Latest / AI DTC WTF / Will Sartorius's The 3-Part Framework That Tells You Exactly Which Ads to Make Next

Transcript

Will Sartorius's The 3-Part Framework That Tells You Exactly Which Ads to Make Next · AI DTC WTF

• 0:00
  If you work in D2C and you use AI and
• 0:05
  you're wondering what the
• 0:06
  F is going on every week,
• 0:08
  this is your podcast. The AI D2C WTF
• 0:11
  podcast. Your home for
• 0:13
  tactical tips, strategies,
• 0:14
  and ideas that you can implement right
• 0:16
  now in your AI workflows to make your
• 0:18
  brand or agency more money.
• 0:22
  Welcome to a new episode of AI D2C WTF.
• 0:26
  We have Andrew and Will here.
• 0:29
  Andrew is working at 50% capacity because
• 0:32
  he's been going to the opera too often
• 0:34
  and taking stage and
• 0:35
  singing a little bit too much.
• 0:37
  Yeah, people don't know this, but I
• 0:39
  actually smoke a pack and a half a day
• 0:42
  and it's finally just affected me.
• 0:45
  You know, I'm also a
• 0:46
  part-time performing artist.
• 0:47
  You're a Marlboro Red guy, right?
• 0:49
  Big Marlboro Red guy. Yeah. And it's
• 0:51
  finally affected me.
• 0:52
  And the reality is I have a six-year-old
• 0:55
  and, you know, she licks my
• 0:59
  face, you know, or something.
• 1:01
  You know, it's like kids do gross stuff
• 1:02
  and that's what happens.
• 1:03
  So I'm sorry. Maybe
• 1:04
  some people like this.
• 1:06
  You know, maybe I'll have to like 11 labs
• 1:07
  and like keep this on because people are
• 1:10
  going to like the smokiness of it.
• 1:12
  You know, some people like a smoky me and
• 1:14
  some people like a smoky voice.
• 1:16
  And that's kind of what we're doing here.
• 1:17
  You're the most
• 1:18
  interesting man in marketing.
• 1:20
  Yeah, exactly. Exactly.
• 1:23
  Yeah, well, we're glad that you're here.
• 1:26
  Thank you for listening.
• 1:26
  There's been a lot of interesting new
• 1:28
  stuff that I would love to
• 1:32
  go through with you, Will.
• 1:33
  What's some of the
• 1:34
  newest stuff that's come out?
• 1:35
  I mean, there's so much.
• 1:37
  Claude came out with
• 1:38
  a new model, BT Dubs.
• 1:40
  What's your take on it?
• 1:41
  What can it do that the others can't thus
• 1:43
  far based on what you're seeing?
• 1:44
  Yeah. So, yeah, the two big drops this
• 1:48
  week or I guess over the past week and a
• 1:50
  half were Google I.O., of course, which
• 1:53
  released Omni, which is
• 1:54
  their new video model.
• 1:56
  And then obviously Opus
• 1:57
  4.8, which came out yesterday.
• 1:59
  So I can't say I have a clear sense of
• 2:03
  how much better it is than 4.7.
• 2:05
  What I can say is it uses a lot of tokens
• 2:08
  for commensurate
• 2:09
  tasks and compared to 4.7.
• 2:12
  So, for example, last night I was
• 2:16
  generating like 20
• 2:17
  static ads for a client.
• 2:19
  They were just in Wirecutter.
• 2:20
  So we just wanted to spin up 20 ads with
• 2:23
  the Wirecutter branding.
• 2:24
  Just get that out
• 2:25
  there sort of ASAP Rocky.
• 2:27
  And I think it used like 850,000 tokens
• 2:30
  to write copy for 20 ads.
• 2:33
  Oh, my God.
• 2:33
  Which is pretty crazy.
• 2:34
  I will say, though, we have a pretty
• 2:37
  intricate process where we have 10
• 2:39
  copywriting agents and each of the agents
• 2:41
  have to give the copy a score of 90 or
• 2:44
  higher or else it doesn't pass.
• 2:47
  And so historically, I think what was
• 2:49
  happening is it would just
• 2:50
  be like, ah, you know what?
• 2:51
  This is good enough.
• 2:52
  But I actually think I was on Opus 4.8
• 2:55
  max at first and then I went down to
• 2:57
  extra and I found that actually the copy
• 3:00
  that it wrote, granted it took three
• 3:02
  hours, was very good.
• 3:04
  Like probably the best copy
• 3:05
  I've seen come out of Claude.
