Latest / The Marketing Brief / Jimi & Brian Discuss AI Implementation and the New Roles Within Organizations (S2 E15)
Transcript
- Brian: Hello and welcome back to the Marketing Brief Podcast. I am your host, Brian Aquan, and this is a podcast where we get to talk about really anything in the marketing realm and kind of keep it loose here. but I don't talk to myself. I usually bring in a guest. I always bring in a guest. And today I'm very happy to be joined by an old friend and colleague, Jimmy Smoot. Jimmy, welcome into the podcast. Jimi: Hey, what's up, Brian? Thanks for having me. Excited to be here. Brian: Absolutely. You are someone I've known for a long time through a lot of changes that we've seen in this industry. You're also someone that I really I I trust like how you approach things. Like I think you're very much a maker, a tinker, and you you figure things out. So I know we had a couple conversations about AI already, just kind of like phone calls, because I was like, hey, what are you working on? What are you building? how would how how would you like to introduce your yourself to the to the pod? Jimi: Yeah, I think that's that that's pretty on brand. I think you nailed it. So Brian: Nice. Jimi: I must be doing something right, you know, to kind of put that out there. But yeah, I mean, I'm a I'm a software engineer. I've been a software engineer. Actually, when I first met you, even though even though I worked in ad sales, I was a software engineer. I I I'd been writing code since I was fourteen years old and I kind of took this like side quest when I kind of worked at MySpace and a few other places, Brian: Yeah. Jimi: during sales in in the advertising industry. And You know, and and and you know, I've kind of have always been at home writing software and it's kind of funny because you know, I worked in in the agency world and the ad agency world for little while, and I was kind of known as the guy who could figure out the Excel stuff, you know, so like the weird kind of crazy Excel formulas people would come to me before. And it was funny because it was I was like, this is so simple. This is like way less, you know, complicated than the stuff that I do at home. and yeah, and and just so to kind of put an icing on the cake here, you know, just to kind of frame it up. I interviewed for a job at DoubleClick, or not DoubleClick. God, was it was DoubleClick? It was I can't Brian: Clip clip mark? Jimi: no, it was one of the it was one of the the ad networks in that era. I just can't remember the name. And I I walked into the interview and I was I was interviewing for a sales role, but I had Linux installed on my laptop and I couldn't get the projector Brian: Yeah. Jimi: to work. And the the the VP, I forget his name, was just like why do you have Linux on your laptop? You know, and that was kind of like, you know, One of those subtle hints where you know I probably should have been writing code and not, you know, trying to go out and sell people on ad ad impressions. so quite a bit of side quest there. But but yeah, I mean that's that's you that's that's me. I'm a software engineer. I've been interested in and kind of working with AI since twenty fourteen. So quite a bit of time now. I I started a company back then called Vesper, and we were doing It was like an AI assistant for executives. So for scheduling specifically was the was it was the Brian: Mm-hmm. Jimi: job. And during that time, there's a bunch of scheduling AI companies that popped up, you know, over the course of like two years. we were one of them. and you know, it was just like a really hard data problem. But like at the time I I kind of realized that what was happening in in the world in 2014 is that there were these these frameworks coming out, TensorFlow and Keras and a bunch of these like deep learning. kind of you know frameworks that would allow you to like work with the GPUs to to use deep learning to to to do certain things. And my insight then was like, okay, well, I think machine learning, which is what you know AI really was called back then, is Brian: Yeah. Jimi: is ready for regular normal engineers. And it's not like this realm of scientists anymore. And it turns out that I was right, I was just a few years early. Brian: Hmm. Jimi: And you know, since then I'd worked on a bunch of different things. That company didn't work out, but I'd worked on a bunch of different things and build a bunch of different, you know, marketing targeting systems using AI. Weird fact, I was the like the the effective CTO of this company called Outer during COVID. there we actually used GPT-2 and we did a fine-tune based on Wayfair reviews to try to build some insights for the product team, which was kind of a crazy fun project. and you know, so I'd always been like trying to find a way to apply, you know. machine learning or AI, depending on who you talk to and their seniority or how long that have been in industry. and yeah, that's what I've been working on just kind of over the last few years. And most recently, the thing I've been working on is a platform