How to Scale Go-to-Market with AI – A Conversation with Jared Brickman (Insight Partners)

Foreign welcome everyone. Who is. Who's joining. We'll give it a few minutes for people to enter. I put in a three question poll here.

I think I've seen a few familiar names in the crowd here. That's the same Christian Horth with whom I, I worked with at snapdocs. Susan Peach. I'm eager to see the results come back on the, on the high traction use cases. It was a, it was a good list. Kind of represented the prioritization we tend to use with our port coast.

All right, so I'd say we get going here. I'm going to end the poll here. We'll see what we got. Last chance to what do we got here? Okay.

All right.

52% are you of you are applying AI in limited in a limited set of workflows. That's pretty good.

Very few.

You know there's a lot of posts online about my org chart. I got these AI agents. You know, I have 100 people in my marketing team but 93 of them are AI agents. But we're not seeing a ton of that yet in this group.

Yeah. And oh by the way, I'm a company that sells AI agents. Come by myself. I've seen a couple of those. Exactly.

And then where are your most advanced uses in go to market engineering workoflows. Little typo there. Okay. Content gen. Does this surprise you, Jared?

Pretty straight ahead. It very much reflects what we tend to see in the portfolio. Agree with the so number one, the maturity curve. Like yeah, I mean mostly in the majority phase, the AI agents very much in the pioneering phase. I mean a few outliers who are bringing it into their org chart Content gen is well worn. I mean AI co writers were around before chatgpt so I'm glad to see that that's continuing to lead. I'm also glad to see the balance between inbound and cold. Right. Inbound, a little bit more routinized, a little bit more narrow that's easier to apply. Cold refining in the portfolio is indeed perhaps a bit too flexible and really quite frankly like it still requires picking up the phone to book a meeting. And then finally like my reaction to the tool landscape, I mean I hear you. Y' all like it is. It is. It is fragmented out there. And maybe the most difficult question that I get from our portfolio companies is like which tool should I use for X? And it's shitty. But I always have to say that's a big depends. Right. And we got to think through that. So excited to talk more with you about that today, Jared.

Cool, cool, cool. All right, well, I'll do the sort of formal intro here since we probably have a good number of the attendees. So everyone, thanks for joining. We're here to talk about something that I know is on every go to market leader's mind right now, which is how to scale go to market with AI. And people don't want to just dabble in the tools. This is what we've been doing. But we actually want to drive revenue, want to improve efficiency, want to get measurable results. And actually that discussion about measurable results and proving this was what really drove me to want to invite Jared Brickman today based on the conversations I've had with him in the past. Jared Brickman is the Vice President of Insight Partners center of Excellence and their operating advisor on all things AI. That's a pretty important role. He's earned his MBA from Duke University. He has been a guest lecturer at Duke, at Syracuse, a half a dozen other great schools and then I know. Prior to joining Insight, Jared led ABM at FinTech, unicorn snapdocs at Insight, Jared has worked hands on with almost half of the 550 insight portfolio companies, helping them embed AI into their sales, their marketing and their customer success workflows in ways that actually move the needle. So Jared brings a rare mix of strategic, clarity and tactical know how here and we're really, really, really lucky to have him here today and to learn from him. We're also joined by students from the Go to Market Engineer School which Matteo is on the call. Matteo and I launched earlier this year. It's a hands on cohort based program that trains operators to build modern go to market systems using AI and automation, including outbound workflows, lead enrichment, signal routing and more using tools like Clay, Octave, Air Ops and a bunch of others. Many of those students are here today. I know some have even accepted new jobs since starting the course. Congrats. And I'm thrilled to have them in the in the zoom as well as all the participants today from Insight Partners Portcos. Welcome to. So here's how we'll run this. I'll kick this off with a few questions to Jared and then we'll open it up to all of you. So if you have a question, drop it in the Q and A box. Can you, can you access that? Is that I don't know if I need to enable that. And then we'll get to as many as we can towards as many of your questions as we can towards the end. So Jared, thanks for being Here, let's dive in.

Thanks for having me.

All right. Oh yeah. Okay, so you can access the Q and A. Thanks, James. Okay, so Jared, I know you've worked with more than 150 insight portfolio companies. It's amazing what's been the biggest misconception that go to market leaders have about AI today?

