Ep. 86: He’s Using AI to Double His Agency Without Hiring

0:00:00Who is Phil Rivers?

Play from 0:00:00 0:00:00 What's up, everybody? Fast Hours Podcast, episode 86. Drew Brucker, Rory Flynn, and we got the ~ future Hall of Fame inductee in the NFL, Phil Rivers, here with us today, who has made an awesome pivot into the AI world. So I can't wait to hear all about that one. ~ but in a on a serious note, kind of serious note, So email marketing savant. What else did you want me to say?

Play from 0:00:26 0:00:26 Nerd? You wanted me to say mouth breather, I think was was the other one? I I'm just ~ I'm just a guy that does email and SMS and I make people a a heck of a lot more money than they pay me. That's the TLDR. I like that. also the sales pitch. Also the sales pitch. simple. It's simple. But yeah, I'm honored to I'm honored to be here. Ro Rory's been big timing me for two years. This and that. Finally I've earned an invitation and I'm just honored.

Play from 0:00:53 0:00:53 Alright, so if we're gonna if we're gonna go with the big timing thing, just for just for our audience's knowledge, I've been fired once in my life. It's the guy sitting right there. So Dude. Yeah. But but what he what he neglects what he neglects to mention is that ~ he was fired because the business ceased to exist. So I technically I was fired as well. It's not like, you know Never never been laid off, fired, whatever you want to call it, it's still fired.

Play from 0:01:22 0:01:22 But I was I was a scrappy freelancing sort of like just psych psychopath. Yes. One of those. And Phil Phil hired me to do to run the social media off Craigslist. That's right. We're we're dating ourselves. ~ know I don't even know. Are we is that a real thing now? Like is that a real story? Off Craigslist? we met off Craigslist.

Play from 0:01:48 0:01:48 I posted a job for a social media person. This is twenty sixteen. Yeah. Yeah. Yeah. So so Phil Phil wined and dined me. He took me to he was, you know, he was probably big big ballin back then. They had he had like some like ~ membership to this club and it was like this it's called like spring eleven or something like that. ~ that was Maggie? All right, whatever. But still I showed up and you were like, you know, Phil's maybe, I don't know, four years older than me, three years older than me, and he's taken me to this.

Play from 0:02:18 0:02:18 ball in place and I'm gonna run social media for their their supplements brand. I'm like, this is sick. Like, whatever. But it lasted, I don't know, like four or five months. Then he's like, Yeah, dude, we're gonna we're just gonna shut this thing down. I'm gonna go do email marketing. And I was like ~ what happened was I started I was new to running ads and I wasn't bankrolled by anyone, so it was all self-funded, but I was playing the game as if I raised money, but I didn't raise money, so I didn't have enough runway to get like the payback period to like figure out the payback period for the LTV.

Play from 0:02:50 0:02:50 And so it just sort of died on a bime, unfortunately. And also if my social media manager had been better And lived up to the expectations in which they communicated during the interview, maybe we would still be around, but you know. grew that channel. I grew that channel with no assets. I got a f I got a file of pictures. we he grew it to a hundred followers in Kume. four thousand in four thousand in like four months with absolutely no like there was nothing.

Play from 0:03:16 0:03:16 You guys had nothing. I was making stuff up. Doing like ROI, Rory. Otherwise it doesn't matter at all. The direct ROI is dead. I th the direct ROI was a thousand dollars a month. I was well you know, I was ri the ROI was there. It wasn't it wasn't like if Phil was paying me a full time salary. I was a it was basically an intern. So attribution doesn't count. It's gotta be direct attribution. regard regardless, full circle coming back around, Phil then, after I don't know, maybe a year or two, was like, You want to come do email marketing?

Play from 0:03:45 0:03:45 I'm like, I don't know anything about email marketing. And then he's like, Well, you can sell. I'm like, I've never sold anything in my life. He's like, Well, we'll figure it out. You just sign up and you just, you just come do this with me. And I was like, I don't want to do this. you had been selling at the time. It was, it wasn't like real selling, though. It wasn't like I was selling, you know, I was doing launch material, which if anyone's not familiar with what like a launch is. It's basically like the there's a sort of internet business model where you like hype up this event, right?

Play from 0:04:16 0:04:16 it's still alive. Look at look at how curated and cool that is. I will I will say for 2016, there's an evil Knievel post on there that I liked. this is a time capsule of twenty sixteen right here. Wow. twenty sixteen, that that's upper echelon, I would think. Look at look at your look at my boy right here. Wow. Hold on, wait, click on click on that Jordan picture. There should be ~ there should was there No, that's not the story one.

Play from 0:04:46 0:04:46 Does back one It's a good one. Found that one. There was some Rory right here. This is Rory's MO, tacos and beer. yeah, that's what it was. And ~ just for anyone that was ~ I I basically I I basically not pulled the sheet up, you know, pulled the the the rug f out from underneath, but I had started a dummy page where that was how I grew a following. It's called Tacos and Techno, it's basically an aggregator page of of music and food.

Play from 0:05:14 0:05:14 And ~ I got it to like ten thousand real quick. is I I'm seeing you some some things are making sense now, right? I've seen these Ray videos, I've seen like the taco stuff before. I feel like there's some commonality, some threads that are taking place over chapters of your life. But regardless, it was a dummy page that I grew to just show people that I could grow a following real quick. And then ~ you know, sort of like roost Phil over there into giving me a job and then he fired me and then he came back and hired me again.

Play from 0:05:46 0:05:46 And then then it got cool. But back to the launch thing, by the way, just for yes, yes, for the sales sales piece of it. A launch is when you just hype up this event for whatever where you could give away a lot of free information. And then like the whole goal after the event of the free webinar is just like Gangbusters sell, sell, sell, sell. So it was like I take 10 calls a day for two weeks and just like try to sell as many either accelerators or packages as you possibly could.

Play from 0:06:14 0:06:14 So I was doing that, but that was all basically teed up. That wasn't selling agency services and actually having to know what I was talking about. So that's when we got into the the email marketing stuff. But Phil and I, you know, I would I would give Phil some credit for me on starting to understand how to be a not a Not just a founder, but just like someone who runs a business because I had noticed how he was doing things and it was radically different than the other companies that I had worked for, where it was super data driven.

Play from 0:06:43 0:06:43 It wasn't emotional. There was no like emotions in it. It was just like pure fact, pure, ~ you know, and always like if if if you're a tinkerer, like it felt it f it felt like Philip was like a tinkerer, like the operation was never sound. So it was always something that was moving. Yeah. Yes. like, you know, from post Tetra days, like then it was like you try nothing. You live in a box and even if the box is decaying around you, you don't you don't try and make any changes.

Play from 0:07:12 0:07:12 So like that was where once when when Roy and I started working together the second time and it really turned a corner, it turned a corner because every day we're just talking about like what's happening on the front lines. And what do we have to how do we have to adapt to like meet the market where it is? And it just, you know, clicked. Yeah. Kobe it was like you know, Kobe Shaq like Jordan Pippen for a while where it was like you were running yeah, you were running point and I was just like in the trenches.

Play from 0:07:40 0:07:40 But you know, there's not a lot of businesses or a lot of ~ you know, owners or bosses or whatever you want to call it that would just take what I was saying and saying, like, this is exactly what I'm seeing every day on the phone. This is like whatever we're doing here is not working and we need to change stuff. And you were just like, Fuck it, change it, let's do it. ~ so you know, it led it led to us being good. Like we got good at stuff. We got good at making offers. We got good at selling. We got good at build yeah, fast.

Play from 0:08:05 0:08:05 So no just a little history for everyone here, maybe for Drew too, who didn't know some of this stuff. But just a little re we'll we'll keep the reminiscing to a to a minimum on this. But it's now Phil has basically which is why I think everyone for everyone here to understand why Phil is here. Maybe not someone that's on the creative AI front, but really on the agency operations front. Like we If you're looking at what Phil has done now, it's basically why I think it's a cool story for us is Phil built an email marketing agency years ago.

0:08:16Why rebuild an agency with AI?

Play from 0:08:39 0:08:39 That there was a lot of, you know, some turmoil towards the end of it with a with a weird acquisition, some some crazy stuff that we don't have to get into. Not nothing of the agency's fault, just a really random predatory sort of POS that kind of came in and then ruined things. Yeah, just business. But built the same, almost the same agency, but with AI now. So, like it's like you're running basically a time machine against each other, like what it's like now versus what it's like then.

Play from 0:09:05 0:09:05 And he's done a lot on the operations side. So just wanna, I think it's it's cool to go into this stuff and see sort of, you know, because we talk a lot about the the creative side and some of the personal ops that we've done, but not someone who's building infrastructure for their entire team to use. So just component in that sort of setting is different. It it's it's different than anything we've talked about here. I think it's gonna be a really, really unique and interesting episode in that sense.

Play from 0:09:33 0:09:33 Philip, I did a ~ I did a quick, you know, ChatGPT research pull for you in terms of your, you know, your snapshot here. I'm happy to read it off if you want to check for accuracy here, but Philip. it, dude. is a New York-based entrepreneur marketer, founder of Growth Point, focused on e-commerce retention marketing, including email, SMS, lifecycle automation, customer value.

Play from 0:09:59 0:09:59 is credited in an interview with two business exits, hundreds of clients, substantial attributed client revenue, spoken about agency scale, founder burnout, near shore hiring, limits of AI-generated marketing. figures such as $500 million attributed client revenue, 15x ROI, 500 plus clients, almost eight figure agency scale. then it goes into growth point, email grow, near shore recruiting talent venture, legendary media, LLC, something else.

Play from 0:10:27 0:10:27 pivotal moments in chronological ~ chronology? Is it chronology? Chronology. I I I don't know why, but I've never had to say chronology before, just chronological. Okay. So chronology. so it goes post, you know, post-college stuff, software company role, early entrepreneurial phase, broad agency services, narrow toward email and retention, ~ scaling phase, burnout phase, two exits, and then starts to get into your public views on AI, gen, copy, agency expertise, creative performance, and quote unquote lazy marketing.

Play from 0:11:01 0:11:01 Pretty spot on. There's more, but I'm not gonna I'm not gonna read the whole thing. But that there you know. Yeah. Pretty emphasis on five hundred million in ~ revenue driven for clients. Emphasis on that. It was receipts. the hey. Yeah, also like and this this AI sort of journey started back when Phil and I first round. There was a lot of things that we started it j everything just came out. We were we were early on it, and that's where why a lot of you know, we're sitting here.

0:11:15How an email agency first used Midjourney

Play from 0:11:29 0:11:29 That's why a lot of you are listening to us. Cause again, Phil was just like, do whatever you want. And it was like, All right, it's like, let's figure out how to get our designers on Midjourney so we can start designing some emails better. Right. Like it was simple things like that. Like one of our designers, Steven, shout out to him. You know, he was one of the first ones that took to it, was just like, you know, like what do I need this for? And then it was before you know it, if you know, I probably can pull some up towards the end. We have I have all the first emails that we ever did.

Play from 0:11:55 0:11:55 With Midjourney version four. and yeah. I help set the stage for you guys a little bit and then you guys kind of interject here? So you guys were together at a place called Tetra. Agency focused a lot on email marketing mostly. Is that okay? Yes. And this is, you know, like a lot of the earlier episodes that we talked about when Rory kind of explains his first moments with with AI, visuals, that it goes back to this moment.

Play from 0:12:24 0:12:24 And he talked a lot about. You know, how they started to play around with AI tools capabilities, with the with the understanding that they they might be able to operationalize, scale these things, become more efficient, ~ you know, eliminate some creative compromises that maybe they they had that they might be able to resolve by introducing AI. And so that's the moment we're starting with here. You two work together at that particular agency. So in those early days, what you're saying, Rory, is like you know, Phil and and maybe others, right?

