Latest / eCommerce Podcast / He Fed 500,000 Emails Into an AI Copy of Himself

Transcript

He Fed 500,000 Emails Into an AI Copy of Himself · eCommerce Podcast

• 0:02
  Welcome to the eCommerce Podcast. My name is Matt Edmundson, and it is great to
• 0:06
  be with you today. I'm very excited for today's episode. We have a returning
• 0:10
  guest to the show. I'll let him tell his story in just a second, but we have a
• 0:14
  great show coming up. I have no doubt. I've been really looking forward to this.
• 0:18
  Always good to catch up with old friends, and that's what we're going to do
• 0:22
  today. If you're new to the show, welcome. We talk about all things. Ecommerce,
• 0:26
  how to build, grow, and do all that sort of good stuff online with your
• 0:30
  business. It's what I do. I run my own ecom businesses, have done since 2002,
• 0:35
  which is a really long, a really long old time. Hence the reason I call myself
• 0:39
  the ecommerce dinosaur. Um, but yeah, welcome to the show. Uh, make sure you
• 0:43
  like and subscribe and do all of that good stuff. Um, but yeah, welcome to this.
• 0:47
  Let's just jump straight into it, buddy. How are you doing? All the way in
• 0:51
  Kansas City.
• 0:55
  I'm doing well. Um, I— it's been too long since we chatted last. And, um, you
• 0:59
  know, I think, uh, I think you might be an ecommerce dinosaur. Um, I'm usually
• 1:04
  not a fan of people like, you know, kind of making fun of their age or their
• 1:09
  experience because I think that it's something to aspire to. But, uh, I feel
• 1:13
  like I'm a dinosaur and you've got me beat by a decade. So You know, I actually
• 1:18
  do, you know, I look up to you as an operator in a lot of ways, you know,
• 1:22
  through our conversations, I've learned just quite a bit and always look forward
• 1:27
  to a chance to getting to connect and just hear what you're working on and what
• 1:31
  the lay of the land is.
• 1:36
  Yeah, nice. I mean, well, for those that don't know, maybe they didn't catch the
• 1:40
  first show, but just tell it, just give us a brief elevator pitch.
• 1:43
  Yeah, so I've been in the Amazon ecommerce space for about 15 years. Um, run a
• 1:48
  full-service agency called Marknology. What's a full-service marketplace agency?
• 1:53
  Um, you know, especially when Amazon was becoming a thing, I jumped in, uh,
• 1:58
  headfirst and, and haven't looked back. So primarily Amazon, um, helping brands
• 2:03
  navigate Amazon images, SEO, running advertising, their reporting analytics, um,
• 2:07
  resellers policing the platform. And it just expanded from there to TikTok Shop,
• 2:12
  Walmart, Um, and in other areas like that. So I think we really consider
• 2:17
  ourselves marketplace-centric instead of just Amazon in that regard. And, um,
• 2:22
  along the way I've launched a warehouse and 3PL. I guess I'm 6 years into that.
• 2:27
  I'm just a small— I, I wanted to build my own brands or acquire my own brands,
• 2:32
  and in that roadmap saw that, you know, I think fulfilment warehousing, what
• 2:37
  happens there is a big profit center, or quality, um, of your experience, you
• 2:41
  can really control it. And so I launched that for better or worse.
• 2:46
  Yeah.
• 2:46
  And, um, and also have a, a portfolio of brands that are, that are mine, um,
• 2:51
  mine and, and other investors, a couple investors with me on those. So a brand
• 2:56
  owner, warehouse and 3PL, and, um, you know, in the ecommerce space. And, and I
• 3:01
  think the last episode I caught of yours was AI, and I know everyone's talking
• 3:05
  AI. It's, it's, um, it's like the general word that, you know, goes so much
• 3:10
  deeper, kind of like when someone doesn't know your music genre, And they just
• 3:15
  say rock and roll or something. And you're like, listen, there's a lot of
• 3:19
  subcategories, okay? There's a lot of niche there. Um, but you know, I've been
• 3:24
  pushing super, super hard there as I try— as I leaned up my agency. Um, '24,
• 3:29
  '25, we, we dropped about, about 7 or 8 people. We're about a 32-person agency,
• 3:34
  just for context. But for us, it was like, it was a bit of a shift. And, and
• 3:38
  when we did that, um, It wasn't because of AI, it was because of just the
• 3:43
  industry and the economy and things were where we were, where I was. But in
• 3:48
  doing so, also kind of put my back against the wall with work and hours in a
• 3:53
  day. And I started digging into AI. I had already been working on something,
• 3:57
  started digging into AI, and the last year has been deep, deep into, deep in the
• 4:02
  deep. So yeah, I'm well-rounded, more of like, Was it a master of many? Um,
• 4:07
  yeah, doing, yeah, doing a little bit of everything. Yeah, yeah, yeah.
• 4:12
  No, I, I love that. I think we're quite similar in many ways. And you have a
• 4:15
  great little office in Kansas, don't you? Kansas City. Uh, there's a great
• 4:18
  barbecue place down the road from you guys. I remember that.
• 4:22
  Yeah, yeah, yeah, yeah.
• 4:24
  And, uh, it was great getting to hang with you and the fam, and, uh, you got to
• 4:27
  meet my daughter, and we just had a good— I mean, it was great.
• 4:31
  And you're wearing the Goonies shirt, which I think you had just had a chance to
• 4:34
  like go in the house or something before.
• 4:38
  That's right, it was on that trip. Yeah, yeah, on the way back from the Goonies
• 4:42
  house, we, um, we stopped off in Astoria and we, we did the whole Goonies thing.
• 4:46
  And, um, yeah, and then we came to see you, didn't we? We flew from there to,
• 4:50
  to, uh, to Kansas, which was great. So yeah, I'm still obviously, you know, this
• 4:53
  was a gift. Uh, I don't know who got me this actually.
• 4:57
  I don't know if you coordinated it or not, but for me I was like Wow, he
• 5:02
  remembers the last time we connected, you know? So no, it's great. And, um, you
• 5:06
  know, I know, um, you've got your hands in a few things, um, not sure everything
• 5:10
  you're touching, but, um, I think that whether you're selling in Europe, you're
• 5:15
  selling in the US, wherever you are, a lot of the same principles are applied.
• 5:19
  Yeah, absolutely. So you've been digging deep into AI. I'm curious, bud, what
• 5:22
  have you discovered? What's been like some of the standout things where you just
• 5:25
  go, I've done this and it's been brilliant, or I've done this and it was a waste
• 5:28
  of time.
• 5:31
  Okay, so for context, anyone listening, like when you say you've been digging
• 5:36
  in, you know, people talk about what subscriptions they're on, whatever, you
• 5:40
  know, I've probably spent $50,000, you know, in learning. So painfully so, but
• 5:45
  also like I'm breaking it, I'm pushing it to break it, I'm pushing it to break
• 5:50
  it, I'm pushing, pushing, pushing. You know, I've been speaking a little bit on,
• 5:54
  as a speaker, you know, as a person in the space, I generally try to choose a
• 5:59
  subject that everyone else isn't talking about and be like, this is what the
• 6:03
  people need to hear, you know, like Moses or something, right? But with AI, you
• 6:08
  know, that's where I'm living. So it's right where everybody's kind of hot and
• 6:13
  what they're talking about. And, you know, for the longest time I was trying to
• 6:17
  build software kind of behind the scenes as a service agency guy. I went to
• 6:22
  school for computer science. Um, networking, security, programming languages.