• 3:07
  And the other thing I found is that if
• 3:10
  you have a system with a handful of
• 3:13
  different skills marked on files, Opus
• 3:16
  4.7, 4.6, they would skip a lot of steps
• 3:19
  and sort of say like, oh, you know what?
• 3:20
  I forgot that step.
• 3:21
  Let me go back and do it now that you've
• 3:23
  sort of called me out.
• 3:25
  What I found is that Opus 4.8 doesn't
• 3:28
  necessarily skip as many steps.
• 3:29
  It's more intricate in reading your
• 3:32
  instructions, which
• 3:34
  is very, very helpful.
• 3:35
  And then on top of that, when you do
• 3:37
  slash goal, it very much, it feels like
• 3:41
  when you do slash goal and give it a
• 3:44
  specific end goal, it feels like it holds
• 3:49
  itself to a higher standard in the sense
• 3:51
  that like I need to do all these steps
• 3:52
  properly and not just like use my sort of
• 3:55
  human companion to bounce ideas off of.
• 3:59
  I candidly haven't
• 4:00
  been in Codex too much.
• 4:02
  I know everyone is telling me I should be
• 4:04
  doing more in Codex.
• 4:07
  I, you know, we use GPT-2 for
• 4:10
  image and static generation.
• 4:12
  I'm just like so, you know, embedded in
• 4:14
  Claude code at this point.
• 4:16
  And I just like Anthropic as
• 4:18
  a company more than OpenAI.
• 4:20
  So that's just like sort
• 4:22
  of like a morality thing.
• 4:25
  Maybe they're both sort of evil.
• 4:26
  But at the end of the day, yeah, I think
• 4:29
  4.8 is better overall.
• 4:33
  I think the reasoning and the token usage
• 4:37
  are definitely correlated
• 4:38
  and they're definitely up.
• 4:40
  But overall, I would say from what I've
• 4:42
  seen thus far, the copywriting is, as it
• 4:45
  pertains to marketing, is
• 4:46
  superior than that of 4.7.
• 4:48
  Of course, you need to be giving it the
• 4:49
  right parameters to get good copy.
• 4:51
  I mean, what about this new Google thing
• 4:54
  that's out with the video?
• 4:55
  What do you think about that?
• 4:56
  Yeah.
• 4:57
  So Google Omni, this is like sort of Veo
• 4:59
  3.4, I guess, technically.
• 5:04
  I think it's great.
• 5:05
  I think it's, again, probably on par with
• 5:07
  C-Dance, which is the ByteDance model.
• 5:09
  They're very, very similar.
• 5:11
  What I will say, though, is C-Dance gets
• 5:14
  pretty expensive pretty fast.
• 5:15
  If you wanted to do a full claymation
• 5:18
  Pixar ad, those things
• 5:19
  are all the rage these days.
• 5:21
  Or maybe they're gone.
• 5:23
  Who knows?
• 5:24
  It probably costs you $20 to $30.
• 5:26
  What's cool about Google is if you use
• 5:30
  Google Flow and just pay for a monthly
• 5:32
  subscription, I think it's like $200.
• 5:34
  You effectively get
• 5:35
  unlimited usage of their models.
• 5:37
  So you're getting $200 and you can use
• 5:41
  Nano Banana, Veo 3.1,
• 5:45
  Veo 3.2, up to Omni.
• 5:47
  And it just feels
• 5:48
  like that's a no-brainer.
• 5:50
  And I think that's just
• 5:50
  like a slept-on thing.
• 5:51
  People are probably just using Higgs
• 5:55
  field or file.ai or going
• 5:56
  directly to the interface.
• 5:57
  But yeah, Google Flow is great.
• 5:59
  We just built an account and that's what
• 6:01
  our team uses when
• 6:03
  we're generating videos.
• 6:04
  So now we're saving money because we're
• 6:06
  not really using C-Dance anymore.
• 6:07
  We're able to utilize
• 6:09
  Omni within this channel.
• 6:11
  I will say, you know, one thing I've
• 6:12
  noticed is that our file.ai where we're
• 6:15
  generating videos and
• 6:16
  static bill has gone way up.
• 6:19
  It's not necessarily a bad thing because
• 6:20
  we're generating a lot more statics.
• 6:22
  And we can sort of get
• 6:23
  into that in another episode.
• 6:27
  But ultimately, yeah,
• 6:28
  the Omni model is great.
• 6:30
  It's super solid, I would say.
• 6:33
  Probably on par, if not the
• 6:34
  best video model right now.
• 6:36
  Like if you want to animate
• 6:37
  statics, it's ridiculously easy.
• 6:39
  Also, product fidelity, it
• 6:40
  does a really good job with.
• 6:42
  Generate the ad first
• 6:43
  in GPT-2 as an image.