Play from 0:12:54 0:12:54 Like gave you a little bit of that creative freedom to explore those options. Where, where did you guys start with that? And and what did that look like by the end of that experience? Well, in the beginning, you know, Rory's most so focused on the sales side at the at that company, right? But he's always been creative in nature, going back to Tacos and Techno, my Instagram feed that you saw before for the supplement brand. Like he is creative in nature. ~ and but he also has the gift of gab and can communicate better than most creatives can, right?

Play from 0:13:25 0:13:25 So, at the time we're all so we're talking pretty frequently every day about the sales and then when the AI started coming about, Roy and I were just talking about it, just as two nerds, right? And then we're like, how how can we start to leverage this? And then Rory's just naturally inquisitive and creative and starts playing with, and he's like, Yo, Phil, I think I got something. We can start to use AI for especially the clients that are problematic with respect to that we have limited assets from them.

Play from 0:13:52 0:13:52 And you'd be surprised, Drew, a lot of brands don't have that much to work with at all. Yeah, yeah, yeah. kind of regardless of their size. And so he's like, this could be a resource that we could use for all say supplement or or you know, creative supp supplementation as it relates to the emails for backgrounds, creative components, things like that. You you like back then the product related stuff, we couldn't even go, you couldn't touch with a 10-foot pole because it was so bad. So all the only thing that we could use, the only thing that we could use the a like Midjourney for was like supplemental stuff, and that's kind of how it started.

Play from 0:14:23 0:14:23 And then once Rory got comfortable with. the the outputs then sort of expose the design team to the tool. Cause at that point everything was so it was like everything was so new. And then we just sort of gave them some guardrails with respect to like this is how you use it. This is how this is what you can use it for. So they didn't go too far into to the deep end with respect to getting lost in the tool and trying to master it. But then that affecting their output of their work because they got slow because they were trying to figure all this stuff out. It's a good point.

0:14:24How do you adopt AI without slowing down?

Play from 0:14:49 0:14:49 So anyone that recognizes her has been around Drew or I for a really long time, those prompt formulas, that was stuff that we used internally. It was like, here's the formula, just go fill in the blanks, right? Like you don't have to go do anything. So it was like that's how we got people to use it without having to try to figure it out. The other, I think the other thing, when but I think both you and I, maybe it was GPT 3.5, when we took, I'll never forget, we we exported clavio data, raw CSV file, just numbers.

Play from 0:15:15 0:15:15 You couldn't even do spreadsheets in GPT yet. And it was just like Just dump this whole thing in here and just push enter and be like, Do you know what this says? And it was like, Yes, I do know what this says. I can see your click through right for the last, you know, six weeks. And it was like, What it's like, no way. Can we develop strategy from this? It was like, Well, it seems that this and this and this is working well. And these days we're like mind explosion. Right. So it it's ~ yeah, that's it.

Play from 0:15:41 0:15:41 This is just like the very infancy stages, like you see it take hold and now it's like to see where it is now. Kind of crazy. That that was just it's just like at by what by the time that you were at the end ~ of your time with that company in terms of injecting that in? twenty twenty four. Mm-hmm. And it was and it was early twenty twenty four, ~ that we were how much of the process then, right, versus like where you started with started to kind of make its way into AI?

Play from 0:16:09 0:16:09 What were s what were then all of you know, or some of the things, right, that you were really then using AI for? ~ you kind of touched on the beginning of like how you experimented, but in twenty four, like everything was still so new and not even close to what it was today. And then so like if you think about like the bot you know, the bell curve of people or like the team in this case, right? Like it wasn't even on most people's radars yet, just at being the average person. So like we're so at the forefront.

Play from 0:16:36 0:16:36 So We would be then that's one thing from an operator's perspective, always being ~ sensitive to, cognizant of, like, okay, well, what do we introduce to the team? Because it can also have the opposite effect of that we're looking for from an ~ from an efficiency pickup. And because they don't understand the tool, they're not inquisitive, they're not tinkerers, then it makes their job harder than it would be without the tool. Right. So that's one thing we were I we were always very cautious of is like, well, how how do we sort of like work on something R and D?

Play from 0:17:04 0:17:04 And get it to a place where like we know we can just serve it up to for them to use, not have to not have to tinker with anything, ~ or slow them down. And so we really held quite I think Drew held a lot back. ~ because I didn't want to I didn't wanna mess up the operational piece, especially at that time. It was mainly copy ideas or or copy outputs and then cr ~ creative elements, but not even the whole email at the time. So it was like this weird to me though, that that to me is very it's a very mature approach and a very I would say like you're you're in the the very, very small minority of the approach, I would imagine that a lot of people that would have played with it would have taken at that time because of that sort of like new shiny toy syndrome, right?

Play from 0:17:49 0:17:49 Where you just want to do everything and you're you're kind of as a as a leader, you're also thinking about the downside of using it too much already, or spending too much time there. In 2024, which you know, just it was that that that wasn't a conversation in in 24, right? It's just because people were the large majority who still hadn't really got into these tools yet, let alone been playing with them enough to really understand the downside of maybe doing this.

Play from 0:18:14 0:18:14 But I I I think that's a really interesting point because, like for me, my last role in house where I was I was sort of doing the same thing, but I was trying to give my people enough time to to play around and experiment with the tools. Right. ~ I saw the value in it and I saw the value of like education, you know, like really thinking about the education, but I wasn't really so much thinking about the downside of them getting too lost in it, right?

Play from 0:18:41 0:18:41 Or spending too much time where it would affect output either. And so I think that's a really interesting thing. same thing now. So like now, copywriters use the tool, especially for like ideation drafts and so forth, and they critique stuff. And then like I'll look at the logs from the writers and what like what what what sometimes people have the pattern to fall into is just like using the tool as a crutch in lieu of their own creativity. Right. So like a writer, for example, I'll read these logs and they're just prompting this repeatedly the same stuff in the convo about one specific email over and over and fifteen, twenty minutes this log.

Play from 0:19:15 0:19:15 this combo. And I'm like, dude, just stop for one second, think of think about it and just write it. You know, and you would it would probably be done in, you know, 10% of the time if you just did it the old fashioned way. And so that's the it's a it's this weird sort of like thing. It's like, yeah, the tools are great, but there's also like a hidden cost to them. And you don't you don't also have complete oversight, nor nor do I want it of 247, 365 about like what's really going on in the conversations and is it, is it more efficient or isn't it?

Play from 0:19:41 0:19:41 And like a lot of times people also the end in the an individual loses their like they sometimes get they don't have the highest like self awareness or like is this the highest and best use of where my time, energy, brain power is going? Like, you know, if I've been prompting this thing for seventeen minutes for this one email, okay, I I I should have the the presence, no like, let me just do it. Let me just bang this out manually. As opposed to keep going down the slippery slope of wasting wasting tokens and more so wasting time.

Play from 0:20:09 0:20:09 what's what's an interesting sort of like split there too is is something that we've talked about a lot, right? Where it's just it even that is interesting because you've got these these two different approaches, right? You want somebody to kind of like work through something and iterate enough times to ultimately get to the finish line with the idea that, okay, if this is something I'm gonna do over and over again, that's worth doing because if I can figure it out, right?

Play from 0:20:34 0:20:34 There's this compounding leverage of time I need to spend doing this in the future if I feel good about the output. That said, if they're not thinking about this from like a recurring standpoint, right, or solving this problem once and really just, you know, putting their put their foot on the gas to achieve it, then it is going to eat up time because they're going to run through that same process every single time they're doing it because they're not thinking about it from a, you know, a package standpoint.

Play from 0:20:59 0:20:59 Yeah. and the like the nature of one email campaign. So you take like, let's just say brand X, let's just say it's Nike, because you have the swoosh, right? Over the course of any given month, let's just say they're sending two emails a day. Each of those emails is completely unique relative to the one before it and the one after it, right? So the the con the not only the creative elements, but the con the copy of itself and the angle within the email and whatever they're driving to, and then the subject line and the preview text of that email are unique every single time.

Play from 0:21:26 0:21:26 As well. And so if if a writer is let's just say there's six emails going out in a month for for a brand in particular, and they're spending an in an inordinate amount of time on the subject line of one email, it's a very low leverage use of their time. Not that the subject line is important, it's super important, but being able to be being able for the the the tool to know like the contents of the email, the angle, and so forth, and then what's something that's actually creative that can be used in the subject line.

Play from 0:21:55 0:21:55 Like that's going on in pulp culture in the moment, it's very difficult for it to be able to do that, right? And have that thing be aligned with the brand and know. That's why some especially subject lines in particular, like, you know, though that's 160 characters. We have to be super creative and it's super high leverage, because if that doesn't pique their interest, they're not opening the email anyways. And so as opposed to r prompt AI 17,000 times to try and get it right, just think about like, what's what's this brand? What's well aligned that's going on in creative.

Play from 0:22:22 0:22:22 That goes aligned that's aligned with the email itself and the brand that we I can potentially use in the subject line much higher ~ higher value use higher value use of their time and and creativity. And and just throwing in some context here for for our audience who might not be familiar with the email marketing side of things, right? Like a lot of times when Phil and I are talking about this, it it it most of it, majority, is coming from the e-commerce side of stuff, right? And there's ~ almost essentially like two or three environments within an email marketing channel.

Play from 0:22:51 0:22:51 You have, you know, typical automation, which is I'm sure if anyone's ever signed up for a new, you know, for a twenty percent off promotion, you're getting automated emails or welcome emails, you get abandoned carts, things like that. So th that infrastructure is one side of it. The other side is the campaigns, the stuff that shows up in your, you know, promotional inbox every day, which is like, you know, it's fall outside here, you know, check out these certain things, right? You know, so there's there's two little environments in the back end.

Play from 0:23:17 0:23:17 And ~ you know, just so just so everyone is can can like contextualize what Phil is saying, like, especially on that campaign side. Typically, you know, your automation emails, you're not overhauling as much as you are your campaigns, which are going out consistently, or now Phil, you might be able to tell me different 'cause I've been out of the game for a while. smart you are. Most a lot of brands a lot of brands fall into the rhythm of like, you know, set and forget or it's a a box check, but where a lot of I'll say alpha or untapped revenue is in most accounts these days is with it like testing and optimizing things within the flows in particular.

Play from 0:23:54 0:23:54 So ~ that's that. so, you know, again, I I think I think if if you guys are okay with it, I want to go back to Phil starting this new agency. You were just doing you were just doing everything by yourself at the beginning, right? Like you didn't have anyone that that was helping you the new new venture. Like, ~ yes and no, like from a fulfillment standpoint, you know, I'm not writing copy, I'm not doing design, I'm not in the tool. So I wasn't doing the I'll say the fulfillment level ~ at the at the very beginning, but I was building it from zero again with a couple a couple of I'll say, you know, stakeholders based on their respective role.

Play from 0:24:31 0:24:31 ~ and so like the AI had come a long way from the end of that prior agency to spawning of this new one. And ~ I was like, all right, well, now we have all these tools and stuff, or I can start to do all this cool stuff. Let me let me start to integrate this on at the ground level in the very beginning. And so, granted, obviously, like, you know, you copy is an obvious ~ use case.

0:24:35How to build an agency on AI from day one

Play from 0:24:59 0:24:59 But again, just going back to what I just said, it's not like It's helpful, but it's not the highest and best you, and it's not the end all be all in all instances for all emails, right? So it's something like a hammer. Sometimes you can use a hammer for a particular job, sometimes a hammer is not well suited. Same thing for the for Claude for the copy stuff. Sometimes it's great, other times it's it's it's not it's not the tool to use. And so, but really what I was most more so focused on at the beginning of this agency is like, how do I leverage AI from an operational perspective across the team?