Sure. The number one misconception is I can buy a tool for my team, Whether it be ChatGPT or Microsoft Copilot, and I can delegate that down to the very frontline personnel and have those folks deliver for me transformational performance changes. I think a lot of us are inundated by these extraordinary case studies where you see a company say fully autonomize their inbound lead flow and achieve a step change in the meetings that they book and the cost of those meetings. Rarely do we find that those types of transformative changes come from a kind of diffuse and generalized effort. At this point, more than 85% of our portfolio is using some form of generative AI in their go to market. But very often it is like we just bought them ChatGPT and everybody's using at the desktop what we like to call staff coordinated type of work is when you shift to system coordinated work. And I know we're going to talk more about this where you have kind of a business process that you're keying in on, one that's tied to a core value lever of your business, whether it be pipeline or retention. And you make it so the staff only have to show up to the system and engage with the system in, in a kind of very defined way that you start to see really strong scalable results out of the application. So core misconception is I can just buy this and distribute it and I can see innovation thrive. It really needs to be very focused and deliberate.

So it sounds like it's not really about I want everybody to just be chatting with ChatGPT and Claude all day long. It's about finding specific processes, picking some KPIs and then building a system for those processes.

Yeah, that's exactly right. I mean look like non controversial to give Everybody access to ChatGPT and that desktop productivity is meaningful. But don't let that get in the way of, let that be accretive or complementary to a much more deliberate and focused approach.

Okay, that makes sense. When we've chatted before, you mentioned that insight has moved from, I think what you said, prompt training to recipes and then deep interventions. So maybe you touch on this a little bit. But what triggered that shift and what does a recipe look like in practice?

Yeah, sure. I mean, frankly, it's exactly what we were just talking about. I mean, in the early stages, we had this Cambrian explosion of generative tools and capabilities. Not a lot of people knew what to do with it. Right. This is vintage 2023. You know, there's lots of prompts flying around out there. And so, you know, we tried to get ahead of that and get very organized and say, hey, here's a book of prompts that run across the entire revenue bow tie from demand generation on through to renewals. And that came from sourcing it from within the portfolio, having a strong perspective on what a strong prompt looks like. And we distributed that, we did prompt trainings, and that was wonderful. And again, you know, that got everybody very excited about AI, but it, it kind of diffused the energy a little bit. Right. We didn't see that then really translate into more than, you know, modest gains in productivity. As time went on, some pioneer outliers started to emerge. Right. Which is this benefit that we have as an insight. Right. We have more than 550 portfolios. It's like a, it's an extraordinary vantage point to see, all right, well, what are those outliers doing? How are they getting the step change in performance? And what we came to understand, and this was kind of our 2024 focus, was this scalability gap, right. On the far side of the scalability gap, where the pioneers were, they had applied structure, they had focused on a particular use case, a particular value lever. They were relatively agnostic as to whether they were using generative conventional automation or both together. On the other side of the scalability gap, on the unfortunate side where people are kind of stuck, you had people just with that generic desktop use. And so we then oriented to say, okay, let's take these pioneers and what they're doing well, capture them in case studies and recipes that we can replicate across the portfolio. And just like we would with sales compensation guidance or customer health scoring guidance, we turned it into a playbook and then it disseminated that. But beyond the playbook, I am on a specialized team that is going deep with some, some of our key portfolio companies in driving like a transformational change to bring them from maybe pre scalability gap to the far end of performance.

And are you and your team building that for them and then training them on that? Or is it more of a advisory? Is it more of a back and forth like that?

