How a $100M+ Company Deploys AI Agents for Real Business Impact

Speaker A: Welcome everybody. This is Ryan and I am back with a very special guest. Today I got Jonathan Moss. Jonathan is the EVP of growth go to market strategy and ops at Experity. I know Jonathan just through multiple circles, community, you name it, right? We nerd out on AI things together. And I was thinking of someone that would be amazing to have on the show because they're, they're quote unquote AI. Ah, nerdery. Like myself. Jonathan, welcome. Happy to have you here, man.

Speaker B: Hey man, thanks, thanks for having me. And yes, I love the nerd in both of us. So yeah, we could, you could do this stuff all day.

Speaker A: We, we literally had to put the governor on the nerdery level with us when we started because we just started talking about a bunch of shit before the show even came on. And I'm like, uh, we're gonna run out of time. Yeah, we get, we gotta start recording soon. So we were just talking like, what about this tool? Have you done this? Have you worked on this? Right, so, um. Awesome as always, man. Good to see you, uh, and excited about what you're doing because like uh, you're working with a larger company. You guys are 100 million plus, right. And one of the things that I noticed from you that I've seen is like, I think you do a great job of deploying uh, AI concepts in your business, but then also still using them. Right? Because I think that's often a big gap where we see leaders that don't use them, um, don't use agents and then they just talk about it all the time. Right. So anyways man, why don't you give us a real quick backdrop on you before we get into it and um, we'd love you to share some of the stuff you're working.

Speaker B: Yeah, yeah, thanks. Yeah, so absolutely. And I think it's important, you know, something that you said, which is, I think there's a lot of people talking about it, but ultimately being able to use it myself daily, but also then building inside of the company that I am and then sharing that externally. I think there's uh, more of us that, that should, should do that if we, if we can. So yeah, so I'm a, I'm a 20 plus year go to market operator. So I've been doing go to market my entire career. I started in publicly traded companies and then I've worked in startups and even started a launch startup. So I've been kind of at every phase, but all everything through go to market. And that's kind of where My passion lies, I think Go to Market's a complex system and AI has a lot of benefits for, for helping us, uh, helping us run it and run it effectively and efficiently. So just a little bit about me, but mainly my experience is in technology, both B2B and B2B.

Speaker A: Okay, perfect. Well, well, let's get rolling, man. So what are you most excited about that you're working on right now? Because like, I know you're working on a lot of different things across different areas. Like what, what has you most excited right now? Um, on the AI side?

Speaker B: Yeah. So I'll actually give you, um, two things. So. And they're slightly different. So the first one is, you know, I'm working with, uh, Expertise, the company that I work for. And uh, we actually just launched a multi or multi agent orchestration platform that uses conversation, um, to really help with, you know, this kind of patient journey. So if you think about healthcare today and you think about how fragmented is from the time you need to find care, schedule care, manage your care, get your information, understand, you know, what your, you know, what the um, uh, you know, what your discharge paperwork says, what, what the labs even mean because it's all foreign language to you. How do, how do I, how do I schedule referrals? Yeah, all the different, how do I manage my own health and wellness? Ultimately, there's a huge gap right there, right? It's, it's fragmented. And so what we found is an opportunity. Uh, you know, the company has over 40 million patients that go through its software and services every year. And so ultimately we've started to build out the, we built the infrastructure and now what we're doing is we're adding what we call skills to that, to that infrastructure. So think of a skill as just a specialized agent that then can, has a specific task or outcome that it's trying to achieve on be on behalf of the patient. And then obviously on that orchestration that I was mentioning before is that it's using all the data and information that we have about the patient, about the provider, about medical history, etc. And just helping them navigate that, that healthcare journey. So that is something I'm, I'm truly excited about and was one of the biggest reasons why I decided to get into healthcare because this is my first healthcare role. Like I said, tech go to market. This is my first healthcare role. So super excited about, about that. I would say internally, if I think about go to market is really just reimagining how go to market works. And I think the, the word Reimagine is important because ultimately, you know, you hear a lot about bolt on AI or people slapping AI on top of their existing go to market processes and things, and I think that's the wrong way to do it. And you got to reimagine what is the buyer journey you're trying to build for? What is the go to market that you want to have? How can it be intelligent, predictive, proactive, autonomous, all those things? Um, and I think we're in the midst of being able to build that type of go to market system for everyone. And so those, I think, would be the two biggest things that I'm most excited about with what.