• 6:45
  Use that as a reference.
• 6:47
  And then upload that to Omni.
• 6:49
  And you can get a pretty good video.
• 6:54
  The other thing that's sort of like top
• 6:55
  of mind right now is,
• 6:56
  was it June 9th?
• 6:57
  In New York, you're going to have to
• 6:59
  start disclosing if
• 7:00
  you're using AI creators.
• 7:02
  So that's just like
• 7:03
  something to be on folks' radar.
• 7:06
  You'll need a disclosure there.
• 7:07
  One thing, you know, I was on a sales
• 7:09
  call earlier and they
• 7:10
  were saying they're nervous.
• 7:11
  This is a, you know, a larger brand.
• 7:13
  And they're very nervous even about using
• 7:14
  like hands or feet for AI.
• 7:16
  So, you know, I did
• 7:17
  some deep dive in there.
• 7:18
  And I know, Andrew, you are, you know,
• 7:20
  you probably have more
• 7:21
  experience with law than me.
• 7:23
  But it seems like there's a lot of gray
• 7:24
  area in this thing where there isn't a
• 7:27
  lot of explicit language
• 7:29
  about certain use cases.
• 7:31
  So it's hard to say for sure, like what
• 7:34
  is going to consist of this
• 7:36
  sort of AI generated content.
• 7:38
  Like if you have an AI persona talking
• 7:40
  and applauding a product like that is
• 7:43
  going to be problematic.
• 7:44
  But if you're just showing a hand, like I
• 7:45
  can't see that
• 7:46
  necessarily being an issue.
• 7:48
  But invariably someone's going to get
• 7:50
  sued or I guess, and we'll have to sort
• 7:53
  of figure that out as we go.
• 7:55
  But yeah, June 9th,
• 7:55
  it's approaching quickly.
• 7:57
  Something just behind folks' radar.
• 7:58
  Good shout.
• 7:59
  The legality thing, I mean, we'll
• 8:01
  continue to see this, you know, change
• 8:03
  and evolve over time.
• 8:04
  I think there's a couple of things you
• 8:06
  never want to under, sorry, you never
• 8:09
  want to overestimate,
• 8:10
  which is the competency of Congress to
• 8:11
  actually legislate
• 8:12
  something, which they won't.
• 8:15
  And you also never want to underestimate
• 8:17
  the power of what lobbying can do, of
• 8:21
  which there will be an
• 8:22
  anthropic and others.
• 8:24
  It'll be, it'll go the way of meta and,
• 8:27
  you know, social networks.
• 8:28
  It's not going to be
• 8:29
  really legislated a lot.
• 8:31
  So I wouldn't necessarily think about it
• 8:34
  too much, but state law will
• 8:36
  become where the issues are.
• 8:37
  So you just have to kind of be aware of
• 8:40
  that, be ready for that.
• 8:41
  Any other major releases
• 8:42
  we want to talk about, Will?
• 8:43
  I think that is sort of the gist of it.
• 8:47
  All right, cool.
• 8:47
  Well, good wrap up and let's go ahead and
• 8:49
  get into the episode.
• 8:49
  Today, we're really talking about, Will,
• 8:56
  what you're doing with AI.
• 8:58
  I think the people know about you.
• 8:59
  They know that you do with AI creative,
• 9:01
  but, but like, let's
• 9:03
  get into the high level
• 9:04
  and then let's go really specific about
• 9:05
  what you think is the
• 9:06
  future of this and all that.
• 9:08
  But first of all, so like, what are you,
• 9:09
  what are you doing with AI
• 9:10
  right now in your company?
• 9:12
  Yeah, I would say AI.
• 9:14
  To sell.
• 9:16
  To sell.
• 9:16
  Okay.
• 9:17
  Yeah.
• 9:17
  To sell.
• 9:18
  Sure.
• 9:18
  To sell first, let's frame it up.
• 9:20
  Yeah.
• 9:20
  Okay.
• 9:21
  So great question.
• 9:22
  AI has made making sales
• 9:24
  proposals so much better.
• 9:28
  So when I get off of a sales call, what
• 9:30
  I've created is
• 9:31
  effectively a markdown file.
• 9:33
  So, you know, we use read.ai.
• 9:35
  It's the same thing as
• 9:36
  Granola, you know, AI recording.
• 9:38
  So what I do is after a sales call, I
• 9:41
  take the recording from the sales call.
• 9:43
  I upload that into a cloud folder that I
• 9:46
  attach to cloud code.
• 9:48
  And then I have a workflow
• 9:49
  in which it reviews the call.