That's a great question. What we've come to find is that there's two ways to work with portfolio companies, and it depends on, you know, their orientation, how technical they are and what they need. One of those avenues is through very classical business process management, right? I don't have to tell the GTM engineering school folks, like, if, if your process isn't nailed down, if you don't have some sense of the inputs, processes and outputs of the various tasks therein, if it's not terribly structured, you're at high risk of amplifying mistakes. Right? And we're not yet at the point where you can just set a generalist super agent loose on a messy environment and have it clean it up for you and then go, right, we're not there yet for that group. That business process management is value streaming the workflow with the functional people just milestone steps. What's happening at each milestone step, what's working well, what's not working well in each milestone step that gives you signals to say, all right, where do we have our waste? Where's our delays, our bottlenecks, our frictions? What's holding us back from really achieving next step performance out of this business process? And then we get into those little waste areas with deep process mapping, root cause analysis, and that ultimately gets you to your system design. There's a long road of very conventional work to do before you're then ready to say, okay, what tool which flies in the face, I think, of the natural orientation of a lot of people in this space, which is if X problem, Y tool. And it's just not that. It's just not like that. So for campaigns processes, for deal management processes, for renewal processes, that's a great approach. The other approach is, to your point, actually flipping the model. Like, insight comes with a strong opinion on the process as to how you ought to structure it, how you ought to interplay generative AI with some of the conventional orchestration technologies. And we help you rapidly prototype to an MVP through like a hackathon workshop, right? So a month, month and a half ago, I flew out to one of our portfolio companies in Los Angeles. They got an Airbnb. The functional people were downstairs, the OPS people were upstairs. And throughout the day, the functional people were tuning the prompts and like getting the campaign research, the campaign tactics prompts together. And then I was upstairs with the OPS people who are starting to scaffold up the orchestration, like the underlying zapier flow and stuff like that. By the end of the day, we hit go on a demo able workflow, and it was largely following a pattern that we had kind of preconceived. So we do both, but it really depends on the circumstance of the portfolio company.

Wow. Okay. And when you, when you come into a company for the first time, there are so many different functions that you could drill into to figure out where the gaps are. How do you do that initial audit to just even triage to know where to go deep?

Yeah, well, it starts with the business goals and challenges. And you know, very often in working with our portfolio companies and this generally at Insight, you know, we have our quarterly board meetings, we have their financials and we're working in close partnership with our founders and our CEOs to really understand like what are the main two or three North Star objectives for the year? What's going to help them hit that, say double year over year growth that they want to achieve? What's going to help them achieve additional points of EBITDA by the end of the year? And then we dig in there and ask, well, at what level of maturity are we applying AI and automation to achieve that efficiency, to achieve that additional capacity to drive growth that then begets you the use cases which we see gain the most traction in the portfolio. And again, a unique insight vantage point is we can kind of cut through the hype and see, yeah, we got 30, 40, 50 portfolio companies that have fully autonomized their customer support for Tier one. We have strong confidence that that is a good workflow. We're going to, in a sense, you need to bring down your cost to serve. Like we're going to, we're going to help you focus on that. And so it's a filter of what are your goals, what are your challenges, what are the various available use cases stacked, prioritized by where we see the fastest impact for the lowest lift.

Yeah, and what I love is that you're really working across the whole boat. Otai, this isn't just about prospecting. A lot of it is about after the sale, working with existing customers. What would you say some of the most common high value use cases are? Prior to sale and post sale. Sure.

So I have a very simple prioritization framework and it matches a lot what you just surveyed. Right. First, make sure that you're hitting every inbound lead that you get. Right. I have some portfolio companies who have that beautiful problem of there's too many inbound leads, it's overwhelming the sales development organization. Like make sure you're hitting each of those like rapid speed, the lead, you know, and with the agentic tools that are available today, you can In a transactional interaction, qualify that, set a meeting, even do a meeting handoff to an AE. For instance, I have one portfolio company that 3x the meetings that they booked at half the cost per meeting by implementing an inbound agent. And they had this great story where an AE went away on vacation and he was in pilot and coming back to the pilot from his tropical vacation and he showed up and he had more meetings booked in the week, like 40 meetings than he had ever had in his long tenure at the business. And it was just like an immediate aha that this was a pilot that was going to bear out for the company. So number one, prosecute all your own bounds, right? Related to that, if you are in a situation where you're doing some outbound and especially where you're doing it based on intent signals, answer all of your signals. We have companies that aren't getting to all of their qualified accounts that they're getting from their account based platform. So take some of the same practices of conversational emailing of sales person co piloting and apply it then to your outbound. Number three, if neither one or two are a problem, increase the throughput in the system, increase your campaigning capacity so you can generate that demand, drive more inbound or drive more outbound signals. And then finally all that should be pushing more throughput into the deal team. I'd love to see situations where the deal team is like no Moss, I got too many ops, I'm working too many because that's a great problem to solve with a lot of administrative backends and meeting co piloting, meeting prep, even contact Lifecycle management is going to help them. Then the short answer on the post sales essentially replicate that same prioritization framework, but run a post sale with the only addition being customer support as we described earlier. So super high traction use case, very clear opportunity to gain efficiency. Gain efficiency there.