• 6:27
  That's my degree. And, um, found ecommerce and kind of put that on the shelf. I
• 6:31
  didn't go the software route naturally. All that to say, um, I've been, I've
• 6:36
  been building software and building systems internally. It started with like,
• 6:41
  what can I do to help myself? So, um, I started OpenCLAW, uh, like when it first
• 6:45
  came out, I think in February.
• 6:50
  Yeah, yeah, it went nuts, didn't it? Everybody wanted OpenClaw on a Mac Mini.
• 6:54
  Yeah, exactly. Um, and people were scared, you know, they were nervous. But I
• 6:59
  think there's very few— I mean, I'm in a unique situation by being a CEO, if you
• 7:03
  want to put it that way, business owner that also can do the coding and the
• 7:08
  programming. And, um, where I'm not having to get permission from this person or
• 7:12
  this person, right? I'm just Going. And, um, because of that, I think I've
• 7:17
  gotten a lot further than other people would, even, even at big companies,
• 7:21
  because they're going to have these, these kind of these roadblocks, you know.
• 7:26
  Yeah.
• 7:27
  So, um, all that to say, I'll try to summarize kind of like what, what I, what
• 7:31
  I've done, you know. So to anyone that doesn't know, OpenCLAW is like an agent
• 7:36
  that, um, has a heartbeat, a soul, these files that like keep it always awake,
• 7:41
  that keep it always kind of listening, always there. You interact with it
• 7:45
  through Telegram or Discord or a channel like that. And the more things you
• 7:50
  connect it to, the more things that it can do for you. So it could be connected
• 7:55
  to your email, to your, you know, your project management, your Slack. So, you
• 7:59
  know, to that end, I started like ingesting the 15 years of work that I've done.
• 8:04
  So podcasts, blogs, articles, SOPs, playbooks, Proposals, won, lost, you know,
• 8:08
  as far back as I could go, I could essentially turn into value. So, you know,
• 8:13
  I'm talking 500,000 emails just under mine, not, not the rest of the team or
• 8:18
  anyone else, just, you know, 500,000 emails over the course of 15 years. And
• 8:22
  then from there it was like, well, I want to get the 300 episodes I have from
• 8:27
  the podcast. And, you know, the podcasting was cool, but the real meat's in the
• 8:32
  emails. So then, you know, I'm segmenting them out. I am, you know, all these
• 8:36
  things and I have all of this data. And the problem I ran into was context. I
• 8:41
  think everyone's running into context, right? It's as a, as a guy, I can know,
• 8:45
  okay, Matt has a daughter. He likes The Goonies. He like, you know, runs a
• 8:50
  couple of brands. He's had a successful exit. He, you know, he runs a podcast.
• 8:55
  He runs a community group. He's, um, he's a Christian. Uh, all these like things
• 8:59
  that I have in my context just from pulling it like this. Right. And with the
• 9:04
  best friend, you're going to know more or less, but in general, like the AIs are
• 9:09
  going sequentially, like, okay, I'm going to go look over here. I'm going to go
• 9:13
  look at Andrew's email. I'm going to go look at Andrew's podcast. I'm going to
• 9:18
  go look in this software data. I'm going to go look over here. And it goes kind
• 9:22
  of sequential like that, but it's not like it knows everything all at one time.
• 9:27
  And I was like, I've done all this work to kind of build these, these brains, so
• 9:32
  to speak. And, and my AI, let's say Claude or whatever on top of that, is just
• 9:36
  missing stuff all the time. And I'm spending all this time getting him to do
• 9:41
  other things. And at first it was like, okay, let me be able to build a proposal
• 9:46
  without the design team. Let me be able to set up an email without, without the
• 9:50
  marketing team. And it could do these one-off things, but like in regards to
• 9:55
  like actually thinking like an operator, thinking like Drew, uh, in regards to
• 9:59
  an e-commerce problem or whatever, it would have some of the story but not all
• 10:04
  of it, and it would have holes. And so went from kind of like this, like,
• 10:09
  winging it system to giving it a set of rules. Like, I gave it basically 350
• 10:13
  rules called Drew IQ, Q, um, where before it even gets to the AI, it's running
• 10:18
  through a series of like, if this, then this. that I already know as an
• 10:23
  operator.
• 10:27
  Yeah.
• 10:28
  Right? So if I'm evaluating advertising, PPC, and I'm like, okay, Claude, uh,
• 10:32
  you know, the general, the general approach is like, should I— what should I do
• 10:37
  with this campaign? Should I raise bids, lower bids, pause bids, add new
• 10:41
  keywords, whatever? It's just thinking about the PPC. Well, I wanted to think
• 10:45
  about inventory levels. I wanted to think about margin. I wanted to think about
• 10:49
  where the competition is. I want to think it— to think about on Amazon search
• 10:54
  query performance report gaps. I wanted to think about organic ranking. I wanted
• 10:58
  to think about, am I launching or, um, am I just, uh, trying to rank for
• 11:02
  keywords or am I trying to clear stock? All this different context. So in order
• 11:06
  to get it like that, I set up these rules and I set up tags. Okay. So the tags
• 11:11
  would get everything a little bit of context. Like this is a launch SKU. This is
• 11:15
  this, this is this. And so before there was a software, there's just the backend
• 11:19
  of OpenCLAFS. So I've got email and I could talk to it and it's doing certain
• 11:23
  things, but I just turned into a glorified, um, delegator, if that makes sense.
• 11:28
  Like I was given superpowers, but I was just becoming the delegator myself. It
• 11:32
  didn't really have the intelligence to route things or really know consistently
• 11:36
  what kind of output I wanted across the board. So I know it's taking a little
• 11:41
  bit, but I, I get to like kind of like where we've ended up. Um, but I wanted
• 11:45
  all that context in one spot and I wanted like, hey, I have all these things in
• 11:49
  my head from our SOPs, our playbooks, our rules that are like, if I'm really
• 11:53
  looking at advertising, I'm like, I need to take— first I'm gonna go look at it.
• 11:58
  I'm gonna get all these tabs open, so to speak, in my, like, you know, normally
• 12:02
  I'd get all these tabs open. I'd be like, I'm gonna look at the margin real
• 12:06
  quick. I'm gonna look at the inventory real quick. I'm gonna look at our ranking
• 12:10
  on this app, Helium 10. I'm gonna pull this data, you know, pull up 5, 6, 7, 8
• 12:15
  different sources. And then I start to get to work on the advertising. Right? So
• 12:19
  that would be what happens normally. Well, if I know I'm going to check all
• 12:23
  those things, I could theoretically write rules that would make it check all
• 12:28
  those things before it came back with an answer. Well, okay, can do that, but it
• 12:32
  was expensive and it took a lot of time. It would sit there and think and think
• 12:36
  and think too long. So I built something called, uh, it's abbreviated WCS, and
• 12:40
  what it does, it's on my machine, it surfaces all of the context in like half of
• 12:45
  a second for for everything around that brand. And then it runs it through. So
• 12:49
  it's run through the rules already and it's surfaced everything all the way to
• 12:53
  the top, meeting ingest everything. So by the time the AI hits it with your
• 12:57
  beautiful prompt that you've written or whatever, Yeah, yeah, yeah. Pre-done, it
• 13:02
  has all of this context there and it's able to like actually start helping you.
• 13:06
  Okay.