Play from 0:25:30 0:25:30 So Not marketing, sure, but more so like operations and fulfillment and all of the communication that goes on internally and externally ~ between the the all the stakeholders that are on any pods fulfilling. ~ and then how do we leverage it to the extent that we can for I'll say comms externally with the client. And so that's really where I've where I've where I've focused and suck my teeth in the most with AI. ~ creatively to a degree, but way more way more on the operation side.

Play from 0:25:57 0:25:57 So right now my main thesis is I want to be able to get you know, the twice, two and a half X, who knows what the ceiling is in terms of the client roster for the team that's affilling it, just because the team is so empowered and pretty much has an extra brain and two extra sets of arms from an execution fulfillment perspective, ~ that we can operate the organization at a very, very efficient level simply because we're using a these tools from an from a operations perspective in all the right places.

Play from 0:26:26 0:26:26 And like starting out, obviously, you know, you did you're much smaller team before it gets there. Was what was ~ what was what was Claude sort of doing for you? What was sort of like the first ~ more operate or maybe a you were always you always had operation. I mean like ~ automations and things in place, which I was like, I I had to ask you questions 'cause I'm like, I don't even know what this means. You, you know, you're throwing out ~ acronyms like I know what I'm talking about back in that point.

Play from 0:26:52 0:26:52 Yeah, I got to connect the DNS to this. I'm like, all right, whatever, Phil. Yeah. ~ but like what was the the what was sort of like that first that first like, all right, this works now. Like this is now part of my business. Asana. Okay. And at the time, like MCPs weren't out yet. There was no Asana MCP yet. Claude Code wasn't out yet when we started. And so at the very beginning, it was like, how do I take all of the there's so many like little minute little things that happen from a fulfillment standpoint, ~ in the interest of like just getting one email conceptualized, briefed, drafted, designed, sent to the client to review and approve, sliced.

0:27:03How can Asana automate campaign planning?

Play from 0:27:33 0:27:33 Waste. into the tool. Like that is the bait, that is this basic order of operations for just one email. Multiply that like times 25 for one client, let alone all of like the one-off stuff that happens in any given cycle. Right. So I was like, what can I do? Like what ~ so my initial sort of like guardrails were Asana. Like how do I make a the environment of Asana and just use what Asana had at the time from like a feature set perspective.

Play from 0:28:00 0:28:00 To make all the stuff that we consistently do over and over and over again as ~ efficient as possible and not have to have some, I'll say super junior person that's just in there moving things around manually. And so the first thing that I built was ~ for any given month, like we, you know, we do campaign calendars for for a client. Well, we'll propose like this is October 9th now, and let's say we're we're right right now, we're doing about to do next week calendars for November and December.

Play from 0:28:29 0:28:29 And so we'll put them all together on a calendar and share with them to review. But then as soon as something is approved, historically they would tell us it's approved. We'd have to go take all that information, memorialize it into Asana or Monday.com, whatever the whatever the ~ task management tool we were using. And that's a like very laborious, detail-oriented, time-sucking, soul-sucking task. Okay. So I was like, I want to not do that.

Play from 0:28:55 0:28:55 And then If you think about just one email and all the steps that I mentioned that have to take place to get the email live sent to the to the world, like setting up the task and all of those subtasks and dependencies and due dates is also very cumbersome. So I was like, I want to have none of that because it's also very ripe for mistakes. And that's the other thing. If you make one little mistake, the the client that they neglect to realize all the other 99.999% of stuff that went right, and they just see the one thing, right?

Play from 0:29:23 0:29:23 And so there's all these little landmines. buried throughout the entire process that one little thing could go wrong. And so like how do I minimize this to the extent that I can? So what I built was let me show you an example. Let me show you the calendar itself. great. This is great because you know, like the I I love that that is a great use case, right? It it's the laborious, like thankless sort of things, but are all they they matter, they have to get done.

Play from 0:29:51 0:29:51 And us like these project management tools, Asana, Monday, you you could even throw Notion in there. I think for like us ADHDers, maybe even outside of that, too, it feels good to just move things around and up. It feels It feels productive, but like in a lot of ways, it's not productive because you're just moving things around, right? And it's like, I did all this. And like again, like it has to get done.

Play from 0:30:18 0:30:18 But it's like, if you've got a like this is to me, is such a great, one of those great sort of use cases across the board. And I'm always looking for things like this. get way more I get way more ~ upside or efficiency from this stuff that we're going over than like the cloud creative out or the design output from Figma, for example. Right? and white. It knows what it needs to do. You've got the you've got the objective data with dates, right?

Play from 0:30:44 0:30:44 Things moving and it can go and do its thing. Yeah. and it just it and it mitigates all of the you know the amount of things that could go wrong because there's a ton of there's like a hundred little tiny things that could get messed up. So here for example, so I was like, all right, if I'm zooming out, I'm isolating the calendar sharing process with the client, getting their feedback, and then once it's approved, having that be memorialized in a sauna in the in the in the specific project of that client, with all of the stakeholders that do fulfillment for that client saved at every role, and then the the the The timing of when this campaign has to go live, T minus X days, so that we know that we're always ahead of the calendar, never or seldom playing with borrowed time to the extent that we can control it.

Play from 0:31:24 0:31:24 And ~ getting things done and scheduled and you know approved and scheduled seven days in advance of it ever being sent. Cause then I have time abundance from an op on operations perspective, knowing that not everything will always go according to plan. But if I if I can control, if I control if I control what I can control more often than not. The the instances in which something goes awry or needs attention because it's last minute or a mistake, then it's w very simple to deal with because the overwhelming majority of all the output falls within the process and we're ahead of schedule.

Play from 0:31:53 0:31:53 That's really what I'm trying to solve for throughout with all of this, right? So here, this is an example like of an email, right? Feature product, the type of the education, the audience we're sending to, the channel, the send date, the links. the our angle for this email, our rationale, any notes we have, and then the client can come in here, give us comments and post the comment. Right. And they can post the comment and just approve it, or they can ~ post the comment request changes, and then it'll be like this.

Play from 0:32:21 0:32:21 Well the team will the team will get a note in Asana saying like, yo, client X left a comment on this campaign in the calendar, go finesse whatever has to be done in the calendar, let them know so they can click approve. And then once like this email, once approve is clicked, this is just demo. But once this is clicked, this just fires off a slew of automations internally, essentially sending this campaign information. So like getting the size right, this email that will be on November 15th, hypothetically, in Asana, in the project of the client, with the title of the camp be campaign being C this would be the fourth campaign of the month, C4, the date of the send date, which would be 10 1115, the title, which is get the getting the size right.

Play from 0:33:05 0:33:05 That's the task title. And then all of the all of the subtasks with all of the assignees and all the due dates so that we're seven, 10 days ahead of this is locked and loaded, ready to send on the eighth. And it's all hard-coded in there with all of the stakeholders for each team. So if I I have like four teams right now, each has a different account lead, a different project manager, a different copywriter, a different designer, different deployment person.

Play from 0:33:30 0:33:30 But all of this is stored in the in the super base. So the system knows. the the the the the team member that's on any any client or in particular and it's just it just runs. So this think of all the all the human error that I just took out of it because the whole process end to end is standardized and then the team can just go. What what was the was this like like how many iterations to get it to this to where you're like, this is scaled across multiple teams now, j not just for this one pilot that I'm So this ~ this calendar was like one of the first things I built.

Play from 0:34:00 0:34:00 So this this version we're looking at now, this is the newest version. The other version worked the same, it just wasn't as smooth visually. ~ but fundamentally fundamentally worked the same. And I think we I've been using this for fifteen months, pretty much since the very beginning. Okay. So so this I mean, this is like a a one of those like great examples of it's like it's worth tackling and it's worth like going deep on to refine, iterate, to get it to a point where just like it feels top notch.

Play from 0:34:28 0:34:28 I'm I'm cu I want to give, sorry, Drew, I want to give the clients the best, most seamless experience possible, right? So it's fundamentally like, yeah, I'm doing email and performance is performance is a factor like driving performance and a return relative to what they pay us to do it. But what I care more about is like this is a service business, fundamentally, right? Like I've lost clients that we have A plus results for and C plus service.

0:34:36Why service keeps clients longer than results

Play from 0:34:54 0:34:54 And I've kept clients where we have C plus results and A plus service. over the years. So I'm like, okay, well again, going back to what I I used to play baseball like you guys did. I was a pitcher. One thing my dad always told me was like, you know, just worry about what you can control. Once the once the ball leaves your hand, you don't worry about anything because you don't know you can't control if it's gonna be a ball or a strike anymore. You don't know if it's been gonna be a ground out, a bomb, whatever or a home run, ~ etc. So I I'm like, all right, what can I control in the process?

Play from 0:35:20 0:35:20 What I can control is how do we how do we what are the optics of what we do for clients and what we share with them and giving them I'll say ease of use accessibility to give us the feedback so that we can go do just execute and be fast. Like efficiency and speed is also a huge factor. Whereas before, this was all done in a spreadsheet manually. Every campaign had to use to think about it, put it down, select select a segment, format it so it looks good, blah, blah, blah. But it took a it took an inordinate amount of time to get the stuff out to them to be able to and then it's friction filled for them because they have to give feedback and then they're emailing back, this is approved or this isn't approved.

Play from 0:35:52 0:35:52 And it's just not a very cumbersome process. So I was like, How do I strip as much out as possible and just make it actually pleasant for the client to give us what we need to to execute for them? Yeah. One of the also one of the crazy things going back to our our initial run at this, not Phil's agency now, is that like we used AI for a lot of the data sort of to identify problem areas. I remember one of the things we found was that people who we had like delivered first drafts to within two weeks, like we had an like an average of like a two month longer LTV on.

Play from 0:36:21 0:36:21 And just like the amount of negative communication was significantly less because we had delivered right away. And it just like set the tone for the rest of it. So like something like this is how you get to that point. Right. Like this is just Yeah. I I think that's another thing, right? It it seems like, especially a lot of the clients I talk to, it's like the the AI component really allows some of these roles to stop being so reactive to just requests, right?

Play from 0:36:48 0:36:48 And then finding a way to integrate AI where it alleviates some of that reactive nature. And now they can actually be proactive, which is not only great for them and the company, but like just from a career standpoint, like career. Creatives wanna be creative. You know, writers wanna write. You know what I mean? Like, and when you're just reacting under a constraint, it's a different operation than when you have the ability to think through it with a little bit less pressure and a little bit more of a free flow state of mind as well.

Play from 0:37:18 0:37:18 Like I think that is also really like one of those common things that I see. I was gonna ask you, Phil, like with something like this, right? You're you're talking about, like, I'm curious, like when you get new clients. or even with with clients that you already have. Like, is there like a a recursive loop or r like a learning element that's baked in where you're able to take, here are all the emails that we've sent that were approved.

0:37:30How do you build client context into AI?

Play from 0:37:44 0:37:44 Here are all the emails that sent like that we we've suggested that were denied, right? Where you've like almost like provided a little bit more of an example objective nature as to what each client gravitates towards, right? To build up sort of this contextual layer. there's that from like their already onboarded standpoint. I'm also curious that from like a new, like a new client perspective, like is you know, do you take maybe a lot of the discovery calls and are there pointed questions that that allow you to kind of like shortcut this process to ultimately deliver more more on brand, more quality, more accepted, more satisfactory sort of ~ engagements that you have with.

Play from 0:38:26 0:38:26 Yeah, so I'll say that on the creative side, like the output, right? Like which is inherently creative with words and designs on it. ~ a lot of times, like let me say this. Feedback will never be zero, right? 'Cause people people always got shit to say. You know what I mean? And a lot of times a lot of times a lot of times ~ clients say like depends on the client and and the team internally and so on, but they just like to say things to say things.

Play from 0:38:56 0:38:56 Right. And a lot of and and beauty is arbitrary. It's in the eye of the beholder. Right. And so what I go back to what I what can I control? Yada yada. And I want to have good good output and give them a good experience. But I what usually so when a client comes on board, they have a body of work that's been approved before in the wild, right? That I can pull from just as reference material. They have, especially the more sophisticated brands, have brand guidelines and tone guidelines and all of these things. and at the beginning it's just ingesting all of that.