Totally makes sense. And thank you everyone for I know some people are putting their questions into the Q and A and I'm going to switch over to those in just a few minutes. You probably have so many specific stories, I don't know how many of them you could share, which ones you feel like sharing. But let's talk about outcomes. Any success stories come to mind where the AI enabled go to market workflow really drove results and then really sort of what made it work in that case.

Yeah, sure. So I know Sixth Sense, a portfolio company of ours, won't mind me sharing this because they've shared it themselves. They're very good at six sense on six Sense, they had an interesting problem at the beginning of last year. They were only achieving 50% SLA on their qualified accounts. Right. This is six out we're talking about. Right. They had, they're generating all this demand and their outbound team just simply couldn't get to all the signals that they were generating. And so they used their own conversational email agent to establish what you might call a backstop. So human, you got four hours to get to this qualified account, four business hours. If you don't get to it, your backup bot is going to jump in, it's going to start that conversation via email. And just, you know, the initial few turns, especially on the enterprise stuff, it's not going to try to carry it too far. It's going to escalate as soon as it finds something that's qualified. But that helped them move from 50% SLA compliance at the beginning of the year to 100% as you might imagine by the end of the year. And, and it also drove about 20% of their pipeline was driven by this one particular strategy. So, you know, great example of them kind of plying their walking their own talk, if you will. And I love the concept of bots as a backup to the humans. Right. Another compelling story. I'll share a couple more. One comes from one of our portfolio companies that essentially had that campaign's capacity constraint. Right. So we're moving from outbound up to number three, the campaign's capacity. Global organization, cyber security, very technical, lots of different products, lots of different Personas. And what that results in is this kind of permutation of segments that you could potentially address. And of course we all know in a good account based motion, you're doing that segmentation really well, you're personalizing really well, perhaps even down to one one. Their team couldn't handle all that and so they implemented a campaigns generation engine and essentially a field marketer in a particular region could fill out a brief, simple form. Here's who I'm targeting, the product I'm talking about, the event I'm attending. And the engine would then compile that all into essentially like a bill of materials, the campaign itself, landing page ads, emails like 90, 95% ready and the global demand gen team's just got to do the final QA and boop. It goes. As a result, year over year they 10x their campaign's capacity. So massive increase and they dropped the turnaround time for those campaigns to 10 days, which was something like a 2/3 reduction in time to produce so huge increase and that's. We've seen that repeated at several different portfolio companies. Now the last one I'll share is for the folks who maybe have more of a product led growth notion. There was a company of ours that was paying way too much to Google in paid ads and it's to supplement some of their weak rankings in organic e commerce provider, you know, long, long small to medium sized business target segment. So they went after the long tail with an SEO engine essentially here's where we're paying too much but we see good traffic from our knowledge base. This is the good content that we could use, but we just need to articulate it into a blog post or additional pages on the website. And they set up that machine and it just ran. And as they started to get more numbers and they started to see the different paid keywords pop, it would go into the top of the pile and it would produce more. So this is an absurd number. They 100x to their first page keyword rankings on this super long tail and within about 10 months organic signups actually superseded paid and they were able to achieve full ROI on the, on the organic program.

Oh my gosh. Wow. It's fantastic. Usually available, I think I actually saw that an early version of that campaigns generator in Austin. That was very cool.

Yes, that's right. We, that's been an inspiring example for us and we built a demo stack that we, that we try to add to some events in the fall.

Yeah, I love it. Those are three really fantastic examples.

For.

Companies that are locked into an existing tool stack. How do you help them adopt as these things change so quickly?