called Assist. and what we do is we allow enterprises to apply AI safely within their company. So what ends up happening with companies that are smaller is it's really easy to kind of you know roll out Claude or you know whatever a copilot or whatever. to your company of you know 20, 30 people. When you get to a hundred people, 200 people, 300 people, 2000 people, doing that kind of thing gets more difficult because you have a lot of questions you have to answer, like how do we manage permissions for 2000 people? How do we deal with the different connectors that people have? What happens when we have 400 different skill files on account reconciliation or generating marketing reporting or whatever that may be, we need to kind of Brian: Yeah. Jimi: create some consistency across the organization. and that's really what we focus on, you know, and the other thing it comes up too a lot of times is people say we want to be agnostic to the model. So we don't want to roll Claude out because Fable's great and then six months from now, you know, GPT six is amazing and we we have to pull Claude, you know, Claude back and then roll like how are we gonna manage that? And what our system Brian: Mm-hmm. Jimi: lets you do is switch and and be more intelligent about how you construct your agents and how you how you would roll it out to your team. So yeah, so that's what I've Brian: Nice. Jimi: been working on. It's been a it's a bit of a wild ride. Brian: Yeah. So is it a layer that sits on top of the LLMs? Yeah. Jimi: Yeah, yeah, yeah. So as layered sits on top of all the LLMs, we can we support both on-premise models. So some companies we work with want to do on-prem stuff, you know, we we did with a heavily regulated company, for example. some people, you know, want to do the major frontier labs, some people want to experiment or use open models for certain types of workflows. And you can pick on a per agent basis, you know, which which model you're using for specific tasks. And you can bundle skills together and tools and all of those things to to to accomplish the task, and kind of right size what you're trying to do. So yeah. Yeah. Brian: Like who do you talk to in an organization to like sell that through? Like what like who is it kind of resonated with within an organization? Jimi: Yeah, you know, it's interesting. We're constantly talking about this internally, right? But it's a it's really like this new kind of role that started to emerge, that I that we've seen emerge, which is it it's it's a kind of a a person who isn't always technical that's inside the company who's responsible for applying AI. And it's usually like an ops role. So Brian: Yeah. Jimi: it's like somebody who previously maybe managing the MuleSoft system, or if the company is kind of in the mid-size, maybe imagining managing zapier or something like that. Right. So not a crazy technical engineer person, but somebody who is more kind of operational minded and understands the company workflows. and some pet companies have like a head of agent ops type role, which is kind of what this is, or kind of a head of AI operations or ahead of AI platforms or that type of thing. and those are the people that we usually talk to. But it's interesting, they don't always report into tech. They report into the the ops team most of the time from what we've seen. So It's it's kind of an interesting new realm. And I think I actually think that there's gonna be a whole industry and a whole job function that's gonna pop up that's gonna be related to this this group of people because you know it's it's almost like in a way it's almost like HR. You know, it's not Brian: Yeah. Jimi: it it's like you know, somebody has to think about like how what kind of skills do we need in the company and then how Brian: Yeah. Jimi: do we construct the agents and how do we get the kind of tooling in in the agents and the capabilities in the agents to to create those skills, which is pretty exciting. is you know, I think operations has been stuck with ERPs for a long time. And to give Brian: Yeah. Jimi: them a new piece of software and this new capability is pretty exciting. Brian: Are are you do you need to get like the technical team, like the IT team involved when when you go in there? Like they obviously probably want to vet it, right? Jimi: Yeah, there's usually always, you know, a few people involved in any any you know sale that comes through. It's usually, as you mentioned, like an IT thing. And that's you know, to set up SSO to make sure the security concepts make sense, that type of thing within Brian: Mm-hmm. Jimi: the company. And then you'll need obviously that operations person. Then you need some kind of usually there's some kind of purchaser, you know, some kind of like senior executive or something who's got to sign off on it. but the good news is a lot of times these companies have some kind of mandate and they're trying to figure out, you know. how to experiment and apply AI in ways to make sense and to construct these teams using AI. So yeah, so that's that's really what we've been able