Ep. 189 - SaaS in the AI Agent Era: Org Chart For Your Agents

Speaker A: Welcome to the SaaS Backwards podcast where we reverse engineer the success of fast growing SaaS firms and explore strategies CMOs and CEOs are using to drive their businesses forward.

Speaker B: Welcome to SaaS Backwards, a podcast that helps SaaS and AI software, GTM leaders and CEOs to accelerate growth and enhance profitability. Our guest today is David Gabriel, VP of Marketing at Rumbix. Rumbix is a field and workforce management software built for construction that makes it easy to capture time and utilization data, uh, around labor, equipment and materials and deliver powerful insights and daily and weekly reports. David, welcome to the podcast.

Speaker C: Thanks for having me back, Ken.

Speaker B: Yeah, well I'm really excited to have you here. And you know, careful listeners might recall that David was a guest on the podcast in August of 2022 when we talked about how important alignment of sales and marketing are and hiring the right people are to the success of an early stage or scale up business. And uh, I think there's going to be some interesting parallels and, and points of departure as we get going. But before we dig in, David, could you please tell the listeners a little more about yourself and the company you work for? Runbix.

Speaker C: Yeah, absolutely. So, David Gabriel, VP of Marketing I work at runbix which is a field and workforce management software for the construction space. We work with a lot of enterprise and mid market accounts and at Rumbix we are really excited about AI both within our operation and also understanding ways to leverage it within the software. About myself, I work with startups and scale ups in the industrial technology space and so that's you know, construction, manufacturing and so love, love the topics that we're going to discuss today. Kat.

Speaker B: Yeah, ah, um, and for people that don't know you coming in cold to the podcast, you're doing some really groundbreaking work in implementing AI within the marketing function and specifically making uh, go to market an agentic process. So I really want to dig right in because we have so much to cover. And just to start off, you've m moved well beyond using AI to help you write content and we're all doing that. I think everybody's doing that. But you've built much of your marketing function around agents and I want to start with like why, like what was the problem you were trying to solve?

Speaker C: Yeah, absolutely. Well, you know, before I came to Rumbix, um, you know, was looking at a few different opportunities and Rumbix stood out because there was a shift from going from problem market fit to product market fit and figuring out okay, we have traction and, and so how do we scale this from a repeatable process standpoint and then an efficient repeatable process standpoint? And so coming in, what was really exciting is I had the opportunity to really rebuild the marketing function. And uh, as I was entering, I'm thinking to myself, what if I could create a go to market function that is 80% AI enabled? And so this was something that the CEO Zach was interested in and curious about. And with all of the tools and new innovations that occurred in 2025, it was looking like it was possible to do this. And so the, the big problem, you know, just coming in that I really drilled in on was their go to market was really dependent on inbound. And the challenge with inbound was they were getting all types of customers, uh, not just, you know, the best customers that convert fast, but they were getting customers that maybe shouldn't have been in the funnel in the first place. And this is very common for companies moving from startup to scale up. And so that was one of the biggest challenges. And so being able to address building out a completely new go to market motion, which for Rumbix was outbound, this gave a huge opportunity for us to execute.

Speaker B: So you also mentioned that there were some constraints on your department. Right. So this was a response not only to the opportunities in front of you, but you were managing some constraints as well.

Speaker C: Absolutely. I mean I think every startup B2B SaaS company is facing constraints. Like if you look at the, the stock market as it relates to tech stocks, the valuations are being affected as well. As, you know, I, I, I am a part of uh, Pavilion and you know, lots of CMOs. One of the things that they share is their, their expectations are going up so high from the board and uh, their budget is staying the same or even decreasing. And so that causes a lot of tension and frustration. But it's also exciting because if you know the tools and if you think like an owner, you're like, wow, this, this could be something where we could actually increase revenue and do it and reduce expenses, which is rare. There's not many, uh, you know, opportunities to do that. And so which ultimately increases the margin, the profit margin, which, you know, everybody in the company benefits when the company is growing and growing more profitably. So I thought that was one of the exciting opportunities here. So coming in, the way that I thought about it differently was, you know, who would be the first person I would hire? And for me that would typically be a product marketer at our stage because, you know, we need to differentiate as we see more Competitors coming in. As we're focusing on icp, also understanding why are we winning deals. That's a key aspect. So you know, looking at that role, that would normally cost me 150 grand base maybe and then you've got bonus on top of that. And uh, you know, thinking to myself, like, could I build an AI agent to essentially do all of the jobs of, of this individual that I would normally hire?