Play from 0:39:21 0:39:21 So in the in in the ~ in the data store. We have, you know, it pulled 20, 30 emails that have approved copy angles, stuff like that, just so we have something to to as like a you know, like a fundamental block that we when we start to draft. And then we also pull in like their most recent social stuff or tone of voice to complement whatever they do or don't give us from a tone guideline perspective. And this is all in there referenced for the team, ~ writers in particular in in the writing case, to like when they start to draft to pull from.

Play from 0:39:52 0:39:52 So it's not like just ramble stuff like one in front of zero, but there's other elements on my six one with strange and ~ in Figma, what we'll do, so like we'll have a sandbox environment internally, we have an approval environment for clients that they see stuff that's ready to be approved. In the sandbox environment, the writers write directly in Figma with the with the designers.

Play from 0:40:18 0:40:18 Whereas before, old agency d copy would design, excuse me, copy would write in a Google Doc. Design would then take that, put it in in the design. And so there's a like a fragment, a misalignment with respect to the structure of an email that a copywriter envisions it on a white background with black text versus what's actually going out in reality. And so it's like, all let's merge these two. So it's more so collaborative now than it used to be. But they'll draft directly in there.

Play from 0:40:44 0:40:44 and on the board, we have all a bunch of references. references from the brands that the clients during the onboarding process told us they really like from a designer or copy standpoint, whether they're, you know, competitors or not. And then ~ a bunch of their historical emails. So the writer has like a quick, a quick reference point like as they're drafting to like get inspiration from or have some structure or guardrails. In addition to like, well say like a notepad or whatever in Figma that's like do's and don'ts.

Play from 0:41:11 0:41:11 Like don't never use these words or the you know, this is no go stuff. This is this is go stuff. ~ so we'll use it a lot for that initial sort of like collection, but like it's not in my opinion, it's not sexy or whatever. It's just that just runs in the background, it scrapes it from whatever the the sources, Clavio or whatever tool they're using, and ~ it's just there ready, locked loaded, ready to go. That's unsexy, but that's that's like again, not sexy. I mean, but at the same time, like when they can just open up and write instead of like, wait, what did they do before?

Play from 0:41:41 0:41:41 ~ I don't know what I'm writing about. what are they like? ~ you know, can I do this? It's just like it's just like you're providing you're like providing a nice little environment just immediately. Right. Like it's the again, like my whole thesis, how to give the client the best experience. I know if I do that and we're not completely ~ out to lunch with it when it comes to performance, which is that's hard to do at this point, then I was like, they'll have a great experience they'll be here for a long time. So how do I just and it all goes back to quality of ~ output, speed, and comms.

Play from 0:42:11 0:42:11 How how fast do we communicate either just updates or sh what giving them updates on what we're doing shipping wise? Contents like context stacking for from a client perspective is I think like I don't I don't know that like people are paying enough attention to that. ~ like I think like the knowledge that you have with all of the communications with a particular client and then having the ability for everybody to almost operate that as a you know sort of this mind hive or hive minds, hive mind mind hive, hive hive mind.

Play from 0:42:44 0:42:44 You know, perspective where it's just like you don't lose any of the thing, right? Because maybe a client says something to you, Phil, but like it's relevant for somebody else on your team that's working on the account, right? And it's like rather that be trapped in that one conversation, right? And it's dependent on you maybe contextualizing that with your team. Now anybody could sort of access those things, right? And it's like the ability that you know everything, right, that is kind of gone into that. for the most part there are still something like it's never perfect you know going back to the mind hive you know it's never perfect and because it's about getting stuff in there but also getting quality in there because not not everything is worth saving and some things that are mentioned are anecdotal but then the AI doesn't always know what's anecdotal and what's of what's valuable to to store right and so like now what I what I really focus on is like the account leads that

0:43:20Who decides what the AI remembers

Play from 0:43:30 0:43:30 that that are the main POCs points points of contact for the client. Are really focused on like service and communication and communication internally, communication externally, and speed of comms, and I'll say thoroughness, more so than like being hands-on-the-wheel expert, savant level with at with respect to email or clavial. Like they have to know it, right? They have to they have to like understand the terminology, they have to have to understand second and third order consequences of certain things and have a fundamental knowledge like of the product and and all of its phalanges.

Play from 0:44:03 0:44:03 But ~ they're I they're I they're they're so much more valuable being with high emotional intelligence being able to serve than being just a nerd in the account all day. ~ and and more so being an orchestrator. So them just stilling down like, this is something we need to save in the data store, save it as opposed to relying on the AI to know all things 'cause some stuff clients say stuff is not important at all and it's just a fleeting thing and it's like mentioned, eat it we don't need it again, you know?

Play from 0:44:29 0:44:29 ~ so that's going going circling back to I'm I'm trying to fill in some some context here, and especially because this is where I want to take the next question. But again, back Phil and I in our in our post post acquisition days when we had we found some issues, right? Like we found things that weren't functioning right, especially on the account manager side for anyone who's worked with agency before, and we had significant amount of like complaints over a certain period of time and we were like, What is going on?

Play from 0:44:55 0:44:55 Like this is not normal. Like we're getting complaints about things that have never run this way. And a lot of it was on the copy side. And like copy side, you know, you have good copywriters that you've worked with for a long time. You're like, why are we getting complaints about copy? And then you start like unwinding the entire operation. And it you know, a lot of it was, you know, to to make a long story short, it was the account managers were slammed, and they were slammed because the copy was bad and they had to do a lot of calls and revisions, and that led to multiple rounds of revisions.

Play from 0:45:25 0:45:25 Instead of one round of revision, it was four rounds. Start to compound that over time. You know, their time starts to dwindle and their briefwriting starts to dwindle, which then hurts the copywriter. So it was like this really weird, like cyclical environment of like how it all sort of like devolved into some problems. But then I was like talking to Phil a couple weeks ago and I was like, dude, like, you know, it seems like the account manager is one of the pieces, like one of the variables that you're always solving for.

Play from 0:45:51 0:45:51 So like this is what I'm curious about now. Cause it seems like Phil has tried to, you know, you try to exit a lot of the like. A lot of the like Phil, how do I do this? Phil, where is this? Phil, what should I do with this? Right. So you're starting to build a lot, you know, you have a certain team structure that we talked about, right? You have your account manager, you know, copywriter, designer, deployment. Is that the is that the pod structure? And then like, where is account lead? Okay. huh.

Play from 0:46:18 0:46:18 copy design, deployment. And then I'll have like a couple, I have a couple, like I'll say floating technical resources. So what I mean by that is like people that have been account managers before, but their emotional intelligence or their desire to communicate intraday with clients is not the highest, but they know Clavio well. So they're there as like a complement to the account leads to be able to like, ~ there's some data we need to go find. Go get me the data, distill it for me so I can wrap my head around it and then communicate back to the client.

Play from 0:46:45 0:46:45 So it's really I it's a assistant d an assistant nomenclature doesn't do justice to the role in which they serve the team and and the account leads, you know what I'm saying? ~ but that that's the that that's the stack. That's the team stack. is the account lead sort of like the position you've optimized for the most now with the AI materials? All right. So like like those two, that and PM. Okay. And the the thing you just went through with the Asana tasks and everything, that's that's more so of a PM responsibility then?

Play from 0:47:14 0:47:14 yeah, well, cause the the the calendar is account lead, but then once it gets into Asana, everything gets memorialized, the PM will review all of the campaigns that are get spawned from that whole process I just showed you and just make sure everything is kosher and then off to the races. So so tell me a little bit more about the the account bleed stuff and how you've sort of optimized them from like the amount of work they were doing in the old agency to like what it looks like now. so like now before they would be pretty much the you know the main point of contact for the client, but the main brain from a strategy perspective and the main the main person from an orchestration perspective to the team to go do stuff and then making sure that ~ everything was done, you know, well and and on schedule.

0:47:38How AI unblocks account leads to just serve

Play from 0:47:57 0:47:57 And so what I've done now is like, okay, again, going back to my thesis is like I wanna have more clients on a pod than relative to traditionally what we were able to do without the AI stuff. And I want to to decouple the what de decouple certain things that the traditionally like the account manager would do and have other people do that so that the account manager can be unblocked and do the what's is of the highest value or highest and best use of their time and and expertise, which is serve clients and communicate.

Play from 0:48:26 0:48:26 And so if they can if they can just do that and they have a right hand man or woman to go get do the help with the Clabio stuff. They have a PM that can just make sure the environment of Asana is clean and that the team on any given day knows what they have to prioritize, then they're they're unencumbered just to focus on serving. And so that that's like my main thesis now is just like just ser like do whatever I have to do to give them t access to tools or human capital so they can just serve.

Play from 0:48:54 0:48:54 That's that's that's very noble of you as well. ~ we we we thank you for your service. so like yeah, I mean, you know, it'd be nice to have a boss like Phil every once in a while just trying to help us out. ~ but I'm like, well, their their their behavior, their trained behaviors like do the most stuff. I'm like, dude, I want you to do less stuff. Like if you're doing ~ stuff, I'm like, something's wrong. Right? So either a c the copywriter's not doing their job, design's not doing their job, PM's not doing their job.

Play from 0:49:23 0:49:23 Like I'm I tell them like you should never have to be in Clavio doing anything. Everything should be given to you either by your right hand or clawed code directly. But you shouldn't have to go in there and click buttons. You shouldn't have to go build a flow yourself. You shouldn't have to set up ~ any any like splits or any branches or any profile filters like that's beneath you, right? Tell someone else to go do that. Make sure that when they do it, it's to your liking. But that's it. I want you to do less.

Play from 0:49:48 0:49:48 Like you should have time in your day where you you're you're not doing anything. And and one of the one of the things I want to get into here, because when you said it, this this made I was just like I have like a moment of like, why have I never thought about that? Was the the way you're structuring like skills and things in a virtual machine. I I think it's very unique. I don't I haven't heard many people doing it. And I think this is probably something that a lot of teams are struggling with, is that everyone's operating in a silo.

0:49:52How can teams share AI skills?

Play from 0:50:15 0:50:15 They're like, yeah, I do this, but I have my own set of skills on my own Claude account and like everyone is sort of producing the same thing in a different way and there's no standardization. So like g you're gonna you're gonna have to give us that sauce a little bit. I'm gonna I'm gonna make you do that. I I was very early like I was very early Claude Code adopter when it came out that I had access to plugging all these tools and access information in a more streamlined way, namely Clavia with API and and writing JavaScript, et cetera, et cetera.

Play from 0:50:43 0:50:43 Right. So it's like, you know, you know the saying of the plumber? It's like, the plumber came to my house and he was there for five minutes, but sent me a bill for a thousand dollars. Then I complained about the bill. And he's like, you know, you're only here for five minutes. He's well, it's not the time I was here, is it what I knew I knew what to do to fit the tool to use and what to do to fix the problem in five minutes? That's the same thing. So like you know, one of the one of the the most frustrating parts about every h all of these tools being democratized that that everyone can use them. Maybe they're not writing scripts with Claude code, but they're using Claybio MCP, for example.

Play from 0:51:11 0:51:11 But they just take the output of whatever Claude gives them as gospel without any like any comprehension of like, this is actually not astute feedback whatsoever, but because Claude said it that it must be right. You know, like So dealing with that sort of like inbound is from clients can can be frustrating at times. But to answer your question, Roar, like I was very early Claude Code Adopter. And then I was like, well, I need to start going back to my how we ran stuff before.