to help with and and kind of, you know, bring those bring that group of people together and and and and execute. Brian: Yeah. It's an interesting idea you bring up that there'll be this whole new set of jobs. And we do see this when there's disruption, technology driven or or whatnot. I I listen to a lot of economic people or economists, I guess, on the on podcasts, and they talk about this thing called the labor lump fallacy, which I thought was amazing because everybody thinks about, we're gonna lose this job, we're gonna lose this job, we're gonna lose this job. But that's a constrained kind of thinking in that. You're not thinking about all the other jobs that are going to exist. And the the the best way to think about it is like we never thought an influencer would be a job, right? We didn't think there would be like an e-commerce manager, right? There's all these new jobs that happen because technology moves forward. So it's an interesting idea to think about what are the AI, what are the AI jobs that are that are being formed right now because all of this because these solutions are coming into enterprise, coming into into companies right now. Jimi: Yeah, yeah. Yeah, I mean, you're you're exactly right. It's you know, this is the thing, like people are always talking about, you know, what's gonna happen in the jobs. The the thing that I the example that I use is ATM machines. So when ATM machines first came out, everybody was like, man, you know, the banks are gonna just shrink to the size of an ATM and all of the tellers are gonna get fired and banks are gonna get smaller and they're gonna be like three people. You know, turns out that if you look at bank hiring, it's up. There's more people Brian: Yeah. Jimi: that work at banks. The people just don't count money, right? You know, it's it turns out Brian: Right. Jimi: that counting money was like not a great job. Like a better Brian: Yeah. Jimi: job is talking to somebody about their business about how you're gonna give them a credit line or helping somebody get a mortgage or something that's a little more impactful than Brian: Yeah. Jimi: you know, counting some money in a window. You know, so you know, and you see this time and time again. You see this, you saw this with agriculture, you know, the ag business, you know, you know, in the 1920s. Most of the economy, mo our jobs would probably be, you know, out in the field, you know, pulling a pulling a plow, you know. And fortunately we're able to sit here and talk on a computer in an office that's air conditioned, about the kind of whatever we're doing. These jobs would be like, you know, if we talked to, you know, ourselves back then, the people would be like, What do you mean? Like, what like what do you mean you don't like hoe the ground every day? Like you you you you type in a computer? What's a computer? So, you know, we just it's it's a failure of imagination and The universe isn't static and the world isn't static and the economy moves and things change over time. And it's just great to remember that, you know, as we go through this transition. Brian: Yeah. Something I've talked a lot about on this podcast is that idea of automation. and it kind of always settled back to like anything that can be automated will be automated. And just recently I've been having these conversations. D do you remember when we were talking about the internet of things? It's probably like early two thousands. Yeah. It's one of those like catch Jimi: yeah, yeah. Yeah, yeah. The lights and stuff, yeah. Brian: catchphrases, but it just meant like, well, you could do it digitally. We're probably gonna do it digitally because it's gonna be faster, it's gonna cheaper. And I feel like we're at that stage with AI of just like. Well, if it can be automated and it's gonna make our lives easier and it's gonna be less expensive, like that's probably just gonna happen in a capitalist society or a global society that's that's just moving in that direction, right? Jimi: Exactly. And by the way, you you probably don't want to do the things that could be automated. You know, this is like this is the thing, right? Like you probably are probably tasks that you do not want to do, you know. Brian: Mm-hmm. Jimi: Like I said, agriculture is a good example. Very difficult work. Thank God tractors came along and made that much easier. Because like Brian: Yeah. Yeah. Jimi: I don't I don't there's there's some people that love farming, you know. We we have a company that that is an ag company and those guys love farming. But they also love just driving the tractor in the field and not pulling a plow with an ox. You know what I'm saying? Like there is there is a difference. And the the great thing about automation is it lets us kind of get to the essence of what we want to be working on, right? So we can kind of choose Brian: Yeah. Jimi: the thing that's gonna make whatever we're doing special, and we can kind of focus on that and kind of delegate the rest into the automation, you know, and there that'll always Brian: Yeah. Jimi: be in style, you know. This