• 13:06
  And so otherwise it was sequentially going down this order. I needed to have
• 13:11
  everything there. That has been like, Painful, but also like, wow, when I, when
• 13:15
  I've got it there. Um, and you know, that's the main issue with, with software
• 13:20
  that's out there or anything really trying to help you, even getting an
• 13:24
  assistant for myself or really getting someone to help me, is they're like
• 13:28
  missing all these context pieces that like would take a lifetime to give
• 13:33
  someone, even, even an amazing person, unless they're just involved in
• 13:37
  everything. So. Um, there's a software front end to that that's like internal
• 13:41
  for our, our company that's like, you know, been amazing. My goal is just to
• 13:46
  reduce software costs and increase efficiency and give the team a lot of things
• 13:50
  that if I was helping them with something, this is how I would help them and
• 13:54
  where I would direct them and what I would tell them. And so it's been, um,
• 13:59
  getting away more so from vibe coding and things like that and more into like
• 14:03
  structured coding and then vibe on the end or like when you're like
• 14:08
  experimenting, but you need these, like the structure in place to kind of get
• 14:12
  you there. So, you know, you can add everything else to it, like cutting up
• 14:16
  podcast episodes and the marketing elements, and that kind of came first. And
• 14:21
  now it's this data layer and interpreting it that I'm kind of going into next.
• 14:25
  Um, but it's been amazing. I think, um, I think, uh, I'm excited finally. Like
• 14:29
  for me, software has always been a pain point and it's not even to me about
• 14:34
  monetizing this and then going and selling it. It's really about adding value to
• 14:38
  my agency, you know, to my customers, to my team, and getting something solved
• 14:42
  that we haven't been able to solve for a long time that's just never really
• 14:47
  existed. It's always been kind of this piecemeal of, of like, you know, 10
• 14:51
  softwares or something if you're really there. And yeah, it's been really
• 14:56
  refreshing, I think, just being able to control some areas that were frustrating
• 15:00
  for me. And an assistant is always I've always struggled with assistants. I
• 15:04
  really needed a copy of me. Like, that's at the end of the day, that's kind of,
• 15:09
  you know, that's really what I needed in order for it to be helpful outside of
• 15:13
  like, you know, basic things. So I've even, you know, wrote an app and got
• 15:17
  approved by QuickBooks. So, you know, my emails coming in automatically draft
• 15:22
  invoices and do a bunch of stuff like that. They don't send, it's always human
• 15:26
  gated, but, you know, um, I've created the assistant and now I'm kind of onto
• 15:31
  like, okay, how do we really like take Everything we've learned across 10, 15,
• 15:35
  20 years. And, and what I think what's exciting is like in the past that work
• 15:39
  was already done. It's like expense, it's energy expensed and bringing like
• 15:44
  value out of that work from the past, old, old podcast episodes, emails that
• 15:48
  you've responded to and being able to see my business like, wow, here's 500 use
• 15:52
  cases of, of objection handling that I, that I did or that I tried and failed,
• 15:57
  or like, you know, having those types of metrics as an agency owner is really
• 16:01
  cool. So I'll set the mic down because that's a lot. But, um, yeah, it's been a,
• 16:05
  it's been a fun journey, I think, uh, and getting to be creative again instead
• 16:10
  of just managerial in my company.
• 16:14
  So isn't it? I think that's similar to you, but in the sense that I am the CEO
• 16:19
  or MD, whichever language you want to use, of the company. And I, I started off
• 16:23
  programming to, you know, 20, 30 years ago, I did all the initial sites. That's
• 16:28
  how the whole thing was born. And then I just hired really good people and I
• 16:32
  just never really kept up the skill.
• 16:37
  Touched it.
• 16:37
  Yeah. And you're just like, well, you know, we've got good people. I don't need
• 16:42
  to do that. But then with the advent of AI, I've sort of picked it all up again
• 16:47
  and rediscovered, I guess, a whole bunch of that stuff. And I just love the
• 16:52
  creativity of it. I think What it's given us the ability to do is to look at,
• 16:57
  you see a problem in your business or you identify something that is, if I could
• 17:02
  get another me to do that, that would be amazing. And within a few hours, you're
• 17:06
  like 80% there. It's the most extraordinary thing. I've never seen anything like
• 17:11
  it. And so like you, we're canceling subscriptions left, right, and center at
• 17:16
  the moment because We don't need them. You know, we, we don't need project
• 17:21
  management software anymore. We've written our own. We don't need, um, this type
• 17:26
  of analytics over here because we've written our own and it's actually much
• 17:31
  better for us because it's bespoke to what we need. And so, yeah, I'm with you,
• 17:35
  dude. I, I, I love the creativity of it. I love the fact that I get to play and
• 17:40
  just have a bit of fun, you know, and just, I, I mentioned this before on the
• 17:45
  podcast. I don't know if you saw it, but I, Um, I was, I was a child that grew
• 17:50
  up in the 1980s, and so I developed KITT, um, on my phone as a sort of a face
• 17:55
  for Claude. And so when I'm driving, um, I have this big microphone button on my
• 18:00
  phone and I can press that button and I can talk to the phone and it will answer
• 18:05
  me. Um, and then it will sort of, it will come back, uh, with its answer, with
• 18:09
  its sort of mock KITT voice. Uh, and the little dials go up and down like they
• 18:14
  should. And now I'm like, I'm saying to AI, I'm like, this is awesome. I mean,
• 18:19
  it's, it's, it's practical in some respects because I can do stuff while I'm
• 18:24
  driving, but it's totally off the charts.
• 18:29
  It's added some fun. It's added some fun to something that as a business owner
• 18:34
  or operator you have to do day in and day out. And, and, um, you know, just
• 18:38
  instead of typing, right, we're talking easily. Uh, and I know some people have
• 18:43
  been doing that for a while, but when you really start coding a lot, it gets—
• 18:48
  and I think one thing that's been really refreshing for my brain, like, is I
• 18:52
  think, uh, I'm a strange person, you know? Like, I, I'm a Pentecostal kid raised
• 18:57
  in Africa, in Congo, um, school in Hawaii, you know, my story, like, just here,
• 19:02
  there, and everywhere. Um, and I'm, I'm a little different. Uh, I think even
• 19:07
  with just my creativity, like, the— my imagination, I literally grew up with You
• 19:11
  know, without a TV in my home. I grew up in— I was born in '86. Um, you know, I
• 19:16
  didn't have a TV. I played outside. My— the imagination, sky's the limit. And I
• 19:21
  didn't have a lot of kids, uh, telling me something wasn't right or whatever.
• 19:25
  There was just no one to tell me that. So, you know, um, I think in some ways
• 19:30
  I've always just struggled communicating what's in my head or my ideas. Um, and
• 19:35
  you spend so much time if you're a leader, like, um, changing what's your words
• 19:40
  to make it fit whoever you're talking to, right? Trying to say that correctly.
• 19:44
  Like, and, um, like, if you get it tuned right, just the— it's almost like a
• 19:49
  therapist in that way, except it's around work things. And, um, you're able to
• 19:54
  say whatever. I can say it in my, my words, I guess, how I want to say it
• 19:59
  without being super descriptive or whatever, right? And and all the, all the
• 20:03
  isms. But, um, that's been kind of refreshing to just be able to be creative
• 20:08
  without having to get someone else to buy into my idea right from the beginning
• 20:13
  when I'm coming up with the idea is like in itself liberating, I think.
• 20:17
  Yeah, no, it is. I love it. I love what you're doing. It's great. Where, where
• 20:20
  do you think it's going to over the next sort of 6 to 12 months? What are you
• 20:23
  seeing?