• 9:52
  Then it goes to the brand's website,
• 9:54
  scripts products, scripts
• 9:55
  product descriptions, and,
• 9:58
  you know, fonts, colors, et cetera.
• 9:59
  It'll generate 14 spec ads sort of based
• 10:03
  on what we talked about in the call.
• 10:05
  Additionally, what it does is like does a
• 10:08
  bit of a very quick
• 10:09
  gap analysis, like what
• 10:10
  are your competitors
• 10:11
  are doing that you're not.
• 10:12
  And that way, like I have one candidly
• 10:15
  going on in the
• 10:16
  background right now from a sales
• 10:18
  call that we went on this morning.
• 10:19
  The full process probably takes five to
• 10:21
  seven hours, like
• 10:22
  doing sort of end to end.
• 10:24
  But what ends up happening is I get this
• 10:26
  beautiful looking proposal
• 10:28
  with 14 spec ads ready to
• 10:31
  go that I can deliver to them same day
• 10:33
  that has like everything
• 10:35
  that we discussed in the
• 10:35
  call embedded within the proposal.
• 10:38
  Insane.
• 10:39
  It's insane.
• 10:40
  So good.
• 10:41
  It's insane.
• 10:42
  And like we are now, I genuinely believe
• 10:45
  this, you know, just
• 10:47
  based on the way our proposals
• 10:48
  look, and I'm happy to sort of share this
• 10:49
  with, you know,
• 10:50
  believe this in the notes or
• 10:51
  whatever.
• 10:52
  Like I genuinely believe just based on
• 10:54
  how our proposals look,
• 10:55
  like we are closing enterprise
• 10:57
  deals at a higher clip because it just
• 10:59
  feels like this is, these
• 11:01
  people really care about
• 11:02
  polish.
• 11:04
  And yeah.
• 11:05
  And the other cool thing about this
• 11:07
  system is if you have the
• 11:09
  brand send over like brand
• 11:10
  guidelines, et cetera, you ingest that
• 11:12
  into your cloud code as well.
• 11:15
  And so like it's you effectively are
• 11:17
  creating an agent to QC
• 11:18
  against brand guidelines.
• 11:19
  So when you deliver those assets for a
• 11:21
  brand that really cares,
• 11:23
  everything's pixel perfect,
• 11:24
  you're going to sort of hit
• 11:25
  the mark right off the bat.
• 11:26
  I mean, it's so it's so great.
• 11:27
  Like it's it's such an easy thing to do.
• 11:29
  But what I was really asking was like,
• 11:31
  you guys make ads for
• 11:32
  people utilizing AI, right?
• 11:35
  But you also wrap in customer personas.
• 11:39
  And like, that's the that's a big, deep
• 11:40
  part of what you do, right?
• 11:42
  Yeah, of course, the ad creation is like,
• 11:45
  it's the last piece of the puzzle, right?
• 11:48
  Like, I think I talk about
• 11:50
  this on most of my sales calls.
• 11:51
  So and we may have talked about this in
• 11:53
  the podcast before,
• 11:53
  so I'll keep it brief.
• 11:54
  But I think there's like sort of three
• 11:56
  pieces of the puzzle
• 11:58
  to generate great ads.
• 12:00
  You know, the first piece of
• 12:00
  the puzzle is a gap analysis.
• 12:02
  You know, I've built a free tool
• 12:03
  adlib.skipper.ai, adlib.getskipper.ai,
• 12:08
  where you can put in your brand,
• 12:09
  it will show you like, these are the
• 12:11
  personas, formats, angles and emotions
• 12:13
  that you're not focused on
• 12:14
  that you should be focusing on.
• 12:15
  And I'll give you 10
• 12:16
  recommendations on ads to build.
• 12:17
  That's sort of piece one.
• 12:19
  So it's like, you are
• 12:21
  focusing on these personas.
• 12:22
  And like, especially for folks that are
• 12:24
  struggling with CAC.
• 12:26
  And it's like, you know, we can't get new
• 12:27
  customers or CAC keeps flying.
• 12:29
  But the issue is you probably keep
• 12:30
  targeting the same
• 12:31
  people again and again.
• 12:32
  So you've trained your pixel to just like
• 12:33
  look for the same type of person, right?
• 12:35
  Like you need to be training people.
• 12:37
  You need to be training Meta's pixel to
• 12:40
  be looking for new audiences.
• 12:42
  So sort of the gap analysis is one.
• 12:44
  And I'll add to that in a moment.
• 12:47
  The second piece of the
• 12:48
  puzzle is time series analysis.
• 12:51
  I think the antiquated way of thinking
• 12:52
  about things is, oh, you
• 12:54
  know, we had that great ad
• 12:55
  in January.