Speaker B: That's a kind of compelling and scary proposition. But tell us just a little bit more about what that meant to you. So this was like your launch point into it, right?

Speaker C: Yeah, absolutely. And so, you know, I, I started with the problem going back to, you know, even just like budget constraints and we wanted to, you know, we want to be a rule of 40 business. Right. And like with everything going on growth rate, especially when you're going from startup to scale up, um, you're normally, you know, if anybody who's done it, you're growing really fast and then you hit a wall and you have to rebuild the business. What got you here isn't going to be the thing that gets you to the next level. And so as a leader coming in, I want to understand things, but I also want quick wins without like blowing out the budget. Right. I'm a big believer. I know Jeff Bezos says fire bullets and then cannonballs. You know, I'm a big believer of coming in and doing rapid tests. And so the AI product marketer idea, you know, eventually we'll probably bring in a product marketer, but I was thinking, hey, why not have one today? And so that, that was part of the thought process of, you know, building it.

Speaker B: I think here the point is that you brought in an AI product manager, right. You created that function first and uh, you know, it allowed you to do more with the budget that you have. I think that's really pretty powerful.

Speaker C: Yep, exactly.

Speaker B: Going back to that thought process, it's a little bit of an out of the box idea, right. That I'm going to come in. You're, you're a, uh, department of, is it one or two now?

Speaker C: So we have myself, we have somebody that manages the pipeline generation aspect. So like demand generation. And then we have another individual that's more on the content side, which is, you know, awareness and consideration.

Speaker B: Great. So we have a small department and now what we're doing is we're able to augment their capability and improve their, I guess, effectiveness. Right. Their ability, the throughput of the organization by starting on this path to build agents.

Speaker C: Yep, absolutely.

Speaker B: Staying on this like track of rethinking our go to market. How much of go to market do you think can be agent led? And I know you're still in the journey and we're going to get a little deeper in that in a minute.

Speaker C: There are some parts of the go to market that I do believe remain uniquely human. And so like a good example is we built a BDR agent and our BDR agent is, is crushing it right now. It's actually for February, it's at 13 meetings scheduled. And this is coming from like zero outbound in September. And so we're starting to see growth and momentum from, from the BDR agent over the past three months. What's interesting is when we first worked with the AI to build this agent, the information it was giving us was just bad, you know, sales tactics. Like as an executive, if I, I mean if. And you, you get this, if I were to show you this, you're like, everybody's sending this. This is like so 20, 22 or 21. And so one of the unique challenges is AI is trained on what everybody else is doing. And part of the go to market is doing things differently, being different, having a unique point of view, especially from an outbound tactics standpoint in order to get attention. A lot of times it helps to be bold or courageous in doing things that are different. And so AI I think, struggles to understand when to make the bold or courageous move. Another example is even thinking about this, the stock market. The moment you have AI introduced into the stock market, it doesn't give, you know, people an advantage. What it does is it just changes the game. And so I think there are unique parts of the business where, and a lot of it is strategic, a lot of it is the delegation and really identifying what is the right process. When is this acceptable work, when is this not acceptable work? Like those are things that are usually human intuition. And so the AI has challenges in my experience in replicating that. And so I do think, you know, you hear a lot of thing things about like, you know, AI is just going to replace everybody's job. There's going to be no white collar jobs. I think that's a little bit of an over exaggeration. I do think like, uh, as I'm working with AI, I do think like working together we can achieve a lot. But it looks different. It's more delegation, it's more stewardship, it's more time management. Because another thing is as an individual, if I'm building all these agents and I'm not good at Managing my time or prioritizing, then I'm just going to compound doing the wrong things. And so those things are remaining uniquely human. How I'm thinking about go to market is really doing the work is you can ha, you can spin up agents. Doing the work is not the bottleneck. The bottleneck is now the delegation and then it's the review on the back end.

Speaker B: Yeah, it makes a lot of sense and I think, you know, a way to also look at the opportunity is you have a team of three that's probably going to start doing the measurable work of a team of five, six or even eight people. And so you're creating opportunity within these smaller organizations to really punch above their weight. So I think it's actually a very exciting thing for career oriented professionals to be involved with in the startup scale up world.