• 20:26
  Well, I think the trend is going to be Look, if you're mediocre, you're in
• 20:30
  trouble in whatever job you're doing, pretty much, unless you're like hands-on
• 20:34
  outside working on the house. Like trades is gonna explode, you know, I think,
• 20:38
  uh, just become more valuable than it was. And, you know, I think, uh, a couple
• 20:42
  years ago we'd have been like programmers are like the set job, right? Like
• 20:47
  just, um, and that's not the case. So, you know, uh, still can be surprised, but
• 20:51
  I think, um, actually like launching an app. Uh, developing it, getting a
• 20:55
  marketing concept for it, going to market, I think is going to be gone pretty
• 20:59
  soon. I just don't think that makes a ton of sense. Like, I think it's going to
• 21:03
  be more of the bolted-together things that really are valuable. Um, um, and, you
• 21:07
  know, so I think one is if you're mediocre or the bottom of the barrel on a
• 21:11
  team, you're, you're going to be in trouble.
• 21:16
  Uh, yeah.
• 21:17
  I think the true, the true operators, like people that are good at their job,
• 21:21
  really, really good people at their careers are going to be just fine. And
• 21:25
  whatever that is, even if, even if AI does it, there's going to be a need for
• 21:29
  the people that can think it out. Um, and I think that, you know, trades and
• 21:34
  in-person events and this authenticity piece is going to even get bigger. I
• 21:38
  think it'll be a course correction, um, which I'm excited about, really.
• 21:42
  Yeah.
• 21:42
  Um, I love the authenticity piece and, and in-person stuff. And I think that
• 21:47
  there's just going to be a— sure, it's hard to tell what's an AI video and
• 21:51
  what's not now, and— but it's going to create this whole movement, counter
• 21:55
  movement to it, I guess, like parallel movement that will be like the real
• 22:00
  original people that are different, that, that really have what only a human can
• 22:04
  do, I think. Um, you know, as far as like unique experiences, because AI is
• 22:08
  really layers of what everyone else has done already, right? It's not really
• 22:12
  creating something. I have yet to see it create something just like brilliantly
• 22:17
  unique. It's, it's really just helping us do things that we're already doing,
• 22:21
  you know. Um, as someone that's been putting out content on Amazon for 15 years,
• 22:25
  I'm sure the AI agents have been learning from us, you know what I mean? Like,
• 22:30
  that's just the truth. Um, so I don't know, it's more of a gauge feeling than it
• 22:34
  is like it's exactly going to happen like this, you know. Um, but I think even
• 22:38
  as the AI gets smarter and smarter and smarter, it's still going to cost. Um,
• 22:42
  you know, the brainpower is still going to cost. And so there'll still be a
• 22:47
  limit in regards to like how much the normal person can use it for and without
• 22:51
  paying quite a bit, you know, because we've got some stuff now that we're able
• 22:55
  to do ourselves, but there's still quite a bit of cost to it, I think, whenever
• 23:00
  you're trying to do anything significant. Yeah.
• 23:04
  And that's something you have to factor. I mean, for the first time ever, we did
• 23:08
  2 new— our last 2 hires were AI guys to specifically board on board. You're
• 23:12
  going to focus on doing stuff on AI to help the operation get better, to get
• 23:15
  leaner.
• 23:19
  Where did they go to college? Uni, I guess, as they say.
• 23:23
  Yeah, well, funny enough, one of them's still at uni. He's doing computer
• 23:27
  science. And the other one did nutrition at university.
• 23:32
  Exactly. My point was like, my point was gonna be, I don't think there's really
• 23:37
  just similar to Amazon, right? And I've built a life for me and my family on a
• 23:41
  marketplace called Amazon that no one goes to school for. I think AI is another
• 23:45
  one of those arenas in a lot of ways. And I think having a programming
• 23:50
  background is incredible because if you want consistent output, you need to
• 23:54
  code, you know? And if you wanna work on a team, Even more so, you need to be
• 23:58
  able to code, um, correctly and organize and effectively and all those types of
• 24:03
  things, right? So, um, those characteristics of that learning, I think, is
• 24:07
  gonna— is a good background for, for AI. But I think AI is self-taught, like, in
• 24:11
  a lot of ways like that. I mean, there's—
• 24:16
  Well, it is. I would say on our team, I probably know more about AI and I can
• 24:19
  see more of the possibilities with AI. Um, and I, I'm quite happy that that's
• 24:23
  the case, uh, certainly at the moment.
• 24:26
  It's brought us relevant again in some way, right? Like, um, You know, I'm being
• 24:31
  honest, like, um, through the pandemic, outside the pandemic, like, as an agency
• 24:35
  owner in the marketplace space, it's been very confusing, you know, um, going—
• 24:39
  teams going offshore, um, the, the number of people doing Amazon or saying that
• 24:43
  they do Amazon in a big way, the competition got a lot louder. It was not as
• 24:47
  blue ocean as it was before. There's been a change of the guard, I think, a
• 24:52
  little bit in people understanding that there's teams that know what they're
• 24:56
  doing and teams that don't. Um, little plug. But also, um, you know, it's like,
• 25:00
  am I going to need to learn a new skill? Is this going to be just like, you
• 25:04
  know, done by agents? And, and I don't think so. I think that there's still so
• 25:08
  much nuance in really understanding how to run a great e-com brand that AI can
• 25:13
  kind of surface it all, but it can't run everything. You still got to be unique.
• 25:17
  But, um, all that to say, myself, like, part of e-commerce, part of what's hard
• 25:21
  is like you're constantly having to learn new things, learn new things, learn
• 25:25
  new things, learn new things. And is this skill gonna be outdated? These are
• 25:29
  like thoughts in my head, you know, am I gonna— is this gonna be something that
• 25:34
  just, you know, Joe off the street can do? Um, and now all of a sudden, I think
• 25:38
  because of all of the previous skills, programming for me in college and, and
• 25:42
  different things, or like building websites, things we did in the early part of
• 25:46
  our career, now that we have a team, all of a sudden again, I'm leading the way,
• 25:50
  um, you know, in my team. And it feels just kind of natural again, like it kind
• 25:55
  of reset to the early days. Um, And I didn't see that coming.
• 25:59
  No, and that's really good, isn't it? I mean, I would say I feel the same. I
• 26:03
  think there are people, there's like, there's a guy called Mark on our team
• 26:08
  who's our lead, he's our head of technical. So he looks after all the coding,
• 26:13
  you know, and all the technical aspects of whatever we do. And I would say he
• 26:17
  is, he still, when it comes to websites and e-com, He is better able to use AI
• 26:22
  to get the good stuff out of it, if that makes sense. I can take it to a— like,
• 26:27
  we're working on a project at the moment. I took it to a point and I'm like,
• 26:31
  Mark, I'm stuck. He can go away. Now, I know more about AI, but he knows more
• 26:36
  about coding, and he can look at what the AI's outputting and going, right, this
• 26:41
  is where the problem is right here. Why is it doing that? And he— I would never
• 26:46
  have seen that, but for him, he can see that, he can understand it, and he can
• 26:50
  get AI to help him. do his role better as the head of technical. I think
• 26:55
  probably where you and I are different, not only, I mean, you're a coder, so you
• 27:00
  would be able to do that much more than me. I guess where I see it is I can see
• 27:04
  again, we have this technology that will make the future different. So I'm much
• 27:09
  more visionary, I suppose, and much more kind of like, right guys, let's look at
• 27:14
  what we— because everyone's got their head down looking at their job.
• 27:18
  Yeah.