Speaker A: What.

Speaker B: What I'm doing today and what the future looks like.

Speaker A: Okay, that's amazing, man. Well, let's do this. Maybe we break it. Let's see how much we could get done, because I know we. We usually cover this in two episodes, right? So you want to show us what you're doing on the internal side first, and then maybe we hit the GTM side or. What's your thoughts on that?

Speaker B: Yeah, so, uh, yeah, so I think maybe, uh, on the internal side. Me. Let me come back to that. I can explain a little bit about it, but I don't have any. I don't have anything particular. Um, yeah, I don't have anything I can particularly share right now. So I think let's. Let's maybe spin on the GTM side, if that's okay.

Speaker A: Okay, yeah, let's riff on that, man. What did you set up? Talk to us about, like, how you. How you set it up, systems used all those kind of things, man. Yeah, yeah, for sure as well, man.

Speaker B: Yeah, I'll share the screen and then let's. Let's dive into a few things. Let's see here. Let me know when you see it. Wrong screen. Wrong screen. One second.

Speaker A: Sorry.

Speaker B: Okay, let's try this again. Cool. All right, let me know when you see it.

Speaker A: Coming up. Thinking. Processing. Okay, There you go. Cool. Cool, man.

Speaker B: All right, sweet. So, um, let's take a look at this. So what I want to showcase, uh, here is, uh, um, using AI agent mode. So I will say, um, it is pretty good. For those that are just getting started or using it, um, I would recommend you do some testing. I think from Ryan's and I perspective, it's okay. Uh, I like to speak openly and candidly about it. I think it's okay. Uh, but I think it provides a lot of value, and you'll kind of see that here. So in this first Example what I'm doing is really I wanted to do um, some in depth analysis around, you know, an area. So if you think about your go to market strategy, if there's an area or a segment or something that you're trying to look at, um, and then come up with some business opportunities based upon uh, demographics, the gap in the market, consumer demand, things of that nature. And then what I asked it to do was create uh, a detailed business plan, a P and L for each of the businesses and like an investor presentation to present. Um, and so what you'll notice here, uh, is it worked for a little bit and just wanted to confirm, you know, something that, you know, hey look, I'm going to confirm this is what you're looking for. I said yes. And then what you'll notice is, is it built out a full market analysis. It gave me its methodology, so the geography, time frame, it gave me sources and limitations. And then it really broke down kind of the demographics like I mentioned. Um, it kind of then talked about the businesses. So showed case like your market gaps and consumer demand. Here's the top five business opportun that it kind of showed. And then ultimately let me, I'm gonna scroll down here a little bit further. It took about once I once it kind of built that out. Uh, I said hey, okay, are you want me to build out the, you know, review this. You want me to build out the business plan, the P L statements and the investor presentation? Which I said yes. And you can see it worked about 25 minutes. What's cool again for those is is you can actually click on this and it will uh, actually walk through kind of everything that it did and kind of that virtual machine that it spread is spun up because when you use agent mode, it's actually spinning up a virtual machine and it's kind of going through the process. So if you want to go back and kind of look at what it did, you can. And then uh, what it did again, it built out this P and L and you can see here some of the financial assumptions. So I could easily kind of put this into uh, you know, a document of sorts if I wanted to. So uh, it did that for all of them. And I'm just going to scroll down here to kind of show you the final. So then it built a spreadsheet for me or sorry, presentation here. So it kind of said, hey, here's the investment opportunities, you know, with your market snapshot. Here's uh, the different areas that I ask it to look at around the area here's the different businesses with the concepts and kind of our, you know, you know, year three revenue, net income, not, um, really sure what just happened. There we go. Um, then it kind of went and broke down a little bit of each one of them. Uh, and then it built out a uh, spreadsheet here around the P and L. So this is not really an advanced spreadsheet. I could reiterate on it, but ultimately it built all that out. So you can kind of see where, you know, where this is headed. Um, and to kind of show you like a real example that I used internally is, um, we have this product called One Packs. And so One Packs is a pack system for radiologists. And um, and so uh, the company I'm with is PE backed. And so ultimately we purchased this, um, you know, we did an M and A with this uh, you know, product and, and said, hey, okay, we think there's a market for it. Um, you know, what is that? And so basically if you take that same concept I shared, what I did was go through a T and now that I know the product, I use the product and then I went through a TAM analysis, then build a go to market strategy, et cetera. So there's so many impactful ways that you can use, um, you know, chat GPT, whether it's AI mode, whether it's deep research or AI agent mode, deep research, et cetera. For, you know, any of the leaders out there, anyone who's trying to set up a business or those that are running businesses today.

