay on top of the churn and things. So that is why, like this was the whole reason, we ended up not being a CSP tool like inSided and a lot of other tools, but the work with them to do all of this, not like a revenue ops is using a different tooling with HubSpot, or account executives using a different tooling for HubSpot. And then there is account managers that have different goals and assignments and different data matters to them. Different signals matter to them. So that's why I said it has to be more modular. And this is the best time to have it with all these capabilities of AI. Now the cost is gonna be a bigger thing. The API costs and all. So how do you manage that? This was also a bigger part of the conversation we had recently. I'm not gonna take you more than five, seven minutes, but just shed more light on the cost side of things in API. And you're talking about LLMs, are you also looking out for some of the Chinese versions that are coming at significantly lower cost? There was a small air that came for five or six weeks about DeepSeek back in the days, like not long ago, but then it disappeared. So how do you see it cost-wise at scale? Carl Carell 36:04 Yeah. Internally from both how we build our own product, we really rely a lot on Claude, both internally and in our own product. We've tried different combinations, but I would say Claude is, I think, really well suited for sales and revenue. And just because, as I said before, I don't want tooling that gives me the answer I want to hear. I want to get the tooling I need to hear. And I think that's how you should reason. And when it comes to cost, it depends on what you do. What we do with our AI is not necessarily a giant cost driver because we're not, to be honest, when you think about B2B sales, what do you need to produce for people? You need to produce short-format text and bullet points for executives with budget who can sign off. They don't want giant branded experiences, a lot of them. They want one page. What's the challenge? How do you solve it? What's the cost and what's the ROI? And what's the timeline and how much investment do we need to make on our end from our team to make this happen, et cetera. And that's about it essentially. So we also want to have the best outcomes. As i said, we are never gonna optimize our AI internally to buy access to a lesser LLM or model because I want our team to be effective. I'd rather have a team that's operating with less people at a higher capacity than buying a little bit cheaper, however, a little bit more manpower. We want to have as high an ARR per FTE as possible internally, and we want to help our clients to do exactly the same. So we work with high quality, high capacity people to get a significant hockey stick leverage on those individuals rather than trying just to do small incrementals that are fighting against each other. I don't know what you see from people you talk to, but I haven't seen anyone else in this space trying to optimize only for cost if they don't have giant cost margins when it comes to LLMs. And I think to be honest, like, even if you have today, as long as you can secure a runway and et cetera, and I don't have issues with that, all the models you're using today will be cheaper tomorrow. We know that, right? So people are Adil Saleh 38:17 They’re getting cheaper tomorrow. They're getting cheaper and cheaper now. I was just reading up last week's Claude's article about Sonnet and how they've improved their efficency close to old Opus. And Opus was freaking expensive. We were using it for different functions and now Sonnet is as good and it's 40% less, cheaper. And then the best part is all these competitive LLMs, especially Google, like Gemini, that's not their core business model, right? Like OpenAI, like ChatGPT is their business model. Google is doing a million other things, right? So Google can any day knock down on the price, and then they have to be market competitive to lower the price. And there comes like, models that we need for us, getting cheaper, as you mentioned. Love it. It was really nice meeting you at first and having all this acumen of a founder and getting to learn from you and your industry was no less than a pleasure and let's keep in touch and have more of these conversations and learn from each other because you don't live long enough to do all these experiences yourself. Carl Carell 39:18 No, no, of course. I mean, we learn from each other and we help each other. That's how we get better. Adil Saleh 39:22 Yeah. Thank you very much. Carl Carell 39:23 Fantastic. Thank you Adil Adil Saleh 39:24 That's so inspiring. Appreciate it. Have a good rest of your day, brother. Outro 39:27 Thank you very much for listening to Across the Funnel. If you got one useful Go-To-Market idea out of this show today, please share this with a teammate and hit follow. Explore Hyperengage at hyperengage.io and Dextego at dextego.com.