Play from 0:51:38 0:51:38 How can I work on stuff in a sandbox environment with me? Cause I'm just naturally super curious, inquisitive, and I know intimately well all of the technical clavial stuff in a way that even team members don't. Right. So how can I use it and like, finesse it so that it's good and then give them something that they can use. So I'm not sending them on a fool's errand every day to like spend hours and hours and hours trying to fit figure this out. ~ and so originally it started with I'll say data analysis reporting, going back to the scales, everything's in the repo.

Play from 0:52:06 0:52:06 ~ I never used GitHub before, I never used Superbase before all this. and then so what I found was that, well, one, getting the team access to stuff hard, right? Just Hard. And going back to like obviously Claude code's matured a a lot over the last twelve, eighteen months, whatever it's been. And you know, the the other frustrating part is adapt it's it's progressing so quickly. So there are things that I'll do, there are things that I'll do to like fix a limitation.

Play from 0:52:31 0:52:31 And then, you know, three weeks later, Claude changes something that renders what I just did null and void or unlocks a problem I was trying to unlock when I w with my gerrymandering. You know what I'm saying? So I would say that like that's the hardest part, but now N what I've what I've ultimately come to is like, okay, we have MCPs, like that is a standard. That will that won't I don't foresee that going away at any any anytime soon. And I need a way to store all the stuff I'm working on in a sa in a safe, controlled environment that only people that I want to give access to can make any changes to it.

Play from 0:53:05 0:53:05 But the rest of the team, they can just use it, right? Like the data store of the emails or the copy preferences, the do's and don'ts that's stuck stored in the data store in addition to the stuff that's on the Figma board. And so like the first version of this was like, okay, we're using the connectors that come with Cloud Code directly, GitHub, Superbase, et cetera. And then Cloud ~ switched up their ~ I'll say the permissions with respect to using connectors. So it used to be organization wide. If I as the admin of these tools granted the org access, every every person on my cloud account can use them.

Play from 0:53:35 0:53:35 Awesome. Right. Then they on a on a win, they turn that off. So I'm like, shit. So then I was like I would but there's you gotta remember, guys, like there's systems that are dependent on this stuff working. And then Cloud makes a change, they don't publish it, I find out the hard way, then I'm like trying to figure out a solution. Meanwhile, in parallel, Cloud's working on their own solution, but I don't know the timeline for that solution, when I'll get access to solution, and so on. And so what I came to was like going back to okay, MCPs are a known known, right?

Play from 0:54:00 0:54:00 That's a protocol won't won't change. And I'm like, I'm w my I was like, okay, let's put everything in like ~ a virtual machine that has a s a snapshot of our repo. It's not even the live repo. A snapshot of the repo. It has access to the superbase, which is the data store. and the team, I'll just create an MCP for that virtual machine, and the team can plug into it via the connectors via Cloud and they can use it. The copy's there, the roles of the team are there, the the connections to Slack are there, Asana, all of the stuff ~ with everything that I built.

0:54:11How does one MCP support an entire team?

Play from 0:54:32 0:54:32 And so everyone can access it now with one connector as opposed to multiple. So less ~ pro less likelihood that something something breaks. or multiple things breaking and it's safe because no one can write to the repo unless I give them direct access to write to the repo. And what's updates there in the virtual machine is I think it's like every three minutes or five minutes or something. It just updates. So if I made a if I if I if I merge a PR in the repo directly because I'm working on something, within five minutes the team will have it.

Play from 0:54:58 0:54:58 I don't have to tell them or I can tell them, but like they there's nothing they have to do. It's just there. So that's that's been also one of the coolest things. It's just like it's secure and it's fast. And it it can you like you're able to tap into it from anywhere. Right. Like it's there's no th you you can take you can use whatever computer, you can use whatever, you know, it's not like a lot of my skills, like my skills personally, right? Like they're on my machine locally.

Play from 0:55:24 0:55:24 Like if I was to go open another machine and be like, open my cloud account, I'm not gonna have access to some of the stuff that I've done on Claude Code because it's on my machine locally. Some of that have yeah. Yeah, they have some GitHub stuff, you but it's not doesn't add you know, it's not I don't have everything on GitHub. But like so if I'm visualizing this, it's Phil orchestrating over here building, virtual machine up here receiving, and then team can tap in whenever they need to and whatever's just it's always sort of like ever producing, but I don't want it I don't unless they have a desire to like work on this stuff, and like they're tinkerers like me, like us, I don't want them getting caught up trying to do it and figure stuff out 'cause then it keeps them from doing what their key objective is, which like the account leads is just serve clients communicate internally, externally.

Play from 0:56:11 0:56:11 You know. is this is something that I've noticed over my over my years of now going inside of businesses and seeing the teams that are doing good with AI and doing well and performing well. It starts from the top. It's like at your level or, you know, right below you, especially in a larger organization. Like it ha like the interest and the grunt work has to be done at the top for stuff like this for everyone else to succeed because it it doesn't flow up. Like for whatever reason, it does not trickle up.

Play from 0:56:38 0:56:38 Doesn't matter how many good people. they don't have the time. Most people don't have the interest. Mm-hmm. And so it's like if you think w it all goes back to incentives, how their compass structure, all these things, but if they have work to get done, but then they're taking th 50% of their time doing AI stuff, but then at their annual review, they're gonna be like, yo, Jim, you were doing way less work this year. ~ well, I was trying to figure out this AI thing. You know, in the in in the in the micro it sounds great, but in the macro, it's like, okay, well, you're doing half your job and that's a that's a problem.

Play from 0:57:06 0:57:06 You know. what what type of thing what type of like systems are running in there? Is it is it just like for the in the virtual machine? Like is it just for the account managers or is it for copywriters and designers and you know like an entire node? Okay. everything. Most of the designer stuff is just in Figma. but the cop everything related to the client. ~ their their I'll say their the links to all their pages, stuff about the products.

Play from 0:57:34 0:57:34 So like we'll take stuff from the product pages and save it in the data store. When they have new products, it'll just update in there. ~ the I'll say the the corpus for copywriting, like their tone guidelines, do's and don'ts, all that stuff is in there. ~ the the pods themselves and all the stakeholders. So like all the roles that fulfill. So like what what the account leads will use will use all this stuff to create tasks. So I go I'm like again going back to efficiency service speed.

Play from 0:58:02 0:58:02 I want the the other big and I told you about this last week, where this the precipice of me being allowed on your show finally, is that is that dude is that ~ I want like I I I identified that like task creation, not the campaign stuff I just showed you, but the one-off tasks, one-off projects is like a very cumbersome activity as well. So like think Drew, there's we have in every client pretty much, we have an external Slack channel with them and their team, and we have an internal channel with our team that they're not that they're not in.

0:58:17Why one-off tasks break agency teams

Play from 0:58:34 0:58:34 But the account lead is in both. So they have to still down, distill down like stuff that happens externally and get it to the internal team. Some of it's like last minute in nature for an email going out today, right? For a change that they made. Others just like bigger projects, but that has to be memorialized in Asana in a comprehensive, well man manicured way so that the team can execute because I tell the I tell the account leads if the if the team that's executing has to ask you one question about your task, the task isn't good.

Play from 0:59:02 0:59:02 They should ask you no questions. And so how do I take the context of whatever's happening externally within these client chats or internally and give the account leads the The ability to efficiently create tasks with all the context necessary in a well-structured way so the team can go do the thing. And if the like AI, Claude doesn't know all the context, it doesn't know necessarily how to structure all the things for because like with some of this clavio stuff, there's second, third, fourth order consequences of things.

Play from 0:59:33 0:59:33 So, like a an example, Roy will understand, it'll be it'll probably be Greek to you, Drew. But like if we're changing a an offer on a pop-up, so you go to a site, there's a pop-up, like 20% off XYZ. And we're changing that offer, there are emails downstream of that that have to be changed as well. And that's like a very simple version. Sometimes there's branching based on a variety of other I'll I'll say ~ filters and and complex and complexities, what have you.

Play from 0:59:58 0:59:58 But the account lead has to think about all these things when they're putting the task together, because if they just put the task, change the pop-up, there's all this other stuff that's gonna break. We're gonna hear from the client, we get fired. Right? And so what I've done is I was like, all right. Like going back to first principles, what do I need? I need an efficient way to get stuff from Slack into a sauna with the context so the team can execute and ask the account lead no questions.

1:00:13Meet Tim, the Slack agent that writes tasks

Play from 1:00:24 1:00:24 And so what I did was to I made an agent, his name's Tim. And Tim is Tim Riggins from Friday Night Lights. Okay. Super, super hot dude. All right. So we got some eye candy in Slack to look at. All right. And ~ we I was trying to keep a straight face there. I yeah, it's it was it it was a little tough. It was a little tough, yeah. I mean, dude, yeah, objectively, he's a very handsome guy. Like, straight up. Anyways, so ~ Taylor Kish.

Play from 1:00:52 1:00:52 But Tim Riggins is way hotter than Taylor Kish. Yeah. Dude, he's so dark and dark and dreamy. Anyways, he ~ so what so I have Tim in there. Tim's in every external chat, every internal chat. Okay. And so I need an efficient way to be able to get, like there's a threat externally to get that information to Tim so he can draft a task, get the approval of the account lead. So to make sure all the stuff is there, put into a sauna again with all of the stakeholders, the due dates, all the everything.

Play from 1:01:22 1:01:22 And so there's two ways that I up until this point, ~ figure this out. Number one, we have a certain one emoji reaction that is a trigger for Tim. So if there's a ~ a thread internally, external channel, internal channel, this reaction is given. Tim drafts a task, sends the DM of the person that dropped the emoji. Hey, this is the draft. Do approve it. Check it. If not, reply in the thread and I'll get it right. That's one way. Those are first, I'll say simple tasks, right?

Play from 1:01:47 1:01:47 Like one stakeholder, like assign this task to this person, do this time. Because there's only there it within the thread, it can't know all of the dependencies that the account manager knows that are going on in the account. You understand what I'm saying? So if the more complicated tasks, instead of the emoji, the account lead forwards it to Tim, gives him the context with Whisper. Right. In when you forward something in Slack, you can type in whatever you want in the box before you send the message.

Play from 1:02:13 1:02:13 So whisper all the context. So think about second, third order consequence. We need another task off of this task because they're connected. Da-da-da. Enter. Tim drafts it, gives us the the task description, all of the subtasks, the the the dependencies, meaning some tasks are blocked because the task before it has to be done before that person can do what they need to do. Right. And the the all the dependencies, all the due dates.

Play from 1:02:38 1:02:38 Everything if there's tasks that are related, like this pop-up welcome flow thing with the creatives, the task of the welcome flow is mentioned in the task of the pop-up, and the task of the pop-up is mentioned in the task of the well. So the team knows all of the all of the stuff. And so I took this down from like, this used to be really a nuisance. And I know this intimately because I lost an account manager. I had to, I had to do it for a while. And I was like, dude, I'm like my reminders in Slack to like remind me to go make a task was like 9 a.m.

Play from 1:03:04 1:03:04 every day. And I'm like, fuck, dude, like I can't do this. I know. so I made this for me, but it makes it so much more streamlined and thorough from a task creation perspective that the team can just go run it. And so now I'm like, my God, I don't even mind. ~ like now I'm on two accounts to like stay sharp and I make them a lot of money and the CEOs are fun to work with. But now I'm like, dude, like account lead ain't so bad now that I solve for this huge problem of synthesizing information and getting a task spawned in a sauna.

Play from 1:03:31 1:03:31 Good to get in the seat and kind of feel it again, though, too, because otherwise maybe you know you're a little bit far removed in a sense. Like sitting back in the seat, you kind of pick back up on some things. Also, really good point there, which is sort of like the second order thinking of using AI to to catch those things. I think that's like one of the more valuable things that it that it helps me with. Like Cause like even even from the beginning, I was all like I don't know about you guys, but I was big into those like mental model books that existed.