• 12:57
  Like, why didn't we ever
• 12:57
  do anything more like that?
• 12:59
  Why didn't we ever hit
• 12:59
  up that creator again?
• 13:01
  Or like, oh, we had that, you know,
• 13:02
  static that did really well.
• 13:03
  Like, why did we turn that off?
• 13:04
  Like, this is how
• 13:05
  agencies and brands talk.
• 13:07
  And it's just like, it
• 13:08
  feels like the stone ages.
• 13:10
  Because like what you can do at this
• 13:11
  point is using the Meta
• 13:14
  API, I would say use the
• 13:15
  Meta API.
• 13:16
  Don't use MCP.
• 13:17
  The MCP sort of breaks.
• 13:18
  Even the Meta API sort of sucks, but it's
• 13:20
  better than the MCP.
• 13:22
  What you can do is you can pull all of
• 13:23
  your ads that you've ever run.
• 13:25
  You can use Cloud Sonnet 4.6.
• 13:29
  And you can tag every ad you've ever
• 13:30
  created with persona,
• 13:32
  angle, format, emotion, and ad
• 13:34
  type.
• 13:34
  And then you can
• 13:35
  correlate that to performance.
• 13:37
  And so you can say like, okay, for the
• 13:39
  past five years, like
• 13:40
  these combinations have been
• 13:41
  my winning formulas.
• 13:43
  And like, here are
• 13:44
  the ads to back that up.
• 13:46
  So if you're creating a system where you
• 13:49
  are never losing
• 13:50
  learnings, like every learning
• 13:52
  is compounding on one another.
• 13:54
  And that's like also why like creative
• 13:55
  velocity is so important, right?
• 13:57
  Like the more ads you're
• 13:58
  putting into the system.
• 13:59
  And like, I know this is
• 14:00
  like a controversial take.
• 14:01
  I'm of the opinion, like for every $1,000
• 14:03
  you're spending, you
• 14:04
  need to be testing a net
• 14:04
  new creative.
• 14:05
  And I don't mean an iteration.
• 14:07
  I don't mean a new hook.
• 14:08
  I mean like a net new creative, right?
• 14:10
  Even if it doesn't get enough spin,
• 14:11
  you'll still get signals
• 14:12
  that will go into that time
• 14:13
  series analysis that will tell you, you
• 14:15
  know, this persona,
• 14:16
  angle, emotion had a great CTR.
• 14:17
  So like maybe there's a
• 14:18
  CRO optimization you can do.
• 14:20
  So anyway, so we have the gap analysis,
• 14:22
  the time series analysis,
• 14:23
  and then the third piece
• 14:23
  of the puzzle, and this is what we're
• 14:25
  building with
• 14:25
  Skipper, is social listening.
• 14:27
  You know, looking at what folks are
• 14:29
  saying in your reviews, what
• 14:31
  folks are saying on Reddit,
• 14:32
  on Trustpilot, on Quora, on Amazon, and
• 14:35
  saying, okay, these
• 14:36
  are like my core eight to
• 14:37
  12 personas.
• 14:38
  This is what they care
• 14:39
  about the most with a product.
• 14:42
  Like these are the angles
• 14:42
  that they care about the most.
• 14:44
  And then what you do is with the social
• 14:48
  listening, the time
• 14:49
  series, the gap analysis, you put
• 14:51
  that together and you have this really
• 14:52
  nice Venn diagram of
• 14:54
  like, here's what the gap
• 14:55
  analysis is telling me to do.
• 14:56
  Here's what my time series
• 14:57
  analysis is telling me to do.
• 14:58
  Here's what people talking about my brand
• 15:00
  are telling me to do.
• 15:01
  Where's the overlap?
• 15:02
  Okay, great.
• 15:03
  Here's the overlap.
• 15:04
  Here are the ads we should be creating.
• 15:05
  And so like, when we now have an AI, we
• 15:09
  call it AI creative only
• 15:10
  team, but there are eight
• 15:13
  people on it, right?
• 15:14
  Like the whole idea is like, there's
• 15:15
  humans in the loop every step of the way.
• 15:17
  Like we are not at the point, and I
• 15:19
  cannot reiterate this enough
• 15:20
  because I get this question
• 15:21
  almost every day.
• 15:22
  We are not at the point yet where you can
• 15:25
  get 10 out of 10
• 15:26
  perfect ads using AI alone.
• 15:29
  You may get really good ones.
• 15:30
  Maybe nine out of 10
• 15:31
  are really, really good.
• 15:33
  But maybe a few won't have the safe zone
• 15:35
  or whatever it is, right?