is why you could either get you can you can go get a You can get a a cup of coffee from one of those coffee machines. You know, I'm at a we work now. And if I go get an ex espresso, I'm getting it from one of those machines. It's not as good as the guy who, you know, is wearing the, you know, suspenders Brian: Right. Jimi: with the with the handlebar mustache and and in Brooklyn, who's like been, you know, doing this for and it kind of just like does a perfect latte for him. It's not gonna be as good as that. Right. So Brian: Yeah. Jimi: even if the ingredients are the same, there's just something about the experience and everything else that's fundamentally human. That we will gravitate towards and we will still make purchases, you know. and I'm a firm believer in that. So I think that I think the world's just gonna shift to things that are more. Brian: Yeah. Human human experience as a premium is kind of the the direction we seem to be headed. what was the other thing I was going to bring up with you? I I feel like we maybe hit it, we maybe at like the rubber hits the road moment kind of at AI. I don't know if you're feeling this, but I think some signals I've been seeing lately of just, you know, there's there's pushback on AI as a concept, whereas the tools are moving forward. And it feels like maybe the honeymoon period of like every CEO. Every kind of senior executive being like, yes, AI, yes, AI, yes, AI, let's bring it in. Now it's like, okay, let's prove it out, right? Is it actually more cost efficient to use these tools? If not, is that something you're coming across as as you're out there talking to to companies? Jimi: Absolutely. Yeah, we talk to customers all the time. You know, there's a customer we talked to, a marketing agency, and they were, you know, they're like, hey, we rolled AI out to our company, and it turns out that people are spending $20 a shot entering their timesheets, you know, and like that is not a really great use. This is not why we decide to do this, you know. So they're just asking themselves, like, what is the ROI on these tasks that are being delegated? You know, I I think I think the answer is for something like that, the answer is. Find a cheaper model. You know, people don't necessarily need to be delegating timesheet entry, which is a definitely a drudgery test. That's definitely plowing the field with an ox. Nobody wants to do that. But you don't need to delegate that to Fable Five, which is like, you know, giving your timesheets to a guy that went to Stanford and got a, you know, perfect and got a PhD. You don't necessarily in in computer science, you don't need to delegate your timesheet work to that guy. You know, you can pick a different model for that. And really, what people just have to do is like start to think about how to construct these systems that allow them to let people do that work in a way that's cost effective, but and then and then kind of orient you know more complex and stronger models towards things that are more valuable. And this is just gonna take time to do. This is why you need the ops teams, you need people sitting down looking at this, thinking about it, trying to optimize for budget, trying different things. You know, this is the thing that's gonna happen. Brian: Absolutely, completely agree with you. Well, I'd love to always ask my guests a question. the question is, what is your superpower? Do you know what your superpower is, Jimmy? Jimi: Well, so this is a good question. You know, I I tend to be really good at unstructured situations. you know, it it's kind of funny, like when I was like I want to say when I was like 18 and I was kind of like, man, what am I gonna do with my life? I I I like it's almost embarrassing to say. I went and I looked at the CIA website to to see like what jobs they had and One of the jobs that came up was like the clandestine operator job because it's it's just like you're thrown into chaos, you don't know what's going to happen, you're preparing for uncertainty. What's really weird is like in those environments, I really thrive. It's kind of like, you know, when my back is against the wall, I'm able to figure things out. When things aren't very structured, I'm able to figure it out. It's just been a trait of mine for a long time. and you know, while I'm going through it, not necessarily great, but looking back, I'm always kind of like, man, that was fun. so like an example of this is I lived up in up by Lake Tahoe for for three years during COVID and we had a crazy winter one year. 50, I think 56 feet of snow fell. Brian: Remember that, yeah. Jimi: It was amazing snowboarding for the first three months. And then for the last two months, it was digging out. It was it was kind Brian: Shoveling. Jimi: of like you just, you know, and what ended up happening is our house. because we were at like 7,000 feet, our house got kind of destroyed by the snow. We had severe roof damage, we had all of this kind of crazy stuff happen. and it wasn't just our place, it was a a lot of houses in the area. They could even declare the area a