Yeah, totally. And this is that classic systems point of entry question, maybe the most common question I get for those that are locked into an existing tech stack, you're hopefully already looking at some of the AI capabilities that are available to you within that stack. And then there are ways about it with particular categories of stack elements that give you a lot of extensibility and I'll talk about that. So you know, I'll be upfront with you. There's some, I won't name them by name, but there's some very core legacy components to that stack that you'd find in just about any go to market workflow where they're being very forward with providing some AI solutions that are integrated therein. I have it in the portfolio, seen a ton of uptake on those solutions, but by my understanding from the providers themselves, they're working very hard at making those highly capable. And, and so again it's the same process, like what's your goal? What's the challenge that you're experiencing the associated business process? And are the capabilities that are available within those tools going to help you apply operating leverage to that process as defined by our key points of friction, like where we have bottlenecks, where it hurts the most in that process. Can, can we at least plug in AI that they have? This is a different paradigm than I need to do outbound. Let me go get an outbound specialized tool, right? Like if you can get focused on the business process and wherein specifically you're trying to solve, it might be that your existing tech stack has a sufficient capability to solve that particular pain. And that's why it's important to bring it down to that level instead of buy it. Like the process level or the tool level. Now if your stack includes a iPass, let's say like a Zapier or a Make, maybe you're really aggressive and you have an nan. Now of course you have massive extensibility because each of those have now built in a like a model context protocol type of offering where you can deploy agents in the most aggressive sense. At the very least, each of them you can call a wide variety of LLM models and it's a great opportunity for you to bring in reasoning. Right, where the rules based structure of your workflow breaks down, where it's too complicated or it's ambiguous and you need to use principles to make decisions instead of explicit criteria. Like just plug in your OpenAI call for your ChatGPT instance, plug in that prompt into that workflow and now you have pretty extraordinary extensibility in your operating leverage.

Okay, makes sense. Thank you. I'm going to switch over and read the the questions that attendees have submitted. See. So Malik asks, from what size do companies tend to build versus buy? Can you share some trends you see, particularly by company stage, early stage series, bc pre ipo.

Ipo, Yeah, I don't have explicit data on this. At least we have plenty of data. We don't have any data that suggests that there is a very clear profile or rule that you might follow. In other words, it's a variety show and it depends. The only trend that I do see is that if they are, you know, let's say 75 million, 100 million plus, they probably do have like a dedicated AI group. That dedicated AI group is probably largely doing build using Snowflake or they're using the OpenAI platform they're building on top of Amazon and various LLMs that are available to them there because they want to work, they want to combine their product and their cloud provider for, for their internal applications. I certainly see that type of profile. But then at the other end we have some early stage companies that are vying for like extraordinary revenue per employee and they are playing with the agentic platforms, the Crewais, which is a Portco, the relevance AI, which is also a Portco. And they're, they're kind of starting first with the bots and building with bots versus going out and kind of conventionally getting that head to see if it can get that leverage. So you see build behavior kind of across the spectrum. Those obviously look very different in kind of who owns it and the way that it's built and the cycle. All the classic things that you'd expect in a kind of a direct relationship correlation between size and agility. But that generally tends to be the profile that we see.

Okay, yeah. Someone asks, so when it comes to the insight Portcos, how do you advise your large horizontal companies on identifying go to market alpha? And then they say, we've encountered a lot of model drift when trying to apply AI pattern recognition to our customers.

Okay, yeah, great question. I think, I think what's being asked and feel free to send another question if I'm not hitting this big horizontal market, which means that we have a wide variety of prospective contexts in which to interact with our customers. I have several of these bigger companies, multiple products, multiple different sectors that they're working in multiple different Personas. And you just simply cannot get away with say for instance, a generalist support chatbot because there's just too much contextually particular elements, the way that you interact with say somebody in oil and gas versus somebody in banking. And so for our, for our larger customers who are working in this context, we see them, you know, not only taking a very focused and deliberative use case approach where they're like, okay, we're going to focus on customer support and we're going to get that right within customer support. We're only going to do tier one, but they're also actually sometimes segmenting that effort. Well, this is, you know, I have a Porko for instance, who is in this circumstance. They're using intercom fin and they first deployed intercom thin within their channel partners for a particular vertical. And now just now they're looking at migrating that over into direct to customer channel and when they do, it's going to be in that same sector of healthcare and then they're going to start to replicate and pivot across. So yeah, if you're in a big kind of highly matrixed market context where you need to, you know, have good contextual wrapping to the way that your buyers and your prospects interact with your business, then you do need to be very careful about the way that you roll out your tools.