• 15:36
  So when we have the center
• 15:38
  of that Venn diagram, right?
• 15:40
  We can give that to a
• 15:41
  creative strategist.
• 15:42
  The model will say, hey, you know, this
• 15:44
  is what I'd recommend
• 15:45
  putting together based on
• 15:47
  those three things outlined.
• 15:48
  And the creative strategist will say,
• 15:50
  great, give me 20 ads, 20 pieces of copy.
• 15:54
  It'll spit out the copy.
• 15:55
  Creative strategist will review the copy.
• 15:57
  Great.
• 15:58
  If they like it, great.
• 15:59
  If not, you know, not so good.
• 16:01
  Again, like I said, 4.7, we were
• 16:03
  rewriting a lot of the copy.
• 16:05
  I just did this wire cutter
• 16:07
  ads for this client today.
• 16:10
  And the copy is really good.
• 16:11
  And so I think Opus 4.8 really, you know,
• 16:13
  helped out with that.
• 16:14
  Granted it took like three, four hours.
• 16:15
  And then the ads are generated.
• 16:19
  We're just using an NFL.ai API.
• 16:22
  So it lands directly in
• 16:23
  our Cloud Code folder.
• 16:24
  Of course, you can
• 16:25
  use Higgs field as well.
• 16:27
  And then I think where I was being really
• 16:30
  too stubborn for too
• 16:31
  long was I was saying like,
• 16:33
  we don't need to necessarily have an
• 16:35
  editor touch these up.
• 16:36
  But as we have been now selling this
• 16:40
  workflow to enterprise, that's just like
• 16:43
  not, you know, you
• 16:45
  sort of need to have that.
• 16:46
  So we...
• 16:46
  Looking at it, yeah.
• 16:47
  Yeah.
• 16:49
  So we have, you know, three people in the
• 16:52
  U.S. that are generating statics with AI.
• 16:54
  And then they go to our editing team.
• 16:57
  We have a 24-hour SLA.
• 17:00
  And what the editing team effectively is,
• 17:01
  they're just like brand guideline Nazis.
• 17:04
  They're just ensuring that,
• 17:06
  you know, the fonts are perfect.
• 17:08
  The colors are perfect.
• 17:09
  The product looks right.
• 17:10
  Everything's in the safe zone.
• 17:12
  And that's their job.
• 17:13
  So, all right, it's just like a QC team.
• 17:16
  Because when we're at the scale now that
• 17:18
  we're generating so many
• 17:20
  creatives, it's hard to be like,
• 17:21
  is that, you know, is that going to get
• 17:23
  cut off by Meta's UI?
• 17:24
  Like we just like needed a
• 17:25
  dedicated team to do that.
• 17:27
  And AI will get you very, very close.
• 17:29
  I would say of all the AI
• 17:32
  statics we're creating, 90...
• 17:36
  We're getting to 90 to
• 17:38
  95% of the way there.
• 17:39
  It's just like those like
• 17:40
  sort of final touches, right?
• 17:42
  And especially for the brands that really
• 17:44
  care about pixel perfection.
• 17:45
  Like if the logo's size is like two
• 17:48
  pixels off, they're going
• 17:49
  to be like, that's wrong.
• 17:50
  So that's where having an editor come
• 17:52
  into play really makes a big
• 17:54
  difference at the end game.
• 17:56
  Yeah.
• 17:57
  So this is how you've built what you're
• 18:00
  doing and selling now.
• 18:03
  And hopefully those of you that are
• 18:04
  listening can take pieces of that and
• 18:06
  maybe even think about
• 18:06
  it in your own agency.
• 18:08
  So like, how are you as a creative agency
• 18:15
  utilizing AI within
• 18:18
  your creative systems?
• 18:20
  And also like, what is
• 18:21
  your creative system?
• 18:23
  Because honestly, you
• 18:24
  talked through some of it, right?
• 18:25
  You're going through the process of
• 18:26
  that's for each client.
• 18:28
  But how are you utilizing AI in the
• 18:30
  business to make more money
• 18:33
  or have more time or, you know,
• 18:35
  be really like be making
• 18:36
  great stuff for these folks?
• 18:38
  Yeah.
• 18:39
  Again, it's a good question.
• 18:41
  Some part of me, I feel like since, I
• 18:45
  don't know what we're calling it.
• 18:46
  We call it the industrial revolution.
• 18:48
  Are we calling this AI revolution?
• 18:49
  I don't know.
• 18:51
  Some part of it, I think I'm working more
• 18:52
  hours than I ever have in my
• 18:54
  life right now, I would say.