• 27:19
  you as the CEO can look up and you can survey what's going on and going, right,
• 27:22
  this is a big area of opportunity for us right here. And if we do something
• 27:26
  there, that's gonna, that's gonna make a big impact. Let's go there first.
• 27:29
  Yeah, it's like that, that learning. That's exactly what I'm trying to say, is
• 27:34
  it, um, is exactly that, which is gone a little bit from, okay, trying to master
• 27:38
  these skills to like being able to see visionary again instead of just
• 27:42
  perfecting what's here. It is like refreshing, it's super refreshing, and it
• 27:46
  feels like, ah, empowering to me in the e-com space. And I think the other part
• 27:50
  is I was talking about the communication, how refreshing it is to communicate an
• 27:54
  idea, you know? And when I think—
• 27:59
  Yeah.
• 27:59
  Whenever you're on a team that's working really hard, and my team works really
• 28:03
  hard, whenever I have ideas and I've— I'm just the idea person, I've always been
• 28:06
  that, but I trailblaze the way usually, and then the team comes and manicures
• 28:10
  the lawn, but I cut down the jungle, you know? Like, that's generally how the
• 28:14
  approach has been in every area, whether it's TikTok. I mastered TikTok.
• 28:18
  Walmart, Shopify. I jumped into these areas and I bring them to the team once
• 28:22
  I've kind of mapped it out. But it can be kind of like you hold yourself back,
• 28:26
  uh, in a lot of ways when I sharing it with the team, because I know in their
• 28:30
  head they're thinking, okay, if I come up with a new way I want to do proposals
• 28:33
  before AI or whatever, right? Or I have this idea for them, they're just like,
• 28:37
  how much more work is that for me? Right?
• 28:41
  Yeah.
• 28:42
  And, and so it gets into your head, you're like, I'm not going to share those
• 28:45
  because I know that's, that's how they're thinking. And I can't just brainstorm
• 28:48
  and go like this because for every single person I say it, okay, they have to
• 28:52
  make that come to life if they're on my team. And now I'm able to bring it to
• 28:55
  life without burdening them with that. And then, you know, and then can hand it
• 28:59
  off if I get it there or whatever. And that's, that's something that I think
• 29:02
  has— just to add a little bit more to why it's so liberating is because I'm not—
• 29:06
  my, my idea is not creating heaviness for someone on my team. And that's what I
• 29:09
  like.
• 29:12
  Are you almost handing them a finished product, aren't you?
• 29:16
  Yeah.
• 29:17
  Yeah. Like, here guys, I've done the heavy lifting. Let's, you know, and you're
• 29:22
  right. I mean, when I used to go away, like when I came to see you guys, I was
• 29:27
  away for like 5 or 6 weeks, I think in total. The team here, there's a part of
• 29:32
  the team, Mark, Michelle, and a few of the others, they're like, right, when you
• 29:37
  get back, please do not download every single great idea you had. Whilst you
• 29:42
  were away, right? We just don't want to hear it.
• 29:48
  They're already saying that. That's crazy.
• 29:50
  Exactly. Yeah, yeah. Because I would like— you would go away, you'd have like,
• 29:53
  you know, 25,000 brilliant ideas, and you come back and go, right guys, we're
• 29:57
  going to take this hill, this hill, this hill, and this hill. Let's go. And
• 30:00
  after a few years, they're just exhausted. Exhausted.
• 30:04
  Yeah.
• 30:05
  And they're just like, dude, please do not. We know. It's really funny, isn't
• 30:09
  it? But you're right, with AI now, I can go to the team and go, right guys,
• 30:12
  listen, this is where Like with Mark, this is where I think we should go. Here's
• 30:16
  code. I'm 80% there. I just need you now to go the 20%. And I've taken the 80%
• 30:20
  off his plate. And you could argue, well, why am I doing that 80%? Well,
• 30:24
  actually, I think by doing that with AI, it'll have taken me a few hours, but it
• 30:28
  was a hell of a lot quicker.
• 30:32
  Yeah.
• 30:32
  And we've got a much better outcome.
• 30:35
  And with AI versus other things in the past, like always trying to improve
• 30:39
  Martinology's branding, our assets, our things we're handing out. Uh, Mark, our
• 30:43
  own marketing would take a back seat a lot of times, and I've always been
• 30:47
  pushing that we need more on our own stuff. Um, is I think what's been amazing
• 30:51
  is I focused on one first. I did it for myself, like getting the agents to
• 30:55
  really help myself, where like, you know, it's helping me write emails and not,
• 30:59
  not drafting a generic thing. It's helping me pull all the facts from the
• 31:03
  meeting transcripts or whatever to be able to give them the best answer. I used
• 31:07
  to need to have a team meeting And so, you know, hey guys, I need— this client
• 31:11
  wants to talk to me. I'm not up to date on what's going on. Let's have a
• 31:16
  meeting, a meeting before the meeting, right? To get, to get my ducks in a row.
• 31:20
  Now I don't need them for that. I can get my ducks in a row kind of without them
• 31:24
  and save them that time, save the company time. And then once I started figuring
• 31:28
  out things for myself, my bookkeeper, the design team, like as far as things
• 31:32
  they used to help me with, I started helping them. And it wasn't to replace
• 31:36
  employees and save a dollar. It was if I was in their shoes, because I've done
• 31:40
  most of those jobs, what tools would help me be more efficient? What could I
• 31:44
  build for them that's going to make them better? So as far as getting them to
• 31:48
  adopt things, I think it's been easier too, because I'm not necessarily asking
• 31:52
  them to do more. I'm saying, here's a new tool. I found a drill for you or a new
• 31:56
  chainsaw or whatever, like, right? And so it's like, this is an easier way to
• 32:01
  get review aggregation, or this is an easier way to come up with new brainstorm
• 32:05
  ideas for PDP pages or, or concepts, or, hey, here's 8 different areas you can
• 32:09
  pull from to really get to the meat of why a listing's not converting. And, you
• 32:13
  know, handing them things that made their job easier versus handing them things
• 32:17
  that, like, they have to go build, you know. Yeah. Um, that's also made me feel
• 32:21
  good as an owner, as a leader, is like the work that I'm doing is making my
• 32:25
  team's quality of life better, or the work better, or one of those things.
• 32:30
  Yeah, everyone wins in that scenario, right? Everyone wins.
• 32:33
  Like, like a simple example, just because like, okay, the podcast episode comes
• 32:37
  out, right? Before, I had to almost get someone off of marketplace work. You
• 32:42
  know, I didn't have like 5, 6 people on my own team working on our own
• 32:46
  marketing. They generally were like on marketplace, Amazon or TikTok, creating
• 32:51
  content. And then we're repurposing our skills for, for Marknology marketing.
• 32:55
  And, um, you know, you'd have to If you sent me, hey, here's the podcast, here's
• 32:59
  some thumbnails, here's like when it's coming out, like send an email to your
• 33:04
  list or post on LinkedIn or share it. Okay, that's an action. So the email would
• 33:08
  come to me, I forward it to my team. Hey guys, can we get this loaded in the
• 33:13
  socials? Um, you know, maybe that's LinkedIn, maybe that's an Instagram post
• 33:17
  about it, maybe that's, uh, an email in Klaviyo, maybe that's a blog post, maybe
• 33:21
  that's a newsletter, right? There's like 5, 6 things. It's an extra— it's extra
• 33:26
  work. Right. And it was, you know, on top of their deliverables all the time.
• 33:30
  But there's some things that happen ad hoc. You can't plan out your whole year.