Play from 1:04:02 1:04:02 First principle, second order thinking, all the little like raise oxum drackham razor, ox occams razor, you know, like all the stuff. But like the truth is they're mental models for a reason. And it's like with with things like that, with second order thinking and anticipating your needs, finding things for you, it just unlocks a whole nother element of the usefulness with that stuff, right? Because That's where AI is so great.

Play from 1:04:29 1:04:29 It's recalling all of this information, all of this context. And that can help you on the front side. It can also help predict failures and obstacles and all these things that dependencies that could come up later down the road, where, you know, when you're going back and forth and maybe you've got multiple threads open, right? Like you're just speed running threads. You know, you're just like, okay, I'm gonna do this one now while this is going, I'm gonna work on this one. And for it to be thinking proactively for you to catch these things is super, super important.

Play from 1:04:59 1:04:59 Yeah. It's been it's been a a breath of fresh air. Well, this is also to to say it, right? To like to to summarize a lot of this, because you know, number one, what we talk about a lot on this channel here is like domain expertise, right? Like if unless you've run an agency with forty, fifty plus clients, the the stuff that Phil is talking about right now, it might sound it might sound like it doesn't, you know, like that's all that's like overkill.

Play from 1:05:24 1:05:24 It's not. To keep fifty clients organized, to send a million pieces of creative like yeah. clients, five stake five internal stakeholders, external stakeholders, right? And let call it like twenty five units of output per month. Right? So like how how many clients could you have right now without AI as a part of your process? Right now I could have with the current team, I probably have twenty percent less clients than I have right now.

1:05:41How many clients can AI add with zero new hires

Play from 1:05:54 1:05:54 And with no no additional headcount. And but in terms of the the ceiling now, I think we could two X the client roster with no new headcount. There you go. And that's the difference. Cause a lot of times when when Phil and I talked about scaling the agency, we thought, you know, there's there's sort of like a point of like diminishing returns where the more clients you have, you're just gonna have to get up more people. And it starts to become like the head count doesn't start to make sense with the margin, right? Like and then like what you're making at a certain point.

Play from 1:06:20 1:06:20 It was like we got to a certain point. Like, why is this why is this not making sense? But going back to what you just said, right? Like this this ability now to scale beyond with the same headcount, but also because you're optimizing for all the little agency problem areas, which at the end of the day it's n never about the creative, right? It's like how much like it's what is this ninety percent ops that's done that that optimizes it? And you think about it, like how many businesses are like that where it's like it the end result might seem like, the creative's the reason that we're losing clients or the creative is the reason we're not doing well.

Play from 1:06:54 1:06:54 It's like it's all the little stuff. It's all the little stuff. like I was talking with the design team yesterday, and you know Roar, we talked privately about like the the design process and my and and and my whole vision for that. And I'm talking with them, I'm like, actually, like, I don't know if this is a problem we're solving right now, the the the automation of or at least the first drafts, because they're like, when a new client comes in, we we we t make we use we'll use AI to make components and like make the structure of their emails.

Play from 1:07:24 1:07:24 And then so when we have like camp that campaign I just showed you for 10 1115, for example, you know, 20 minutes ago in the screen share, like all the blocks are there, right? All everything they're just dragging and dropping, like, okay, these are the five blocks we need. We're good. That takes 15 seconds. So it's not like an area of like there's tremendous friction or inefficiency from a design to get something propped up, you know. So I'm like, maybe you know, there's from my standpoint, time, bandwidth tokens that are spent do it trying to do something else, not the design piece, cause it's really no there there.

Play from 1:07:56 1:07:56 You know what I'm saying? At this point. ~ as much as I want to. I'm like, I'm all super hyped on it and stuff. And I'm like, I don't know. I think this is just like a I'm just going on a wild goose chase. Yeah. It's a, you know, it's and this is this is what again, like domain level expertise starts at is, you know, Phil can run a single email account and knows how to run it really well and knows the ins and outs of it. How do you run 50 of them but also understand all the problem areas? But then once you understand the problem areas and have the experience of all these things that go wrong on email and how sensitive email is for for our audience who's not familiar, it's a very easy sort of you know, third party agency to cut for whatever reason.

Play from 1:08:35 1:08:35 People love to cut email marketing agencies because it's not as sexy as paid. And you know, there's it's just a lot of like, we could do this internally, we'll try it. And then they they come back and it's like, well, we can't do it internally. You know, you guys are actually right about that. ~ yeah. So there's like all these it's all super intricate, but it can generate 30, 40, 50 percent of your monthly revenue just from actually sending emails. But the the whole point of that, what I'm saying is like to take everything that is super important to understand.

Play from 1:09:04 1:09:04 From a context standpoint for an employee so that they don't have to go and ask Phil what to do or someone else what to do. Like how many like inter everyone here is thinking about this probably right now. It's like how many questions do you get from your coworkers that are just like, you should know this or go here? It's somewhere over here or whatever, right? Like, yeah. the first couple weeks to a month or month plus for some of these people to come on and be onboarded, they're they're messing everybody's time up because they don't, to your point, they don't know where anything's at.

1:09:19Can AI shorten employee onboarding?

Play from 1:09:35 1:09:35 They don't know what tool should be used for what. They have to rely on the, you know, default old school style recommendation, like onboarding system, right? Like I've been in where it's like it's a notion database and it's deep, you know, and it's like I don't know the naming conventions. I don't know what tools we're using for what because we've got, you know, like what's our tool stack? What do we do for this? Right. Like that, dude, that that is such a good point.

Play from 1:10:01 1:10:01 And it's exacerbated when you get on like new employees and activation. And you brought up something too when when you were doing the what it was Tetra, right? Tetra is the name of it. The beginning, right? Where Roy, you were you said something about like we were already thinking about this in terms of like prompt structures and and that kind of thing, right? These Templates, right? Same thing is true there. It's like imagine us banging our heads against against a wall, trying to teach people all the vernacular of photography, all the things you need to include lighting, you know, camera angles, right?

Play from 1:10:35 1:10:35 Like these types of things versus just developing the template or the structure, right? Which has sort of the brand knowledge or visual identity baked into it and can put the prompts in that structure. Now you're activating employees first week versus trying to train somebody with so many reps on any given tool, right? To get, you know, good to great output. It's just a long haul.

Play from 1:11:01 1:11:01 And that and even then you need a curious person and a proactive person to accomplish that versus meeting people where they are and activating people right away. Yeah. know? And I th I think it's like I think like again, ~ because I f I feel like it's sometimes it's hard to maybe visualize for people who d you know, maybe haven't run in the agency world or really don't function in sort of like this team structure, maybe individuals, whatever.

Play from 1:11:27 1:11:27 Like the way that you've built it is is what I just keep racking my brain as to how intense it is from an experience standpoint to have it as here's like Everything you need, you have client knowledge. So if you have to go in and find things about the client, you can just tap in via Claude. If you need to know who's on the team that you have to talk to, that's in Claude. If you need to know, you know, how like where to task or how where things are located or what information is what or previous performance or data or results or whatever, it's right there.

Play from 1:11:59 1:11:59 Yeah. D know. and if you need to find, you know, if you need to figure out ~ you know, what happened six calls ago. Files are the files are in the computer. You know what I mean? Sorry. That's one I haven't heard in a while, Phil. Thanks for bringing that one back. But there. it's it's just like that's that's you're created a digital organization, essentially. Like you created a digital knowledge base for the entire ops. Is this ~ is this what it looks like?

Play from 1:12:26 1:12:26 No, just to visualize, this is how complicated things get the more you add more people. Right? So like, okay, we started it's you know, me and someone else doing copy and design. Easy. Add a project manager, add a copywriter designer, add an account lead, add an external one external client, one one team with all external clients here. This is forty-five lines. It's it gets it gets super hairy super quick.

Play from 1:12:54 1:12:54 Yeah. this is just communication. It has nothing to do with the the the task creation and execution. And then all think like, dude, just for one email, the amount of lines of some stuff that can go wrong sometimes, or people need information, or a link changes, or an image changes, it's like it's it gets it gets complicated quick. There, I mean, you're i I would say the people that are on in in your organization are lucky to have you as somebody that knows so much about the capabilities of AI.

Play from 1:13:26 1:13:26 Cause we've to we talk about this too, right? It's we we were even alluding to the people at the top leading the way, that kind thing. But most companies don't have this benefit. They do not have this person that can facilitate the change management of integrating AI in, which is why like they're hiring. you know, consultants and people that can actually drive the change, right? Like it's just hard to get it. It's hard to get it used internally with my small team. I I can't I can't imagine doing this with a team of a hundred or five hundred or five thousand or fifty thousand.

Play from 1:13:55 1:13:55 You know? It's just like yeah. too, right? So it's just like, you know, you're it it it's a big bite to take and and chew off there because it's just like you these just companies don't have that luxury, you know. And again, it just kind of goes back to the time, the time thing we talked about, we talk about all the time. It's just like people don't have time to learn this stuff when they've when they're they're they've got responsibilities and they're in between meetings and they got a 30 minute gap before their next meeting and they got another deliverable that they need to produce same.

Play from 1:14:25 1:14:25 Like where's the time to learn AI and experiment? There's not. Dude, the other thing I was gonna say too is like, I've I've fallen victim to this as well, and but I'm super plugged in. But like doing stuff that isn't worth doing. Right. So like there'll stuff there'll be stuff I try and do what they have. I'm like, pie in the sky, this sounds great. If we could do this, and then it's I I do it, I'm like, ~ I can't really use this at all, or the team can't use this at all. So it's just like, you know, again, just you know, on the tinkerer side of me is just I'm having fun, right?

1:14:27Which AI projects aren't worth building?

Play from 1:14:54 1:14:54 If I just like, but in terms of being getting value from it, it's like a lot of the stu especially earlier days when I'm just like having fun or just like ~ you know it's the shiny object, it's easy to like not everything we're building is building is worth building because we can build everything now. And so that being able to discern what is valuable to build and what isn't ~ is a task in and of itself.

Play from 1:15:19 1:15:19 Well said. So what are you building next? Give us the tinker. So right now, ~ refine Tim, even though he's hot, looks don't get you everywhere. You know what mean? So that's right. That's right. That's why I got that's why I got Tim when he was, you know, senior in in high school. All right. Don't get us don't get us blocked on YouTube, dude. guys, watch Friday Night Live so you'll understand.

Play from 1:15:45 1:15:45 Let me just show everyone Tim right now. Okay. ~ boy. Just for for comedic for comedic purposes here. Just I'm just just like, hold on, let me find the avatar that we use in the chat. ~ show though, wasn't it? yeah. Dude, this is a b this show is like it it's twenty twenty year anniversary right now, by the way, if anyone wants to watch. I think it's streaming on Netflix. Like, dude, I mean look at this guy. Specimen.

Play from 1:16:10 1:16:10 Specimen. Look at that. that dude telling you to to change the UTM tracking link on one of your emails. Exactly, dude. Dude, the team thought he was a real dude. Like a lot of people, the team's younger, they didn't watch the show. clav, bro. Did you tell my god Dude, a lot of team people a lot of people on the team didn't know who he is or was the character. They didn't watch the show because you the show's 20 years old now. They're like, who's this new guy in the chat? Who is this? Who's Tim?

Play from 1:16:35 1:16:35 ~ yeah. So I'm I'll I'll say re refining Tim more so to make sure like the zero shot of the drafts are just like over the target more. So they're like, even if we have to give one reply to finesse something, it's better than before, but still like inefficient. Two replies, I'm like, I want to get rid of all two reply friction on Tim. so the initial output is good. And then the other thing that I'm working on right now is you saw the calendar.

Play from 1:17:02 1:17:02 I didn't show you the output of Asana because all this sensitive client info, but the output of what comes from the calendar to Asana historically is very cumbersome. It would do some initial drafts. and ~ in like working through it over the course of the year, it's like it's a very low value to the team. And so I'm reformatting like the output of the calendar as it relates to showing up. In Asana, so that the copy and design can I'll I'll say be able to get to work faster in Figma.