Speaker A: Love that, man. I think it's great and I agree it's always good to look at because we spoke about this. I wasn't blown away with the results from Agent mode that I experienced when I tested it across multiple different use cases. But the part that you're talking about, well, first of all, one is OpenAI has classically been, as of today, the leader always over everyone else in terms of what they're doing from um, at least just pushing the limits of the models. So even if it's not amazing today, I could see it, it becoming good. Right? And I love the fact that you had to do three unrelated tasks of like market research to a P and L report to, you know, the doc. Like I think those are great. Ma'. Am. What about the one where you source the candidate? Like that was really interesting and a good use case too of like, you know, you had a candidate you need for someone use Agent mode to transource that. Like, would love to hear what your thoughts are on that.

Speaker B: Yeah, absolutely. Let's. Let's show you. So, um, so again, this is another use case of agent, uh, mode. Um, and so ultimately what I did here was I just said, hey, I've got a job description. And I copied the job description from LinkedIn here. Uh, and I said, please search, uh, and provide me the top, the best 10 to 15 candidates you can find in Nashville. The reason I selected Nashville, um, you know, not only did I just, you know, move here, but we're setting up an office, so we're looking for talent in the Nashville area. And, and so I was kind of specific. And then what it did was you can kind of see it worked for 15 minutes here. And again, same thing. You can, you can click and kind of see what it did. I think it's actually thing. It's kind of interesting, um, if I play here, is it. It actually you can kind of see it go through and look at different people's profiles, um, you know, and kind of some of the things that it did. So I thought that was, you know, kind of interesting again, to kind of watch through it. But ultimately what it came up with was it looked at that, uh, they looked at that job description and then it basically gave me back candidate. Here's what their current role and why it aligns to our requirements, and here's kind of the evidence and notable details. And so kind of just broke this down for me. Um, and so that way it gave me a very quick and easy, I mean, think about how hard it is to navigate LinkedIn. Um, but ultimately it gave me a quick and easy way to kind of just have a list of people that I could just go and look at and see if these are people that I may want to proactively reach out to for a role, etc. So I thought that was a, a pretty good and simple, you know, work for 15 minutes. So it was a very simple prompt. Like none of these. I try to show things that are simple as far as, you know, complex, not. Not complex prompts. And, uh, you know, so you can kind of see how simple this one was to just, you know, give actually pretty, pretty good impact, uh, here as far as what you, you know, taking time away from your day, uh, and allowing the agent to do the work for you.

Speaker A: Awesome, man. Uh, what did you think of the results? Did you like when you went through them? Are you impressed? Not impressed.

Speaker B: Yeah, so I think. Oh, yeah, I think overall, so. And this goes back to the prompt because this was just an, you know, an example. So I think Overall, you'll notice if you look, I've got a senior manager. So I think it did. So here's where I think it did well. It aligned with what the role, uh, you know, kind of, uh, the overall level of what I was looking for with RevOps, experience, etc. What it didn't necessarily align well with. And I, this is where I could improve my prompt is around the level of experience that I was looking for. What you'll notice here just in the first three, you've got a senior manager, you've got a senior director, and you got a CRM, CRM administrator. So very different levels of, of what, of what you're looking for.

Speaker A: Right.