• 33:35
  You've got episodes and things like this, you know. So now it's an email comes
• 33:39
  through, I forward it to my agent that's in charge, right? And he automatically
• 33:43
  knows from a mixture of the routing rules and a mixture of like his own
• 33:48
  intelligence, right? Like how I like my Klaviyo emails built, or, you know, he
• 33:52
  knows that if it's a If it's a podcast, it goes, it goes on my media site, on my
• 33:57
  media page, it goes into a newsletter. We wait till I have 2 podcasts, I kind of
• 34:01
  blast them out. We create a blog on it. You know, I also check if I'm a guest, I
• 34:05
  check for backlinks on their site when they post it. And if not, I draft an
• 34:10
  email to request them for that, tell them I built a page and done the same thing
• 34:14
  for them. That's like, you know, 10, 12, 15 actions. And I cut it up into minis
• 34:19
  as well in case they don't come. And like, that happens with an email forward.
• 34:24
  Yeah.
• 34:24
  And that's like, to me, that's incredible because I'm just— I'm getting more out
• 34:28
  of the work I'm already doing. And that is like more, almost more exciting to me
• 34:31
  than net new, if that makes sense. Yeah, yeah, yeah.
• 34:35
  No, totally. I mean, we do the same here. Remarkable what it is. What are your
• 34:39
  AI tools of choice right now?
• 34:44
  Well, um, I went with OpenCLAW. You know, I know there's Hermes, there's a lot
• 34:49
  of other ones that people can use. And what I also saw when I was really
• 34:53
  studying all of this was I was watching all of the YouTubes. I spent like a
• 34:58
  month just learning, ingesting before I started my OpenCLAW. And then because of
• 35:02
  security, and that's what I went to college for, I did do it on a local network.
• 35:07
  And I started expanding it out to Slack through my— to my team could interact
• 35:11
  with it through Slack. And then I I kind of had some stages that I wanted to be
• 35:16
  safe and whatever. But, um, one was like, okay, I would read. And a lot of the
• 35:20
  YouTubers or the people out there putting out content, they're smart. They know
• 35:25
  how to create content. They know what subject is popular, but they don't have a
• 35:29
  use case like I do. I think that's a main game changer in what I'm doing versus
• 35:34
  other people is my use case is my everyday work. It's right here. It's 60
• 35:38
  Brands. It's my stuff, right? And so watching the YouTubes was informative, but
• 35:43
  it wasn't the same as just jumping in. So I also got OpenCLAW and then you see
• 35:47
  OpenCLAW versus Hermes, OpenCLAW versus Gemini, Gemini versus Anthropic. Like,
• 35:51
  and there's something lost in jumping. There's like a lot of time and whatever.
• 35:56
  Just even if this has a little advantage, like I need to deepen this
• 36:00
  relationship, you know? So I've kind of committed to that, I guess. I use
• 36:05
  Anthropic as kind of the backend and Um, but I have some local machine stuff
• 36:09
  and, you know, but, and then, um, and then I— OpenClaw is not, is not, uh, stuck
• 36:14
  to one or the other. You could have 8, 10 different, like, you know, memberships
• 36:18
  or APIs that you're connecting to. Um, so I'm using OpenClaw, Anthropic. Um, I
• 36:23
  have lots of little tools like gamma.app, which is like kind of a proposal
• 36:27
  builder, um, and moved away from them. Honestly, I, I've brought them in and
• 36:32
  then kind of moved away at the same time on a bunch of stuff. Needed it at the
• 36:36
  beginning to learn it, then realized like I can build that myself or, or I don't
• 36:41
  need that thing or I want to do custom. So it's a lot of custom as well. Um, and
• 36:45
  then I've got like, you know, the SQL servers and the, and all that kind of
• 36:50
  backend crap happening. That's like not necessarily AI, but it's, it's been
• 36:54
  integrating with AI. And so.
• 36:58
  Yeah.
• 36:59
  I think that's kind of where it's, it's at is like some of it's AI. Sure. What
• 37:03
  is actually AI and what is AI helping you build? Like the APIs has been the
• 37:07
  biggest leverage for me. Um, it used to need almost like a specialist in that
• 37:11
  API to get, to get stuff. And now you can just plug in, which is incredible. Um,
• 37:15
  but to that end, when something new comes out that I really like, like, like
• 37:19
  let's say Hermes or a competitor, that's like a switch. What I do instead is I
• 37:23
  go and I have the machine learn everything it can about that build. And then we
• 37:27
  determine kind of like, is that better than what we have? Is that something we
• 37:31
  should ingest? Is that something we should bolt alongside? Is that something we
• 37:35
  should try to replicate? Is that worth the build? And so instead of changing
• 37:39
  mine as these different good options have come out, I've just kind of custom
• 37:42
  built mine based on those new things, if that makes sense. So, um, some of them
• 37:46
  will be like, well, it's cheaper because it compacts the information so much
• 37:50
  better. The other one might have a dashboard. The other one might the agents
• 37:54
  better, you know, these types of aspects. And what I've done is just learned
• 37:58
  from those ideas and then built them myself.
• 38:02
  Yeah. Well, that's fair play because I think you're right. I think there is a, a
• 38:05
  massive opportunity cost when you move.
• 38:09
  Yeah.
• 38:09
  That we don't talk about. And we did the same. We would jump between Claude,
• 38:13
  ChatGPT, and it's just like, oh, now this is better, and ChatGPT-5 is better
• 38:17
  than Opus 4. And then you're just like, and after a while we're like, No, let's
• 38:21
  just nail our guns, you know, let's just nail the mast and let's just, let's
• 38:26
  just stick with one system and really get to know how that system works well for
• 38:30
  us and really hammer that thing home. Is it the best system? I don't know. I
• 38:34
  don't care. It's the best system for us right now. And the, we've just got
• 38:38
  really good at using that one thing.
• 38:43
  What, what, what I will say to that is there is a little like caution for me,
• 38:47
  which is like, if let's say Anthropic gets shut down by the government. Because
• 38:51
  it refused Trump on some stuff, and then Trump's coming down hard on him and he
• 38:55
  switches to OpenAI. And, you know, that's a lot of funding and a lot of change.
• 38:59
  What if I had built my whole system on Anthropic and then it just gets like its
• 39:03
  feet, you know, the rug pulled out? So with the OpenCLAW system, or, or what I
• 39:07
  call Winston— my system is called Winston.
• 39:11
  Love that.
• 39:12
  Um, um, it's, it's, uh, it's AI, um, agnostic in regards to like, okay, if
• 39:17
  Anthropic went down, which is what I'm comfortable using, and, and, you know,
• 39:21
  we're all using Claude, whatever, I could switch to OpenAI and my system would
• 39:25
  be okay.
• 39:29
  Uh, yeah.
• 39:30
  So it has like fallbacks, if that makes sense. And that was important for me as
• 39:33
  much as I'm com— very comfortable with Claude and I've, I've built all that all
• 39:36
  into that, that I, that if it, if it did go, everything I've built wouldn't be
• 39:39
  for, for naught.
• 39:42
  Yeah. Yeah, so that's very wise. Very wise. Um, it's—
• 39:48
  I—
• 39:48
  have you played around with— so you've obviously got the frontier models, right?
• 39:52
  So like Anthropic, like ChatGPT, but you've also got the open source models
• 39:56
  which you can install on your own network, on your own system. Have you played
• 40:00
  around much with those?