FEMA zone. So what was kind of crazy is, you know, coming out of this whole thing around the springtime when things started thawing out, my family and I we went down to the FEMA tents because we were just like, what do we do? We can't find any determined contractors, we can't, what do we do? We went down there and they didn't have any answers for us. They were kind of just like, well, you gotta find a contractor. and by the way, you know, we're only gonna give you like 10 grand, which is like nothing. it's it's it's Brian: Yeah. Jimi: actually pretty incredible what FEMA will pay. Brian: Yeah. Jimi: and you know, and by the way, it's not even 10 grand in cash, it's a loan. So we're gonna give you a loan for $10,000 through the SBA, which you know was great and helpful. and you know, we had to kind of be put in a situation, we had to figure out how to deal with insurance and everything else ourselves. In an environment where you couldn't find any contractors, to the point where I had was calling people, they were they would come and they'd give me a bid and they'd disappear or they just wouldn't show up. This went on for months and months. And my wife and I, we were in Europe. My brother lives in Europe, so we were in Europe visiting my brother, and I was like laying there and I was like, my wife looks over and she's like, you know, what's wrong? And I was like, I think I'm gonna have to do this. And she's like, What do you mean? I'm like, I wasn't I said, I think I'm gonna have to do it. She's like, what do you mean you? And I'm like, me. Like, I'm gonna have to figure it out and get it done. So we got back and I just kind of like went into crazy mode and started. I I figured out how to pull a a permit as an older builder to reconstruct the house. I found subcontractors where I could find them. If I couldn't find them, I learned the job on YouTube. So I did everything from plumbing to electrical to hanging drywall to and you know, what what people told me was gonna take. Around a year we did in about six months. And that was just through just straight written determination. so in that kind of environment, you know, when I'm kind of, you know, forced to figure something out, I tend to do that. So Brian: Yeah. Jimi: anyway, long answer to that question, but yeah. Brian: That's all right. I remember that time. I remember you giving me a video tour of your house and like what you were working on. I totally remember that. Jimi: Yeah, yeah, it was nuts. That was a crazy, crazy time. That wasn't a fun remodel. That was a gun to your head remodel. There's all kind Brian: So yeah. Jimi: of weird things that you learn when you do stuff like this. Like the insurance companies, our our underwriter or our adjuster was great, but the insurance companies you gotta have you kind of have to deal with them in certain ways. And then they will give you money to move and somewhere else because our house had to get gutted completely. We couldn't live there. no kitchen, nothing. but they only give you money for a certain amount of time. And we burnt through about six months of that time. It was like a year, just over a year they gave us. So we burnt through six months of that just trying to find a contractor. And I had to do the rest in six months. So, so with a gun to your head, if you're kind of forced to do something, you tend to find a way. at least I do, and that's I think pretty much my superpower. So Brian: That is it's quite a superpower. Jimi: Yeah, yeah. Yeah. Brian: Well, Jimmy, I really enjoyed talking. How do people get in touch with you if they're interested in assist or if they're interested in I think of you as a thought partner that I that I really rely on? So how do people get in touch with you? Jimi: Yeah, I mean feel free to read you know reach out. You can follow me on Twitter. my Twitter handle is JS4, so J S F O U R on Twitter. I do a lot more reading than I do writing there. but you also could check out assist at withassist.xyz. So go ahead and check that out. You know, we're looking for customers. We're also looking for people who want to join the cause and and and help us out. So feel free to feel free to jump in there. Brian: Dot XYZ, that's a new domain. Jimi: X Y Z, yeah, yeah. We try to be a little different. Yeah. Yeah, I was it's funny. Brian: Nice. Yeah, it's definitely different. Jimi: I you know, I was at a I was at a a builder conference, like a home builder conference yesterday, just you know, talking to customers and stuff, and I was wearing these green, neon green shoes. And I got so many people that were like, Hey, what's up with those shoes? And I was it's I just try to be different in ways to kind of pop a little bit. what the it's the Seth Godin purple cow concept that try to apply that in real life. So Brian: big fan of Seth Coden. Well, it's working. It's working for you. Jimi: Yeah. It seems like it. Yeah, let's go. Brian: Well, thank you so much for being a guest on the Marketing Brief podcast. I'm your host, Brian Aquan. Our guest has been Jimmy Smoot. And to all the listeners out there, stay curious.