Right?

Yeah. And I guess the alpha part of that question is start where it hurts the most and the value is the best. Right. Like that, that example is a good one because. Because that, that for that company that, you know, channels 50% of their business and then that particular sector for them is, is, is a high margin sector. Logical. Okay.

Kaylee. Hey, Kaylee asks when you say you have poor codes that aren't making it through, excuse me, through all the leads generated from their ABM platforms, what specifically do you mean by that? She follows up by saying no way that you have brands finding success with the 1.0 ABM stack, like 6 sense demand based. Right.

Interesting. Okay, I think again, I think what, what you mean by this is sounds like maybe the Kaylee's a little bit surprised that people have this too much of a good thing problem. And believe me when I tell you it, believe it or not, it exists out there. But the good thing quickly turns into a bad thing when it's 20% qualified and 80% not qualified. Right. And for some sectors like you're just, that's just the reality of the way that you do business. If you're going to exist in that sector, you're going to get a lot of shack for the week. And so it only becomes more critical that you have like strong qualification criteria and that you roll it out in the agent so that the agent can split that up. I don't know. What do you think Jared was I, did I hit that question or you think she was asking something else?

I, I think you hit it, but Kaylee actually threw another question in here. Let's see if it relates. So we'll give her a follow up here. She says a lot of spaces are talking about AI agents, but fewer building in public. Okay, slightly different topic. But what are your Go to resources to learn, see or build for this new future. And then she says aside from that, go to market engineer school. Of course.

Of course. Well, it starts with go to market engineering school. Yeah. So totally get this problem. Right. And it's been a problem in this sector ever since really our chat GPT moment where there's a lot of people who are talking about it, but very few of them are bringing the type of clarity and structure that Jared and his team have done with go to Market Engineering School. On specifically the agentix side. The, you know, Andrew and his group at DeepLearning AI have long been leaders in machine learning and you know, we partnered with him on, on a few companies and his, his material is great. You know, they do have some entry level stuff but especially if you have some more technical people, it can get into the intermediate and the advanced and super strong library of content there. Beyond that I come with a little bit of a bias or a benefit because I learned quite a bit directly from our portfolio and so we do have a couple of agentic tools in our portfolio. Crewai for instance is one of them and he partnered with Andrew in a course on his offering and so I've benefited from that of course relevance. AI is another among many others who are rolling out agents and frankly just going direct to them. The material tends to be pretty strong of course, kind of steeped in their version of how you roll out agents. But unless if you're in a situation where you're doing agents from scratch, then your deep learning, your OpenAI documentation is going to be your friend. But probably for most of us we're using some type of interstitial platform and I'd probably start you there and ignore a whole lot of the rest because just there's a ton of hype. It's very easy to, to build a very thin agency platform and I would just orient to the ones that tend to have market leadership.

Fantastic. Okay, good. John asks how much collaboration do you have across Portcos? You sort of touched on that just a little bit in your previous answer. Maybe you could drill into that a little more. And he asks do you share automation formulas and prompts to help drive use cases, especially at early stage companies?

Yeah, sure. So benefit of being in an inside Portco is you're part of this massive family and I would say that the family is very good about sharing amongst, amongst themselves. And so it's been a wonderful lab in the past few years where we've had classes not unlike go to market Engineering school where we have cohorts from various Portcos and they have indeed shared prompts, zaps, you know, recipes. They're very good about that. We also have a series called the on site Expert hour where every week we want somebody from our more than 100 person growth platform team, kind of a team of experts that have functional expertise not just in AI, but sales, marketing, hr, finance, what have you. They'll host roundtable, many of them as you might imagine have been about AI. And we'll get 60, 100, 150 portfolio companies that are showing up to those. And as the name would suggest, they're sharing five or six feature panelists who share what they did, where they fell short and then people get to pipe in. And we actually just published on our blog one of our recent on site expert hours and so I encourage you all to check that out if you want to see what that's all about. So lots of collaboration and we even see commercial partnerships between some of our portfolio companies if they're somewhere in the same sector. And those things complement. So those are just some of the benefits of being an inside port company to be sure.