• 40:05
  Yeah. Um, here's the thing. I do feel like sometimes people are pinching pennies
• 40:09
  when they— when I hear them talking about AI, and I'm just like, you're not
• 40:14
  really about it. You're talking about saving cents. Like, you know, what are we
• 40:18
  doing? What are we talking about? Um, and I just— I do— I hear people and I'm
• 40:23
  just like, we're talking about $20. What are we talking about? Or like this
• 40:28
  membership or this membership, right? Um, and the local models are great for
• 40:32
  like, for example, I, I pull all my emails, my meeting transcripts every night.
• 40:37
  Um, and I'm not using Opus or Fable 5 to do that. I'm using a local LLM on my
• 40:41
  server that's pulling those emails and ingesting them or pulling Amazon stuff.
• 40:46
  It's not using Opus, but I have a different process for what's getting ingested
• 40:51
  and tagged and organized in the system versus what goes out to customers or
• 40:55
  what's client-facing.
• 41:00
  Yeah.
• 41:00
  So cheap model in, expensive model out. When you think about like what, let's
• 41:05
  say it's helping me write an email or build a deck, like what am I doing using a
• 41:09
  cheap model to help me do the math behind like a proposal deck or the build? No,
• 41:14
  if it's recreating myself, I'm not going dumber. Like if that makes sense,
• 41:19
  right? So, um, I've always felt that on the, I try to get, that's why I try to
• 41:23
  get all the way through the rules, that example I gave you, the rules, the
• 41:28
  routing, the context. All of that is actually free, the pool. So by the time it
• 41:33
  gets to my, to my agent who actually does the, the reasoning, which is like the
• 41:38
  very final layer, so to speak, um, I do it at the basically the most expensive
• 41:42
  or the, the most suited, not, not every task. And then, but all of the work
• 41:47
  before that, the coding has to be built and all of that. But like once it's
• 41:52
  built, that context that it surfaces is actually a free call on my own machine.
• 41:56
  Wow.
• 41:57
  So that's, that's also what gets me excited about that context build I built is,
• 42:01
  is like all of that work is— it's not pulling the API, it's not searching the
• 42:05
  MCP, it's not searching my chats, my pro— it's like already there from my
• 42:09
  system. Uh, I think I use Ollama, um, locally. Um, but, uh, then it gets to the
• 42:13
  AI and I'm using Opus and I, and I kind of don't, um, compromise on that. I just
• 42:17
  think anything that's client-facing that's that's trying to be me is not
• 42:21
  something I want to go cheap on.
• 42:25
  Yeah, no, that's very, very wise. We're the same. I just use Fable all the time.
• 42:29
  Yeah, and I know it burns through your tokens, and I say all the time I don't—
• 42:32
  that's not true. Um, but the important stuff, I just turn it on. I'm like, I
• 42:35
  don't— I just want— I just want it done as best as I can with the model that I
• 42:39
  have at the moment, right? Why would I not?
• 42:42
  If it's saving me time, I'm— I'm not, you know, I'm not pinching pennies there.
• 42:46
  And the difference is drastic. Like, and I'm sure that's like the Apple iPhone
• 42:50
  when it comes out, right? They start degrading your old one, right? And I'm sure
• 42:54
  that that's the case in, in the AI models. Like, this is still business in the
• 42:58
  world we live in, right? Um, yeah, but how much time is that worth for me to sit
• 43:02
  there and let something like Sonnet reason, reason on, like, you know, the
• 43:06
  Sonnet model or something, reason on something very important to me? Or, you
• 43:10
  know, the difference I see it struggling, struggling, struggling, switch it over
• 43:14
  and it fixes it. You know, so, um, to me it's not just about, uh, I, and I am
• 43:19
  trying to be efficient there all the time, but I kind of go in spurts and it's
• 43:23
  like, go build. And then, you know, um, I think the biggest thing for me was,
• 43:27
  um, I built this OpenCLAW system that runs through an API and it was acting like
• 43:31
  an assistant where it's always on. It's not like coding, but really just
• 43:35
  figuring out Claude Code on the side as well. And like getting, I was kind of
• 43:39
  doing one, getting really good at one versus the other. And just leaning on
• 43:43
  Claude Code to do a lot of my building, you know, like the actual coding of apps
• 43:47
  and my systems and things like that. And then using the OpenCLAW system, which
• 43:51
  is like always on and it connects through, it's illegal to use it on like a max
• 43:55
  account.
• 43:59
  Got it.
• 43:59
  So you're using it through API, which is more expensive, but it can do a lot
• 44:04
  more. It's always there. It can kind of like, it's not gonna push back on you.
• 44:08
  And even in the few months since AI came out to now, um, the limits used to be
• 44:12
  crazy. Um, you know, and if you're doing— there's a big difference in if you're
• 44:16
  doing like proposal building. And where I really learned was like, okay, it just
• 44:20
  keeps compacting, losing my context. This is ridiculous. I'd rather just do the
• 44:24
  work myself. Like, you know, when it was compacting, uh, this is frustrating.
• 44:28
  It's losing everything I just taught it.
• 44:33
  Exactly.
• 44:33
  So then the, the, the mode is to say, okay, stop clearing the context. Which it
• 44:37
  builds, but then it remembered everything, but it was getting super expensive,
• 44:41
  crazy money, but it was remembering everything. And I'm like, oh my God, this
• 44:45
  workload I'm cranking out through this workload, this is amazing. But it was
• 44:49
  getting so expensive. So I'm like, I can't get it to remember everything. Think
• 44:53
  of it like us, you know, like the metaphors are crazy, but like, you know, our
• 44:57
  context, it's big. Okay, it's working, but that's not sustainable. I can't keep
• 45:01
  doing that. That's expensive. And then, um, and if you're like coding with that,
• 45:05
  that chat interface, it's like, uh, it gets confused if you're switching
• 45:09
  conversations and the compacting. So Claude Code and, and to build things out
• 45:14
  with GitHub and the branches. And for people that don't understand that, like,
• 45:18
  you know, work on a branch at a time and code and, and then come back. And, um,
• 45:22
  yeah, it just depends on what your use case is for, you know? And for me, um,
• 45:26
  Yeah. Creating high-level— like, I needed to do high-level things. Um, I think
• 45:30
  one thing I want to share is that, like, a lot of people talk about creating
• 45:34
  multiple agents, that, like, you know, you've got a CFO agent, you've got a
• 45:38
  design agent, you've got this, you've got that. And there is value in that, in
• 45:42
  having a prompt where you're giving them context about what they're doing. But I
• 45:46
  think with my rules and my code and the way I've built it, where, like, the AI
• 45:50
  is just really acting at the end of a lot of things Um, and inside smaller
• 45:54
  sessions. I've gone away from that. And instead what I've built is, is a, is a
• 45:59
  super agent, a Drew IQ.
• 46:03
  Um, yeah.
• 46:04
  Like that really has everything that I want in one super agent. And I've spent
• 46:08
  my time to get that agent like really smart and know where all my stuff is and
• 46:12
  really be able to help, help me instead of this idea of like, you know, 20
• 46:15
  different specified agents that do separate things and jump in on, on separate
• 46:19
  things. I, I had the Discord where they were all communicating and arguing and
• 46:23
  chatting and Tried all of the things. I thought that was kind of the only way to
• 46:27
  do it was kind of have these specialist agents with names and part of a team.
• 46:31
  And that is, that's just like if you get on YouTube, if you start watching
• 46:35
  stuff, that's how everyone's doing a lot of things. For me, it was just easier
• 46:39
  to get one that was like as close to me or brilliant as possible instead of
• 46:43
  working on 20 different personalities and, and all the skills. And, um, it's
• 46:47
  worked a lot better for me.