Play from 1:17:31 1:17:31 but it really that's more of like culling of information as opposed to adding more or making it different, just like ~ getting rid of stuff that we don't need. That that's the that that's really the big thing. And then the other thing is ~ also I'll say architecture related in Q4. So like I'm not releasing anything that the team will use from a development perspective between now and January, but like. our the process that we're using in terms of the plumbing of the calendar, Asana, Figma is like old, right?

Play from 1:17:58 1:17:58 And it's sort of like pieced together from like V one AI when we had to use NADN and all this stuff, but we now everything is more mature and we don't have to rely on on like the piecemeal of of duct taping all these tools together. So the big project internally is nothing that anyone will ever see besides me. And just in terms of the the the technology stack that we're using for the pipeline of everything that we discussed today. Sick. I always get amped when I when I hear the some of the you know, dude again, for anyone who missed that, like go back and audit your shit.

1:18:20How do you keep AI workflows current?

Play from 1:18:30 1:18:30 Like, you know, I have to do this all the time now because there's stuff that I start on like Opus four point six, and I'll just go, you talk to Opus five five and be like, How does this look to you? And it's like, this is garbage. This is garb. I'm like, it's been working for me for six months. They're like, it's like, no, I'm gonna redo this right now. thing is is key, man. I mean, like even that, right? Like that would be worth setting up for a lot of people just as a schedule task. Like recurring skills, schedule task, right?

Play from 1:18:57 1:18:57 Like audit my skills. I I have something like this. Audit my skills. Look at staleness. Like let's let's be a skeptic, right? Like red red team. Dude, I've been using red team like a mug. it's great. ~ but you know what I mean? Like those things go, yeah, they they like. Everything's got an expiration date on it. So, I mean, you do have to revisit these things. I do have one question. I'm thinking like I'm thinking of this question in terms of something, ~ people that are in-house, okay?

1:19:13How to prove AI ROI to your boss

Play from 1:19:25 1:19:25 And and you are the top of an org, right? Somebody's playing with in with AI on your team. How do they what is the best way to present ROI or wins or the effectiveness of somebody that's working on something with AI to Prove value to you or that it's worth it. Right. Like how does some how should somebody think about reporting up in terms of what they've accomplished with AI and and maybe how that would impact the company, or just communicate in an effective way to make them look like, ~ I'm not just fucking playing around.

Play from 1:19:58 1:19:58 Like this actually does have some value. I can only speak to it from my perspective of my org. And I'm not I have no corporate background. I know nothing about traditional kneecap sucking or anything like that. Okay. that kneecap'sucking boy. Oof. Mm. Yeah. so I also look at like, okay, who on my team would be doing this the research or the stuff to present? And I did I'm like Rory knows I'm not super corporate.

Play from 1:20:24 1:20:24 Like I don't need a I don't need a ROI calculation. Like if something in your life sucks day to day and this and you built this thing to make it better, awesome. I'm happy to pay for the tokens. You know? So I don't I don't like have this rigorous ~ review process or whatever. Like you guys all have your allotted budget in Claude. Do whatever the hell you want. If you need over request, I don't even ask. Just do it. Like just do whatever you want. I'll say the biggest thing is like from a with design, because I'm like not super intimately involved on the day to day of what they're doing and the outputs of the design.

Play from 1:20:58 1:20:58 Is again just like what I do is just give them the freedom. Like you have access to stock tools. You have access to the the image gen stuff. I pay for it. You have your tokens. If you need more, ask me. If it ever gets to a point where it's overkill, I'll I'll raise the issue then about inefficiency or whatever. But it seldom ever gets to that point. And then what we do is every other week, I either meet with the design team or the copy team. So every week one of them. And people will just walk through what they're doing, how they've been using the tools, new new tool that they use, how they're creating components or little things they've the copywriters are doing.

Play from 1:21:30 1:21:30 Cause the c to Rory's point, like we have the whole system, but different copywriters will like prompt and do stuff different. And so if someone's blocked they you know they keep the going back to the log that I mentioned at the beginning, raise the issue and be like, and then someone else says, I'm doing it this way. But it's just like everyone's remote, everyone's distributed. So if we have one hour a ~ every other week for that group of the designers that are, you know, of of like mind to get together and just jam and have an open form, there's really no structure for the calls.

Play from 1:21:56 1:21:56 And we just get on it and just talk of that. that it's just like, what are you guys dealing with? What sucks? Like What how's the feedback been from the clients? Any patterns? And we just talk for, you know, sometimes thirty minutes, sometimes an hour. ~ but it's just a good environment just format for them. And I'm there 'cause I so I that's why I like to keep my thumb on the pulse. just like just talking about stuff. I think that's great. Like just the ability to show and tell and and yeah, get people thinking and and seeing somebody else and think about, I I might have solved something like that before.

Play from 1:22:26 1:22:26 Here's how I did it. Right. And you start to like, you know, again, sort of like become less fragmented because you're sharing the wins and losses and you're just being very open about how you guys are using those things. I I think like at a at a basic level, ideally companies are doing that. At a bare minimum, they're doing that, right? Because You're facilitating ~ and urging people to use the tools. You're providing the resources for them.

Play from 1:22:51 1:22:51 And then you're giving them a space to talk about it. Right. And I think like just that and building the reps in and then sharing the wins and the losses does a lot. It does a lot for a team, not just contextually, but from a confidence standpoint. to like giving people a giving people like the freedom to do it, then come out come back to me to make the business case. Like for me, I don't it at this stage it doesn't make sense. But like having someone what larger companies will do is have some like what they're calling now like forward deployed engineers where they're just like on the fringes of the of the dark web building stuff, right?

Play from 1:23:19 1:23:19 But effectively for us, like that's what I'm doing with the other gentleman that I have that helps me with the AI stuff is more technical in nature. So like when I hit a block or like on on the virtual machine or Some of the security stuff or whatever, like that's his expertise. So I can like lean on him for that stuff. but like that's my that's what I'm doing now. And so like I'm not too reliant on the team to do it at this point. But some other people, other orgs that don't have a maybe a guy like me at the helm.

Play from 1:23:45 1:23:45 It's like getting someone that's curious, but the the tough part to solve for is the subject matter expertise and knowing where the body's from. Yeah, exactly. With those that goes back subject matter expertise though. correct. But also like I've if I'm if I'm glazing you for a second, I always I always notice just like you can you can see a problem very clearly. It's not like hidden. You know, like you go to go to a lot of other companies and I'm just like, What like why why this?

Play from 1:24:12 1:24:12 And they're just like they they never thought about it. Like I feel like you have a really good eye for like this is something that can be fixed. And like it's a it's either like it annoys me or it annoys someone else, or it's been a you know, it's caused an issue in the past and you've like, okay, I'm gonna fix that. Right. Like I just I have to have to fix that because it's it annoys me. Right. So and that's that's a skill set like problem identification. We talk about that when we'd audit accounts. We'd be like they think it's they think it's the automations, but it's the you know it's the pop up window.

Play from 1:24:39 1:24:39 It's like they they have like point one percent, you know, acceptance rate in the pop up. They're like, my welcome flow sucks. It's terrible. It's like, well, no one's signing up for the offer and they're not coming downstream. It's like is it yeah, it does suck. Yeah. So it's but dude, it's like the you just mentioned this. I don't think we've mentioned this really like one time other than in a macro standpoint. Like you just said you just said security. Like what it what what is that what are you doing with like is that client data?

1:24:50How do you secure shared AI infrastructure?

Play from 1:25:06 1:25:06 Is that like client assets or like internal stuff? Like what are you what are you doing security wise? yeah, but more so just like the repo, the super base, the data store, right? Like the team not being able to write to it and break anything or or put save anything there that isn't worthy of being saved. So no no one can save anything anywhere besides me and one other guy. That's it. So anyone like people can People don't even have direct access to the repo.

Play from 1:25:34 1:25:34 They have access to the virtual machine because no one's on the front lines actively building stuff. So they don't need to be able to create write create a PR and merge anything. You know what I'm saying? And so whereas before old cloud code, the team had access to that stuff, but like they don't need to it's confusing. And they don't need they don't it they they don't need to know, not from like a secretive perspective, just because that's not what they need to be concerned with. Right. And so now what they can do is they can W with our MCP and the way things are structured, if they're doing on the same thing and they want to memorialize it, they can what's it called?

Play from 1:26:10 1:26:10 there's a skill in there or or whatever that they can invoke to like summarize what they're doing, make sure it's good, and then send it to me or the other gentleman that can write to the report. Okay. That's cool. Did it take you I don't want them worried about making changes, thinking about is this worth saving? Isn't this worth saving? Because more often than not, nothing is worth it's not worth saving. So I'm just like keep just executing the tasks that are assigned to you in Asana with the with the pimp ass automations that I built and just and keep running it up.

Play from 1:26:37 1:26:37 And when you if when you do want to memorialize something, tell me and then we'll figure it out. But it's happened so infrequently that giving them the ability to do it would create more problems and then it solves. Sounds you build like franchise as an internal structure. It's like just like here's all the op here's like the entire thing, here's the whole playbook. I've actually built it for you. Just like open up shop and go. You know? Yeah. be efficient and and just serve more clients with less headcount. And I don't want the biggest org. Like right now, the I I would rather brag about have like back in the day, people bragging about having the most people.

Play from 1:27:08 1:27:08 I want to brag about having the least people. Granted, for agency business is super inefficient from a human capital perspective, because you need people to fulfill, right? So it'll never be like, you know, these billion dollar one person AI companies, but that's all right. But from an agency perspective, I'm like, all right, if we can have If I can finesse two X the two X the client load with the same exact team, I mean, ~ that's fucking crazy. And so rather than having the rather than having double the headcount to service the same amount of clients.

Play from 1:27:38 1:27:38 And you're asking your employees to do their specialty, which I think is really important. That's it. we can finesse this from an op standpoint, then it's like I can pay everyone well above market. They'll have the best quality of life, day-to-day ~ experience working because we solve for all the stuff that sucks and they can just focus on their thing. And then what this just a it's just a nice, nice way to live. Yeah. That's that's what I mean. Like an an AM can be an AM.

Play from 1:28:03 1:28:03 A copywriter can be a copywriter. Designer can be a designer. Yeah. and a deployment person. Yes. And again, just it's it's a lot of insight to get to that point. It's a lot of experience failure, ~ frustration, you know, maybe a couple of assholes along the way that that that make it, you know, that get you to that point of being able to structure it even. But you know, it's for anyone who's thinking about this, I mean you can take the same take the same thought process and apply it to a team structure internally.

Play from 1:28:32 1:28:32 It doesn't have to be your own out agency. Like whatever what you've just sort of like laid out is an infrastructure for any team of like how, you know, you just need a point person. You need a point person or or point two people, you know, whatever it might be. But dude, it's it's super cool. And like I I I've always been ~ you know, now that we now that we don't sort of operate together anymore, we talk all the time and I'm just like, what are you doing?

Play from 1:28:58 1:28:58 And then easy we we end up we end up like forty five minutes to an hour down the line. Yeah. on creative and I'm so indexed on the other ops stuff. You know, so like when I same thing when I ask I'm like, that creative shit, that's super cool. 'Cause I like I it's it I appreciate it, but I'm so far from it. And then opposite for you, you're like, dude, you're doing all this crazy op stuff that you're so far from, but it's an entirely different use case for AI than is your day to day. Yeah. And it's that's the that's the one thing because right we've talked about this and you know, you know, probably just glaze over this slightly.