• 18:56
  And I think from folks that are sort of
• 18:59
  doing similar things, like
• 19:00
  I have one client that is,
• 19:02
  he says he works 18
• 19:03
  hours a day in Cloud Code.
• 19:04
  He's just like addicted to it, right?
• 19:06
  And there is, I feel like a bit of a
• 19:09
  paradox where you feel like you're
• 19:10
  putting in more work,
• 19:11
  but you're getting less out of it.
• 19:13
  So you really need to sort of figure out
• 19:14
  which swim lanes make sense.
• 19:16
  I would say the biggest ROI I've found,
• 19:22
  one, the proposal game
• 19:23
  has never been better.
• 19:25
  Two, CRO.
• 19:26
  We never did landing pages.
• 19:28
  You know, we never did landing pages.
• 19:30
  Landing pages took us
• 19:31
  like three, four weeks.
• 19:32
  They were such a pain in
• 19:33
  the ass to put together.
• 19:34
  If you had edits, like it would take a
• 19:37
  designer, you know, four or
• 19:38
  five days to get edits back.
• 19:39
  Like now with like Vercel, Cloud Code,
• 19:42
  GitHub, like spinning up landing pages
• 19:44
  has never been easier.
• 19:45
  So in our workflow, like again, if we see
• 19:48
  an angle pop off for a specific brand,
• 19:51
  we just spin up a landing page within the
• 19:52
  next 24 to 48 hours and start driving
• 19:54
  traffic to that instead.
• 19:56
  Right?
• 19:56
  Like it's work, but like the more
• 19:58
  one-to-one you can go
• 19:59
  from ads to landing pages,
• 20:01
  obviously the better.
• 20:02
  We've talked about this ad
• 20:03
  nauseum on the Cody podcast.
• 20:05
  And I don't want to, you
• 20:05
  know, beat a dead horse here.
• 20:07
  Like from an operational perspective, I
• 20:09
  would say at first I was like,
• 20:11
  we probably don't need to
• 20:12
  have a project manager anymore.
• 20:13
  I was wrong.
• 20:14
  I think having a project manager is
• 20:15
  actually exceedingly helpful.
• 20:18
  I like definitely had a bit of whiplash
• 20:19
  where I was like, I can just sort of
• 20:20
  automate everything.
• 20:21
  And I think everyone sort
• 20:22
  of goes through that period.
• 20:24
  And then you sort of come back to reality
• 20:26
  and you realize like the solution is
• 20:27
  somewhere in the middle.
• 20:28
  So it's AI plus humans in the loop, which
• 20:31
  I've always been a proponent of.
• 20:33
  And especially with generating ads, I
• 20:35
  will say like from an ad perspective and
• 20:38
  like an analytics perspective,
• 20:39
  that is undoubtedly changed forever until
• 20:45
  token prices like
• 20:46
  increase astronomically.
• 20:48
  But like the ability to be able to track
• 20:50
  how certain personas or how certain
• 20:52
  angles are performing over time is crazy.
• 20:55
  It's so cool.
• 20:56
  Like we just have a graph where I can
• 20:57
  just like pull up a client.
• 20:58
  It's like, you know, how is, you know, a
• 21:00
  mom of 24 to 35, how has she been
• 21:04
  performing, you know, recently?
• 21:05
  Maybe she like is an urban mom, right?
• 21:08
  Like how has that been performing, right?
• 21:11
  So effectively, we've created a system
• 21:13
  where when our creative strategists come
• 21:15
  sign on for the day to
• 21:16
  start making their ads,
• 21:17
  they already have a swim lane.
• 21:19
  They know like I need to be generating
• 21:22
  ads for this persona,
• 21:24
  this angle, this emotion.
• 21:25
  The question is just like, what's the
• 21:27
  best copy for it and what's the best
• 21:28
  format to do that in?
• 21:30
  The AI can guide you, but, and I've said
• 21:32
  this to my team, again, at Nage, I'm
• 21:33
  like, the differentiator is taste.
• 21:36
  Like humans have really good taste.
• 21:38
  AI doesn't have any taste.
• 21:39
  There's something soulless about a lot of
• 21:43
  AI ads that I found.
• 21:44
  And that's why having people in the loop
• 21:47
  throughout the process is essential.
• 21:50
  I think it's, you know, one is I think
• 21:51
  from the workflow standpoint, and some of
• 21:53
  you listening to this may
• 21:54
  really identify with it,
• 21:55
  which is I think you were doing a certain
• 21:57
  amount of work because I feel like I'm
• 21:58
  working more as well.
• 21:59
  And then you sort of see what's possible.
• 22:03
  And then you were able to take on these
• 22:04
  things that you never thought you'd be
• 22:06
  able to tackle, right?