Speaker B: And so, but ultimately what I would say is, uh, you can obviously easily fix that with the prompt. This, um, was kind of a test. I was actually just playing around and being curious about what it would do. And I was actually surprised because I hadn't tested AI agent mode with LinkedIn yet. And so, um, I was actually surprised how, how well it did. Um, but ultimately, yeah, so, so really great results aligned with. I, if I, if I make my prompt a little bit more specific around the experience or level I'm looking for, um, I bet it would, uh, I bet it would return the right things.

Speaker A: Excellent, man. Well, that's, I mean, that's good feedback and I love the way that you kind of approach it. You're like, hey, like, the model didn't produce the perfect results, but here's where I could have guided it better. Cause I think, I think that's where a lot of people fall down, right? They like, like I tried it and does it work right versus, like, all right, well, did I give it everything it needed? And, um, I think that was really good. And I liked, I like the use case too. And I think there's a lot of utility in that. I'd be curious, like, how that stacks up side by side versus just like deep research. Because I've had clients that have used deep research for basically evaluating vendors and, and uh, marketing do that at like a $300 million company. And he's just like, this saved me, you know, saved me days of work that I didn't have the time to do. Right, so I'm with you.

Speaker B: So, so, you know, I'm with you. I do that a lot too. So, in fact, I've done a ton of vendor valuations with deep, uh, research. One of the weekend projects that, uh, I'm going to have maybe this weekend or next. And I'll share the, the results with you. So you can, you uh, can feel free to do what you want. Is what I'm curious in. I think the biggest question people ask is I, if, if I think about the business outcome or what I'm trying to achieve and the jobs to be done and I don't have the internal capabilities or I've decided that it's not something I should go build or leverage a current foundational model or uh, uh, a building a custom AI agent for something like that. What is the right tool for me given the tech stack that I have today? And so one test that I'm going to run is can I feed either can I feed deep research and, or, and I'll use ginspark for this as well because I know, you know, you like me, are big ginspark fans too. Is can I give you my tech stack? Can I tell you what it is that I want to accomplish and can you go find the best three to five vendors that would accomplish that given my current tech stack? Right. Because everyone has a different tech stack and I think that's one of the, you know, so I can evaluate vendors all day long but how do they, you know, who works best with what I'm, what, what my current setup is today I think is one of the missing pieces and I really want to, I really want to see if AI can help, help people solve this.

Speaker A: Yeah, that's a great one, man. I love that use case. I, I did a simpler dumbed down version of that. So actually with genspark earlier today, like my computer's been kind of acted up memory wise and so it's still working good. But I'm like I need to get a computer with more memory. I have too many apps open too many tabs, AI models, whatever. And um, it spun uh, up because I was going to use agent mode and I tried it in Agent mode and it crashed. Right. But then I tried it in genspark and basically it gave me an answer like two minutes and it was pretty good. So I'd be curious like what the results are of that, you know, because, um, I think that's another amazing use case that you're thinking of so. Well, unfortunately man, we're up on time. Where can people find you? Where can they find more about you? And then we'll wrap things up.

Speaker B: Yeah, sounds good. So, yeah, so follow me on LinkedIn. I'm typically there posting um, all the time. So uh, with my, with my buddy Ryan, uh, you can find me there. Just uh, look for Jonathan Moss. And yeah, I work for Experity. So if you want to. If you're thinking about, uh, looking at. Looking them up, um, feel free. I can share a video with you about, uh, kind of what we. What we've built, if you want to use it, and post. As far as, uh, the care agent, what I was mentioning earlier, um, and then. And then also, um, also, uh, I have kind of a. An educational research media, kind of podcast, AI Business Network.

Speaker A: So.

Speaker B: AI BusinessNetwork. AI. If you're looking for, you know, ways to help you along your, uh, along your AI journey, just like Ryan does. I really like giving back and kind of helping people figure out their way. It's changing all the time. This is all new to everyone. And so as I'm experimenting and learning, I will. I love to share that with everyone, so.

Speaker A: Nice, man. Well, I appreciate it. Appreciate you. Thanks for being on the show, man. And definitely, uh, we'll share that link in the notes so people can check out what you're doing with, uh, that agent that you guys built internally. So thanks for being on the show, man. I really appreciate it. It was good seeing you.

Speaker B: Thanks, man. Same to you. Cheers.

Speaker A: All right, and we will see you on the next episod.