Play from 1:29:28 1:29:28 But the creative side of email is it's it's hard. It's not very easy. Like if anyone's not familiar with an email marketing sort of campaign and what it looks like, there is multiple blocks, there's multiple points of messaging, there's multiple images, there's multiple links, like there's a lot of stuff. It's not just like, you know, s go into ChatGPT, give me an email marketing campaign and then like drop it in and like, you know, attach a link. You can do that. But if you're doing it for the highest level of like really decent brands that, you know, email is a very important touch point to people, like there's a lot of creative that goes into it.

Play from 1:29:58 1:29:58 It's not simple. And I think that's one of the things that ~ you know, we were talking about is how to automate that, you know, sort of process, not to write and design everything, but to get blocks, to get, you know, the infrastructure in to let people do things. Again, like you know with data. O B E L L O. Yeah. So they got, ~ I know the founders of there, they do some really, really cool stuff.

1:30:11What can Obello automate for brands?

Play from 1:30:23 1:30:23 They're they're they're pilled on the the AI sort of creative component. But they were showing me something that they have too, where it's like you you can basically customize. Here, let me see if I can kind of pull this up really quick. And ~ but this is really interesting too, because I I had a client was was basically a a portfolio holding company and they had like 19 or 20 brands across it, right? So it allows you to bring, you know, the brand guidelines in.

Play from 1:30:49 1:30:49 You can scale all these things with aspect ratios, ~ email builder, but the email builder, it's like a it's like a design tool the way it's built. It's like a Canva meet Figma kind of thing. And all the changes you can like set up governance with with you know, like who has permission to what, but then also like you can apply all these changes globally or through a certain campaign or whatever it may be, and allows brands, I think, to speed up like the email like design process.

Play from 1:31:19 1:31:19 Also, like when you're thinking about all these display, remember we were talking about the display ads, Rory? Like this was sort of the tool I was pushing my my one client toward because I'm like, this is. Insane because it allows you to basically take one piece, you can lock in components, and then it'll auto-adapt to whatever display, banner size, aspect ratio you're looking for, even the weird ones. ~ this is just like a very, very cool tool that like I I want to use.

Play from 1:31:46 1:31:46 I want to use, but I have no yeah, Can I speak to this for a minute? yeah, go for it. All right. So there's a bunch of people that have taken swings with a tool like this before. Right. Some of like it's it looks fucking dope, right? But then it comes back to like, okay, doing this for clients and their creative preferences and like how picky they can be about certain things. Like there's just things that looking at the template or just I'm just looking at your side and then I pulled it up on my on my end just to look at it, that like from a structural component standpoint, like can't can't do.

Play from 1:32:16 1:32:16 For example. Usually like in ecom world, the headline, subheadline, call to action is on the first block that is above the fold, right? And every every re every reference image that I saw on their site that is an email doesn't have that. And it might be a creative limitation with respect to the tool. Maybe it's not. But all these tools are just like little things or preferences or just like little industry things that don't get factored in.

Play from 1:32:41 1:32:41 So like some most some brands it's sufficient. I'll say for mo for the smaller end of the spectrum, 100%. But then once you start getting into with these fucking brand directors and creative directors and brand guidelines and all this stuff, that it's like, ~ it's like you need the flexibility of Figma to be able to execute what they need. ~ otherwise, I'm gonna check it out. one because the I think the guys that made this were creative directors.

Play from 1:33:07 1:33:07 I mean, they they have sort of that background. They're thinking about this maybe a little bit differently than some of the maybe the other players potentially. But I I saw this and it was like this, it was somebody that like understood the things that we understood with a lot of this stuff. And I was like, dude, that I wish I had that. when I was in house. Like I was just like, man, this would have just taken me fucking forever. And like the email builder is just like one part of their thing.

Play from 1:33:33 1:33:33 Like they have a whole like dam system set up in there. You know, brand guide like all the brand guideline build outs, the governance layers. It's it's it's pretty dope. Like they just I think they're like a newer company they just raised, but it and then if I use a seven year affiliate link, all right, bro? me know what you think. no, let me know what you think for sure. That's that's ~ you know, we we could use more affiliate links on this show for anyone who's listening.

Play from 1:33:58 1:33:58 You know, anyone who wants to, you know, provide some some access to some tools too. If you're interested, we can always test them live. ~ a little a little subtle plug here. Look at us. But look at see, that that's that's the interesting part. The Figma stuff is is really ~ you know, I think that was pretty that was also pretty revolutionary for the agency stuff at the time. Just like having the clients being able to be in there and see the stuff and not just like, you know, being able to comment directly in it and not having to like at least export shit and send it back and forth.

Play from 1:34:23 1:34:23 It's like so many little things, but dude, it's it's awesome, I think, like what you've built. And it's like it's cool to see sort of the side by side as to like what it was like then versus what it's like now. And I mean what what's the client total up as we start to wrap things up here? I think what's the what what's what's your client total right now? Forty in like it's been like fifteen months, sixteen months. And what's the what's the head count?

1:34:42How many clients can the team handle?

Play from 1:34:49 1:34:49 If I think of core team right now, we're at sixteen. Sixteen, okay. So what? Mm-hmm. So this is like again, and that's not a small team, that's not a you know, single person operation. Yeah, you got and then you've organized it into No, right? to forty, and you said you had room too. So I mean, that's th I I think look, I think I don't right now I'm like, I think we could double.

Play from 1:35:19 1:35:19 Same roster with the bandwidth availabilities, with the efficiency pickups of the A of all the stu all the ops stuff that I showed you. Like none of the stuff we went over is creative or copy. Everything is ops. Yeah. Everything stops. And that's and that's where what how much of your time is spent on that at this point now? all of it bes besides sales, all of it. Okay. So I mean and then I'm sure you put you have sales ops stuff that you've you've plugged into it as well.

1:35:42How does AI automate client onboarding?

Play from 1:35:45 1:35:45 Yeah. What's different? it, but the onboarding piece, once a client says, All right, let's go, ~ in this in the CRM, once they hit a a ~ a stage, that it will s ~ everything is stored in there with respect to the service, the price, the team that they're on or will be on, ~ so on and so forth, sends them contract. I don't even have to I don't look at it. Everything because everything comes from the CRM.

Play from 1:36:11 1:36:11 ~ and then as soon as they sign, send Stripe, Google Drive, folder share drive, Asana project spun up with all of the rules and everything that we talked about that are all baked in there from what that I built 16 months ago. And ~ internal Slack channel, external Slack channel with the invitations that they give us from their onboarding form of who they want on their team in the Slack. Like that's day zero, as soon as the contract is signed.

Play from 1:36:38 1:36:38 Did as a client you gotta feel pretty damn good about that. When it's just like all this stuff just happens right away, it's like all handled and taken care of. Like you I wish we had that back then. That was sick. And then next part in the project in and of itself that's assigned to all the stakeholders, yada yada. In the sales process, we have what we're going to be doing for the next nine hundred days pretty much from the sales call. So that gets memorialized, put in the asana task. So the account lead knows handoff from me, call transcript or transcripts, the game plan.

Play from 1:37:07 1:37:07 So they can they can get ready for the kickoff call, have a hundred percent visibility, and then also simultaneously s the scraping of their store emails from that they've sent it before, the tone guideline, brand guideline, when they send it all saved to the data store as soon as as soon as as soon as that contract is signed. What? Why like like this would it's easy but it it's easy. but still from my three years ago, Yeah.

Play from 1:37:35 1:37:35 like the amount of work that went into just scraping the store. You know what I mean? Like just it's things like that that it makes me, you know, even on the the onboarding side that just send preparing the s the contracts, you know? Just throwing it out there. All right. dude, I don't know if you can I don't know if you can afford me anymore. I'm a little bit more than a th little more than a thousand dollars a month, Phil. Well, you know, we'll figure it out, dude.

Play from 1:38:00 1:38:00 We'll we'll we'll we can account for inflation, etcetera. We can get the gang back together and but I'll let you know when the when the J D is live, okay? I don't I don't feel like getting fired again. Can we can we, you know, put that on my contract? Can't fire me. I didn't fire you the second time. Damn, got him. Semantics. Semantics. of like those kind of like a working the fire you I quit situation. ha ha. Wait, no, come back.

Play from 1:38:26 1:38:26 No, absolutely not. That's the All right, I the AI stuff with you back. No, okay. I I I have ~ I have just a I I have a random question. So as we wrap up. ~ Phil, give me give me an unpopular or or just a hot take about AI in your world when it comes to agencies, creative ops, whatever whatever it may be. What what's something that like you're thinking about a lot that maybe is something that people haven't heard of before or is a contrarian sort of take or whatever it may be?

1:38:33What's Phil's biggest AI operations lesson?

Play from 1:39:01 1:39:01 I think I already I think I blew my load on that ten minutes ago when I was like, Not everything that should be can be done. Pause. Mm. Not everything that not everything that ~ can be done should be done. Just be got another one. Come on, dig deep, man. the other one I'll say is just like do less. Do us. Do less. Keep it simple. First principles. First principles.

Play from 1:39:27 1:39:27 It's simple stuff, dude. But is, dude. Going back to the mouth breathing nature that is me, it's just keep it simple, dude. I'm not that bright. strips for that, right? They do. But I I don't more so mean it like and I really am a mouth breather. More s I don't want to drop the R word. I don't wanna drop the R word here. So I'm I I I ~ you know it's spectrum oriented, yada yada.

Play from 1:39:52 1:39:52 You know what I mean? Well you did you did make your you did make your you know internet debut of you know selling t shirts on obscure message boards so I mean that it's a it's a great story. But we won't I forgot about it till right now. So he my origin story. did he's used to used to do this on sales calls and like I'd hear it and I'd just like I'd try to like bite my lip because he just said it with so much enthusiasm.

Play from 1:40:19 1:40:19 You know, you sell t shirts on obscure message boards. I'm like, what the hell is an obscure message board? point ~ That's how long it since you ~ that but then he hired you off ~ Craigslist though. So makes sense. I mean the the the breadcrumbs were there. Yeah. Sh shadow flows. Shadow po dude, they're real. They're real. But yeah, I don't know, man. Just like don't worry, but it's just this is this d this is this is like I I I I used to listen to Shadow on sales calls and like dude can sell things.

Play from 1:40:52 1:40:52 He's very good, very straightforward, to the point, no bullshit. But he would use terms every once in a while like you know, purchase horizon. I'm like, you just come up with that? I don't even know what the hell that means. You know, I'm sitting there, like I've heard the same conversation 40 times. I've never heard the word purchase horizon come up. And like the other people look at him on the phone like, purchase horizon? Yeah. I I love that. You know. of those in every once in a while goes a long way. Dude, I don't even know how to read, you know?

Play from 1:41:18 1:41:18 Yeah, we we know. We we know now. But I think we dev less, dude. Keep it simple. Like it. Rory, who we shouting out, man? Who should we shout out this week? What are we doing? local local account manager. Maybe you're ~ you know, Drew ~ Phil, if you're not familiar at the end of every episode, we tell everyone that they gotta do something if they sat here for at least an hour and forty minutes. If it's if it's not like or subscribe, it's you gotta tell somebody.

1:41:24Who should viewers tell about Fast Hours?

Play from 1:41:44 1:41:44 But we try to find a you know, maybe you're obscure message board operator. Let's tell them. That's a that's a good one. ~ yeah, you your subreddit. founder on Reddit, ~ whatever that whatever that would be. you know, you better tell somebody. I like that. Bring it back to how the call started. Outcast, goody mob, you better tell somebody. better tell somebody. It's a good one. I think that's how we ~ appreciate you guys joining us.

Play from 1:42:12 1:42:12 Hope this is a great episode. having me, Rory. I really appreciate being allowed on the show. I feel ~ I feel vindicated and I f you know, I feel like a million dollars cash, honestly. So thanks, man. Thanks, Drew, for having me as well. when you said vindicated, we needed some dashboard confessional right there. But ~ maybe maybe we can do that in post. Anyway, we'll see you guys next week. Thanks for joining us. See ya.