• 46:50
  That's a really interesting idea. Yeah. Yeah. Wait, what, what are you finding
• 46:55
  with AI and Amazon then in the Amazon space? What's working well there?
• 47:00
  So, um, you know, I think before there was like, before I plugged everything in,
• 47:04
  I've got Amazon Ads API, I've got Seller SP API, which is the other one. I've
• 47:09
  got Amazon Marketing Stream, which is the ad spend every 15 minutes. Amazon
• 47:14
  Marketing Cloud. So all the data points. Okay. Now, and I used to get them
• 47:18
  through software, other software. I still do in some aspects, like, right, I'm
• 47:23
  pulling through their APIs, they clean them up, they manage them. On the Amazon
• 47:28
  side, for me, first I started with my work, which was kind of different. And
• 47:33
  it's really been the last few months I've built something called Ecom Pulse. I
• 47:37
  was building it for 2 years before AI came out. Which is really grabbing all the
• 47:42
  data points from all the important marketplaces like Shopify, Walmart, TikTok
• 47:47
  Shop, Amazon, a P&L, for example, and pulling it all into one spot so you can
• 47:51
  see the halo effect. You can see like my margin on this channel versus this
• 47:56
  channel. You can see your Google and Meta ads. Um, and people say they have
• 48:01
  omnichannel, but they don't. There's not really software out there, in my
• 48:05
  opinion, that does it great, at least not with Amazon forward thinking. And so
• 48:10
  for me, I've been working to kind of pull all that together with AI on top of
• 48:15
  it. Because the truth is this, I always kind of lead with this. Drew, if you had
• 48:20
  all the time in the world on this thing, what would you do and how would you do
• 48:24
  it? Because I do believe that the real leverage pieces are the non-scalable
• 48:29
  things.
• 48:35
  Okay?
• 48:35
  Like in general, like if I could grab this report and this report, and I could
• 48:39
  do a Reddit search and I could grab all the reviews from me and all the reviews
• 48:43
  from my competitors, and I would know what pricing they're at today. And
• 48:47
  tomorrow I'd have a refresh and know what pricing they're at tomorrow. And I
• 48:52
  would have my margin information, my inventory information, my ranking report,
• 48:56
  my search query performance organic report. I knew what the Meta and Google team
• 49:00
  was doing as far as pushing off Amazon traffic. I knew what was happening in the
• 49:04
  economy at the time. I knew like when our shoppers were buying, what age our— if
• 49:08
  If I had all of that, like in a perfect world, if I had all the time in the
• 49:12
  world and I could cross-reference it, what would I do? You know, and that's what
• 49:16
  a good Amazon pro has been doing for years with spreadsheets is like, okay, I'm
• 49:20
  looking at this keyword ranking and we're advertising on this exact match. And
• 49:24
  at just at scale, building an agency past an operator of one, that gets very
• 49:28
  hard to teach. It gets too— that we don't get paid enough to do that at that
• 49:32
  level. You know, there's a million reasons why. That's like the rocket science
• 49:36
  that we can't do at scale. But at one, a one-off project, if I wanna spend 4
• 49:40
  hours on an afternoon on a Saturday, I can get a lot of that together on a
• 49:45
  product, right? And so you say what AI is doing, I think it's helping me create
• 49:49
  what we would only do like as a pro, like once in a while when I can, to being
• 49:53
  able to do that more often. And really, really helped me like really
• 49:57
  troubleshoot and give people insights on their brand that I don't think in some
• 50:01
  ways they've ever had before. I've been having a lot of fun really deep diving
• 50:05
  some, some projects where the click-through rate's high, they're clicking on the
• 50:09
  main image, but they're getting in and they're not converting. And why are they
• 50:13
  not converting? And what are, you know, what's the difference in the keywords
• 50:17
  we're going after versus what they're saying on Reddit or what they're saying in
• 50:21
  our reviews or like the kind of the tech speak behind the products. And, um,
• 50:25
  just that layer of like knowing what questions to ask the system, you need to
• 50:29
  know all those places to go to get all that information and what it needs to
• 50:34
  think about, but being able to really pull that together and then put that into
• 50:38
  a presentation that's not just like, you know, ChatGPT agreeing with you. Okay.
• 50:42
  Like, 'cause it's got, you know, some pushback and really being able to find
• 50:46
  some killer insights that I feel like have always been there. I've just never
• 50:50
  had the time to get them. So that's a long answer, a long way of saying like, I
• 50:54
  think it's helping us do things that we couldn't do at scale before. And that's
• 50:58
  why Amazon and marketing and a lot of these things got so SOP driven and so Just
• 51:02
  templaty because we're cranking out work. No one wants to pay for that type of
• 51:06
  like almost like advertiser type of work to really bring a product to life.
• 51:10
  Yeah, I mean, I think that's really what it's doing is helping scale the
• 51:14
  unscalable.
• 51:18
  That's fantastic. Drew, listen, man, I'm aware of time and how it's rapidly
• 51:22
  gone. We've just got started, I feel. If people want to find out more about you,
• 51:26
  find out about Marknology, what you guys are doing on Amazon, TikTok, etc.,
• 51:30
  what's the best place to go to?
• 51:34
  Yeah, you know, if you go to marknology.com, I have like newsletter signups
• 51:38
  throughout the site. So it's just easy if you want to kind of get in when I'm
• 51:43
  speaking or a webinar, blogs and stuff I'm putting out. I'm definitely trying to
• 51:47
  bring a lot more of that forward and kind of just share. I like sharing what I'm
• 51:51
  working on when I'm working on it because it's my most passionate content. You
• 51:55
  know, I think Versus kind of the, the old. So marknology.com. I'm also on
• 52:00
  LinkedIn. I'm on Instagram, Andrew Morgans. Um, Instagram's a little more
• 52:04
  personal and fun entrepreneurial type of thinking and, and sharing. And, um,
• 52:08
  LinkedIn is more of the business side for sure. Um, to me they're both, they're
• 52:13
  interconnected, you know, like, so, um, love to interact with anyone there.
• 52:17
  YouTube, whatever channel you guys are on, um, would love to hang out.
• 52:22
  Fantastic. Well, we will of course put all of those links in the show notes. Uh,
• 52:26
  so whatever podcast player you're on, you'll be able to see it in the
• 52:29
  description. If you're on YouTube, you'll be able to see the links for Drew in
• 52:33
  the description. Of course, if you're subscribed to our newsletter, which will
• 52:36
  not be anywhere near as good as Drew's, I'm sure, uh, but if you're subscribed
• 52:40
  to that, it will also be in that newsletter as well. But, um, Drew, listen, man,
• 52:43
  thank you so much for coming back on. We should do this again sooner. We
• 52:47
  shouldn't wait so long for the next one.
• 52:51
  I would love— I got about 50 questions for you. As well. So, um, we'll have to
• 52:53
  make it happen.
• 52:56
  Absolutely. Drew, you're a legend. Thanks, man. Appreciate you. Thank you for
• 52:59
  coming on. Uh, it's been, it's been brilliant.
• 53:03
  Thanks for having me.
• 53:04
  Wow, there you go. What another brilliant episode of the eCommerce Podcast. What
• 53:08
  a legend Drew is. Uh, but that's it this week from me. Thank you so much for
• 53:11
  joining us. Like I say, make sure you like and subscribe and do all of that good
• 53:15
  stuff because I'll be back next week with another great conversation. But from
• 53:18
  me, until then, that's it. Bye for now.