Speaker C: It is the opportunity. When I was 22 years old I started my own business and it was around digital marketing and uh, 2012. What's interesting is as that decade continued and Even looking into 2020, the leaders were those that embrace digital. Right? And I think AI is the same opportunity for young executives that are rising stars embracing leaning really, really hard into AI, it's going to be an advantage. There's a book called Rookie Smarts which really I think valuable for everybody to think about and it's that you know, even those that have had success, you've gotta continue to reinvent yourself. And you know the advantage that a rookie has coming in, they, they don't know if something's a dumb question or oh, I can't do this. They just are like, I'm just gonna figure it out and I'll ask people and you know, that's the fastest way to, to learn these tools. Like it's not reading a book, it's not, you know, there's a lot of YouTube videos, it's hard to, you know, sift through it all, but it's really having a community of people that are doing it and doing it really well and just like grabbing their coattails and asking them questions and playing around with the tools yourself that I think is going to help any go to market executive have an unfair advantage.

Speaker B: I agree and I want to get a little more specific for listeners. So I think you need to walk us through what you've built. Horror roles that agents are performing at RBX today because it's, it's a pretty broad set of mandates so it would be great for you to walk us through what you've built and how you organize them. Much like it was an org chart of human workers. So let's, let's talk about what you've built and um, how you're organizing them.

Speaker C: Yeah, well, you know, the big aha for me, Ken, was In the first 60 days I was able to, I was on a call with uh, one of my team members, Brett, and we were working on the BDR agent. And I set up four different agents. And uh, while I was on the call, not you know, typing things into chat GPT or you know, clicking, clicking buttons, I was just laser focused into the call. While I was on that call, 25 hours of work was being completed in, in the background in just two hours. And so it created, it went through all our GONG transcripts and created case studies based on a template that I had given it. We actually had another BDR agent that was reaching out to 20 prospects. We scheduled a meeting with the CFO. The product marketer was working with one of our team members to redo our homepage and align it with some of the gone transcripts. And then we also had a LinkedIn connector finder agent that was helping us find connections that are in our network. And so all of this happened in the background without me even doing anything. And so that was like the big aha, uh, moment of wow, this can do work while I'm doing other stuff. And so from there, you know, one of the biggest challenges is not necessarily building the agent, but it's figuring out where do you want the agent to fall focus. And, and it goes back. A uh, good like litmus test is, you know, whenever I think of an agent I want to build, I ask myself like, how is this going to help increase topline revenue? How is it going to reduce expenses, increase margins, reduce risk in the future, or improve the culture of the company? And so those are, those are the things that I'm thinking about as I'm building these agents. It's tempting to build just because of fear of missing out, but really the power of these agents is to just completely obliterate the bottlenecks in your business. And so really, uh, highlighting what's the bottleneck that is restricting growth, I think is a good question. But anyways, going and going back to the org chart, you know, another question I asked myself was if I had 40 more hours, what would I spend my time doing? Or if I could hire one person? And that was the product marketer, you know, in this case. But if I could hire one person, who would I hire? And so that helped me think of these agents as Specified roles. And from there, you know, and I, and I've written a lot of this I'm trying to build in the open on, on LinkedIn. But from there you know, you, you realize that if you give an agent too much work, it gets overwhelming and hallucinates just like a person. And so, you know, you may want to split it into what's called sub agents. And so you have a manager agent has a lot of different sub agents that are specialized. By doing that, that agent can, you know, send it to one of its agents on its team multiple. Those agents can work together on a project. Or if you have six tasks, you can replicate a specific agent six times and then do the task, you know, instead of once, one at a time, all six are being worked on all at the same time. And so those were a lot of, you know, aha's.

Speaker B: Ah.