Fantastic. Hugely valuable. All right, I got a couple more questions here and then we'll wrap it up. So Manin asks for companies in the 100 to 250Arr range. What kind of org structure is needed to successfully drive AI adoption across the go to market function? Leave it in IT or a separate cross functional center of excellence team?

Yeah, that's a great question. I mean having a cross functional AI center of excellence, AI steering committee, some type of coordinated body at that scale. A non controversial. Yes, like 100%. You ought to form that when you get beyond that you do tend to see kind of a variety of ownership structures. You know, one Porco for instance that I work very actively has a data analytics and AI group. So they've kind of lumped that all together. Another Portco app that size that I've worked with directly has AI group that is solely focused on internal operations and leverage type of projects versus a separate group which does AI within the product. And then you have kind of the third structure of you know, chief of staff or innovation office at the very top owns the strategy and each C suite executive is responsible with their teams to have their own kind of distributed center of excellence. So it's kind of again a bit of variety. Any one of those structures is functional. You know the way that you make the decision as to the best one for you is to observe where your technical proficiency is, where your data proficiency is and how well those centers of proficiency integrate with your functions. That's, it's, that's the most critical part is as we talked about earlier, technology in some ways is the easy part. It's the business process management, the business change that critically has to be a part of the AI change. And so what I've seen not work is an AI group or an IT group especially it's as just kind of out there doing project based work and they're not sufficiently connected with their, with their business partners. Quick story. I was on a call with one Porco, I will not name them the AI crew demoed for us and the cmo, this multi agent architecture that they developed from scratch. And look at this. It grabs info from zoom info and like it goes and searches the contacts on LinkedIn and gets additional enrichment data and then it like drafts an email for them and then they could hit this button and it's like this like frankly kind of homegrown interface. Very janky. I'm like, everybody on this call knows what that sounds like. That's, that's like a clay, right? And the CMO literally was like, oh, that's funny. We just, that looks a lot like we just bought this tool called clay. And I think it does the same thing. I was like, yeah, like you just burned like 300 hours with your AI crew. Like you got to talk, you got to coordinate. So again, just whatever structure you pick, the coordinative mechanism, whether it be a committee, a top level office, is absolutely critical or less. Unless you want to waste time like that poor code did.

Funny story. Yes. Oh my gosh. Okay, well I'm just going to throw you one, one last question to wrap it up and that is to ask you Jared, what are you most excited about or worried about when it comes to AI in the next 12 months?

Yeah, I mean everybody's talking agents and I am starting to see some extraordinary applications of agents in the next 12 months. I expect to see that more of my portfolio companies have flipped their human roles from frontline and coordinative work up to director level work. And what are they directing? They're directing relatively broadly scoped agents that are responsible for not only a business process, but perhaps multiple because they're in the strong role scope. So we're just starting to see the early versions of that in relatively narrow stuff. We're starting to see them build out a multi agent architecture that starts to look a lot like a norg chart. And I think for the human staff it's just, it's, you know, I'm hearing nothing but excitement because it is pulling them out of the stuff that sucks. Building marketo programs, you know, scraping from the web. Right. All the things that I'm sure you covered in the GTM engineering school, but they're packaging it in a way where it's more like you're working with an intern then you are necessarily having to do a lot of deep like systems in mechanical built. So I like much of the sector. I'm very excited to see this next chapter bear out.

Fantastic. Good to know there's actually something there and it's not just a lot of hype.

It's not hype, no. I'm starting to see traction now. The folks are out there who are saying, you know, a general stage agent is going to build all your sub agents and it can be your whole team. Slow your roll on that. But for. For the folks who are, you know, especially in the bigger organizations down at the front line, like, there's. I think there's a very exciting transition that they're going through.

Wow. Fantastic. All right, well, I think that is all the time we allotted here. Thank you so much, Jared, for spending the time with us. I know I learned a lot and I'm sure all the participants did. Thank you, everyone, for attending and keep learning, keep experimenting, and we'll talk to you soon.

Yeah, for sure. Thank you, Jared, for having me. Thank you, GTM Engineering School. We'll catch you guys soon.

All right, take care. Bye. Bye.

Bye.