Speaker C: And what the challenge that I had is, I, it's like I had this vast amount of workers that I can build and I wasn't really using them every single week. And so that's where I decided to map out like via an org chart and also have like, you know, the agents and like what is their performance? It's funny, I, I have it actually right here. So across the org chart the different departments I have, so I have product marketing, content campaigns, bdr, lead gen, customer communication, field events, partnerships, and then I have revenue operations and then custom agents just for specific use cases. And, and so this was really helpful because a lot of these agents, you know, you may have three that are working really well, but then one is not working well. And because of that it's, it's making it so you're just not using that agent. But by being able to review their performance, you can update the skill data uh, on, on that specific agent, which is basically, you know, it's, it's so funny because it's like, it's similar to how you interact with teams. Like I've uh, managed teams of over like 20 different people and if somebody is going in the wrong direction, you either reprimand or you know, you basically tell them, hey, we need you to be over here. It's the same thing with the agents. And so if an agent is going in the wrong direction, you don't throw the baby out with the bathwater, but you tell it, hey, like I would like for you to do this instead of this. And what it does is it updates the skill data for that specific agent. And so it's really interesting because like we're building so many of these agents. And there, there would not be a person on earth that would hire a bunch of people and they just never talk to them. These agents need to be managed. They need like weekly, you know, hey, course correction, you know, especially as they're getting started. That, uh, that's one of the interesting things. Like as you're building these agents, they need a lot of care up front. But over like, you know, after the first month and second month, you're like, wow, this is, this is really like, I don't need to do much with this agent anymore. And so a lot of times the challenge is just getting started.

Speaker B: Well, you know, that's a great segue and it's really not what we prepped. But I think you have to give people just a little window into what it takes to get started because to many we're speaking Greek, so perhaps it would be good, just step back and say, hey, here's a good place to get started. Might even be with your LinkedIn profile, because I know you put a lot of data there, a lot of steps are outlined there. But just give people a little bit of the world we're talking about because it's, it's not ChatGPT, it interactive chat to write a blog.

Speaker C: Yep. Yeah. And like, like Ken, you just said, you know, I've been putting these articles out just because it's, it's hard. You know, I'm on LinkedIn and it's, you see these amazing resources and then it's like, type, you know, Claude and I'll get you everything. And I don't know about you, they, they never respond to me, they never send it to me, or if they do, it usually goes to like a, uh, you know, $99 course or something that I need to buy. I've been trying to just put it out there, you know, the content, because I think there's, there's just a gap of practical information. So specifically, you know, these tools are changing constantly. So really depending on, when you listen to this, that's a huge factor. But as of, you know, February 2026, the two tools that I would highly, highly recommend is Claude, specifically Claude Code and Lindy. These are two tools that have really been, you know, helpful in building an AI enabled organization. Claude code is a little bit, can be a little bit intimidating because it has the word code in it, but the interface is very similar to, you know, if you were using ChatGPT, you just have almost like a coder that you can, uh, delegate to and, and ask to do things for you and there's so, so many things you can, you can do with that such as like connecting to Canva and asking Canva to do something for you or a very specific uh, you know I think the best, one of the best places to start is connecting an MCP into your CRM or into Gong or, or whatever you use for your call recording and being able to chat with transcripts. So that I think is just low hanging fruit. And like I said with the CRM, being able to chat and say hey, I want you to summarize the activity feed and all the notes and the emails on this account and you know, being able to chat with that I think is really powerful. I know some of these, some of the tools they, they have products for that but I think those are two use cases that can be really helpful. And, and I, and I use the word mcp. It, you know, many people, uh, like I didn't even know what an MCP was, you know, eight months ago. And so I went into Claude, I'm like what is an MCP like? And it told me what it was. I'm like okay, like how do I build one? And then Claude code was like oh, I'll just build it for you. I'm like okay. So like what's amazing about these tools is they'll tell you how to use them. And so if you don't know how to use them, ask how do I use you? How can I leverage this in my business? And so my background, I'm not a coder, I'm not a developer. And so all of this I've learned by just asking the AI how to use it.

Speaker B: Yeah, so I think that's a really important point. The encouragement here is pretty much anybody with an inquisitive mind and willingness to keep trying is going to get somewhere valuable. Right. It's, it's not a matter of if, it's a matter of when. And I think the idea of being able to chat with your transcripts for example, could be fundamentally game changing for a lot of companies. Especially if you have a decent sized sales team or customer success team, you can start turning up insights that are very valuable and in a novel way and demonstrate the value of an AI investment to your, you know, your peers and management.

Speaker C: 100%. I, I know that you know, talking with some of the customer success side, not, not at Rumbix, but just in general, you know, I've, I've heard over the years like, you know, the, the sales reps are not giving me enough information on this account. And now you could just look at all the emails, activity feed, you can look at the gong transcripts, you can look at usage data. Uh, and what's amazing is you could also automate this. You could ask the agent, hey, Every Monday at 9am I want you to send this report or before the qbr, I want you to build this report. And then it builds it and has it prepped to you. A lot of times, like the first try, it doesn't get it right. I see a lot of people stop there. But if you work with it and give it a template and update the training data, giving it, you know, criticism where it needs to improve, you can really make it work. Which is exciting, is super exciting.

Speaker B: Um, and I think this leads into our next topic really well, which is you've said to me on multiple occasions that agents really only work when there's a repeatable motion, when there's some guidelines to follow. So what are the disciplines that have to exist before AI can truly accelerate? Go to market.

Speaker C: So it's funny, I was just talking with somebody about this and what's interesting is there's a lot of executives that are, that are unable to communicate the process, which is revealing that they may not understand the process or there may not be a process. And so a lot of times that's the first step in this. Another tricky thing is in a startup sometimes the process is to be agile. And so how do you create a repeatable? And there are ways to do that, but it doesn't look as simple as like a, I shouldn't use the word simple, but it's not as clear cut as within an enterprise company. You know, here's the process. They, they may already have those processes documented. But specifically looking at, uh, you know, it starts with what is the pain point? What's the opportunity? Why are we doing this in the first place? Does this matter? You know, so I think that's the first question. From there, one of the helpful things is to look at, if I were to hire a person to do this job, what would be the individual steps that that person would do? And so breaking that, if it's a specialist, you know, breaking it into individual, you know, parts of like, okay, they're probably spending an hour on this, you know, another two hours on this, you know, almost looking at a job description of that individual. Another thing that's been helpful for me for crafting a process is looking at if this was an enterprise account, like if this, you know, product marketing was a team of 50 people who would be, what would be the roles across this organization? And so you may have somebody that's dedicated, specific to competitor intelligence, for example. So it's not just, hey, product marketing, but it's also like, okay, under product marketing, you have someone that's responsible for competitive intel. You might have somebody responsible for, you know, positioning. You might have somebody responsible for why do we win deals in doing those interviews? Why do we lose deals? And so those are just a couple of examples of like, how do you break this down? So once you have that, you can prompt the agent. And going back to practical, it's as simple as saying to the agent, hey, I would like to build a sub agent infrastructure for, let's say, product marketing function within product marketing. These are the individual roles that I would like within the sub agent architecture. And then you can list those roles. If you have further documentation, that's great, you can provide it. But it's that simple of a prompt to get started. And then the agent will create, you know, those markdown files. It could be a markdown, it could be a word doc, but basically like text files that say, this is the training data that the agent references before it does its job. And so what's, what you'll learn is once you execute that, you'll learn like, okay, how did it actually work? And what I see a lot of times is it works maybe like 60%. There's sometimes it's you, one shot it and it's, it's great. Other times it's like 50, 60% good. And you may, you may be tempted to just throw it out and say, uh, you know, this isn't, you know, revisit AI agents in six months. But going into each of those individual sub agents, what it'll help you understand is like, where are there gaps within the process that you may not understand as a leader? And so it can help. The exercise of going through and documenting a process, I think is almost just as valuable as building the agent infrastructure to do the work.

Speaker B: Yeah, it's really interesting. It's like you're almost setting yourself up to ultimately hire the real humans to do some of these things as, as you sort of need more thoughtfulness and adaptability, right? You're building this foundation layer. I want to kind of change the perspective just a little bit, which is if you're an investor taking a look at a company as a potential portfolio company, how can they, from the outside in, ask the right questions about AI productivity games? You know, how can they tell whether a company's built Real systems that confer advantage or are they just, you know, spewing out blogs on the chat?

Speaker C: Part of the challenge is it's a transformation in the bigger the company, the bigger the transformation of you have to think differently, you know, when it comes to AI and what is possible. And so I think that the key thing, it really starts with the executive team in leadership. If the executive team and the CEO, they're not bought in to this direction, it doesn't matter what people are doing in the organization. And so I think it really starts with the senior leadership in the business after that, you know, let's say senior leadership is bought in really looking at where are the bottlenecks again going to Bottlenecks, uh, where we don't want to just do AI. How does this, you know, going back to those five different things of you know, revenue, expenses, margins, reducing risk or enhancing, you know, culture of a players. How is this, what's the hypothesis on rolling into these and each functional executive should have a vision in how they're going to work with AI to you know, again you're, you know, building two different companies. How's it going to help you today? But then also how is this going to help us in the future and help us be more secure from a repeatable revenue generation?

Speaker B: So uh, I want to, I want to focus this question again on like looking beyond the organization itself and you know, like do you have any advice for investors and boards on thinking about operating leverage and margin expansion? Like what should they be expecting? What should they be looking for? Yeah, what kind of teams, you know, are, are they going to be expecting or looking to be in place? How many David Gabriels?

Speaker C: Yeah, well, well I think what's, what's fascinating. So you've got RB2B. Uh, they've grown to I think 6 million in ARR at this point. They're still growing, you know, who knows what they're at right now. But they have three people and uh, they've been able to grow. So if you think of revenue head per headcount, that's, that's incredible. And it's only because of, you know, AI in the world that we live in that that was impossible five, ten years ago. And so you're going to see lower headcount teams. And so what used to take 3, 5 people you may be able to do with one person now it really does, uh, it, you know, your market and the strategy really dictate, you know, what the org should look like. But this new technology really needs to play a Factor, I think, you know, in every single software business, the most immediate thing is reducing expenses. If the teams are able to embrace AI, then keep the same team and you're just humming and pushing up revenue. But as change happens, you know, there are some people that want to be on that bus, there are others that don't want to be on the bus. I think being patient, you know, change is hard, you know, through that transformation. Obviously if you're in a red alert situation, you have to do what's best for the business. But I think starting today, to empower the employees in your company to really embrace these tools that could look like, hey, every week spend 30 minutes playing around with these tools. It could be doing hackathons. Like, those are ways to help your organization be ready. So those, those would be things I'd be looking for from the investor of like, it's not just the CEO says this, but they're actually doing things within the business that reflect that. From a product strategy standpoint, I think one of the interesting shifts is who, who are we going to be creating software for? Are we creating it for humans to interact with or are we creating it for the agents to interact with? And so this may sound, you know, crazy, you know, if I said this two, three years ago, but looking at my tech stack, I would say 80% of my tech stack, I haven't logged in in three months. But I'm using it every day. And I'm using it because I'm plugging in to the mcp and so the agent is using the tool on my behalf. And so the ux, the UI of software doesn't, doesn't matter as much for certain softwares. And so I think that's one of the things that investors should be looking at from the long term standpoint of like, is it being built for humans or to interact with other AI agents?

Speaker B: So we've gone from like headless, headless content management Systems to headless SaaS. Right. We don't need the interface, we just need the function. That might be a great place to land our episode. David Gabriel if people want to find you on LinkedIn, what's the best way for them to do that?

Speaker C: Yep. So it's LinkedIn, slash, uh, in Dave Dash Gabriel. And like I said, I've been writing a lot on these topics and just as I'm, um, just documenting as I'm

Speaker B: building, that's really exciting stuff. We only really touched on a few pieces of the work you're doing. I'm really looking forward to digging in with you in the future. And also your company, Rumbix. What's the web address? If someone in the construction or manufacturing business is interested in the solution, how can they get ahold of that?

Speaker C: Yep, that's runvix.com that's R H U

Speaker B: um, M B I X. Right?

Speaker C: Correct.

Speaker B: Great. And if you want to reach me, I'm on LinkedIn in Ken Lempitt, My advertising, advisory and demand gen agency for SAS is Austin Lawrence Group. And if David hasn't convinced you to subscribe to the SaaS Backwards podcast, I'm, um, not sure what will. David Gabriel, thank you for being maybe our first repeat performer here on the podcast. Thanks so much.

Speaker C: Thank you, Ken. It's been great.

Speaker A: Thanks for listening to the SaaS Backwards podcast brought to you by Austin Lawrence Group. We're a growth marketing agency that helps SaaS firms reduce, churn, accelerate sales and generate demand. Learn more about us at www.austinlawrence.com. you can email KenLempet@ah klostinlawrence.com about any SaaS marketing or customer retention subject. We hope you'll subscribe and thanks again for listening.