• 22:07
  Like, I mean, and that can mean anything.
• 22:10
  It can mean your creative system based on
• 22:11
  hooks like you're talking about.
• 22:13
  It can be creative cohort analysis.
• 22:15
  It can be, you know, what ads have driven
• 22:17
  the most net new customers.
• 22:18
  It could be, I mean,
• 22:19
  it can be any of that.
• 22:21
  And the volume of work, you know, I feel
• 22:23
  like I'm doing the volume of
• 22:25
  work of like three people now
• 22:27
  because I'm able to do all this stuff
• 22:29
  that just was unaccessible before.
• 22:31
  It just was like, this would take me
• 22:33
  forever to go through this.
• 22:34
  And now it's like, boom, boom, boom.
• 22:36
  We got to do this, got to
• 22:37
  do this, got to do this.
• 22:38
  You know, like I built like four landing
• 22:40
  pages for our own shit like last week.
• 22:44
  And now I've got to tell people about it.
• 22:46
  I got to promote about it.
• 22:47
  And then I'm using AI to
• 22:48
  write, you know, this and that.
• 22:49
  So, you know, I think it's big.
• 22:51
  And we're actually working on a statement
• 22:53
  about AI now in our
• 22:54
  organization about like,
• 22:56
  here's what we believe.
• 22:57
  Like we need to be transparent.
• 22:59
  AI augments our work, you
• 23:01
  know, et cetera, et cetera.
• 23:02
  It needs to be
• 23:02
  transparent when we're using it.
• 23:03
  Because I think we're all sort of in this
• 23:05
  like rush, gold rush moment.
• 23:07
  But it's also, we're going to need to
• 23:09
  come up with some standards
• 23:09
  in the industry of what it
• 23:12
  really means, the changes.
• 23:13
  I think from a creative strategy, or
• 23:15
  sorry, from a
• 23:15
  creative systems standpoint,
• 23:18
  you know, the fact that you're able to do
• 23:20
  deep analysis is massive.
• 23:21
  And if you are an organization that's
• 23:23
  making creative, you know,
• 23:24
  I think one of the biggest things that
• 23:26
  you want to always be looking at is
• 23:28
  not thinking about volume.
• 23:31
  It's not just about
• 23:32
  making a bunch of ads.
• 23:33
  And I think we know this.
• 23:34
  But it's about making
• 23:35
  enough of the right ads.
• 23:37
  And then figuring out what our
• 23:39
  expectation is as it
• 23:41
  relates to testing those.
• 23:42
  And it's also not every ad that you have
• 23:46
  in your, in the community,
• 23:48
  like in the ad account, right?
• 23:50
  It can't be all AI.
• 23:52
  It can't be all founder videos.
• 23:55
  It can't, you know, you know that to be
• 23:57
  successful, we need a mix.
• 23:59
  And this is what that creative like sort
• 24:01
  of analysis and systems can be built.
• 24:04
  I think that what this is allowing, or
• 24:05
  like can do for us, I think that what
• 24:07
  it's allowing us to do
• 24:08
  is build better ads over time and build
• 24:15
  better like way to categorize and
• 24:18
  understand where opportunities lie.
• 24:20
  And it can help you scale faster.
• 24:22
  If, you know, and that's, and I do think
• 24:24
  that to your point, a
• 24:26
  person running project management
• 24:28
  or CLO type of thing or type of role is
• 24:31
  absolutely the most critical as it
• 24:33
  relates to this right now.
• 24:35
  Because what I see in organizations and
• 24:37
  that the places that I'm
• 24:38
  coaching and stuff is like,
• 24:39
  people are doing this stuff, which is
• 24:41
  cool, but it's all running around like
• 24:43
  with their heads cut off.
• 24:45
  And, and you can have a tool like HQ,
• 24:47
  which is cool and puts
• 24:49
  all these projects together
• 24:50
  and you can see what they built or what
• 24:52
  you built and you can utilize those
• 24:54
  projects together, etc.
• 24:55
  But that is not going to solve for the
• 24:58
  why in each of this, right?
• 25:00
  Or in all this, like it's not going to
• 25:02
  solve for like, what's the incremental
• 25:04
  progress that we are making?
• 25:07
  And, and also the
• 25:09
  incremental revenue that we are making.
• 25:11
  Yes.
• 25:11
  And so I think you like, it has to be
• 25:13
  looked through that lens.
• 25:15
  And I understand that there's an
• 25:16
  excitement and like all this, but right
• 25:18
  now it's very much a
• 25:20
  noise perspective, you know?
• 25:22
  And it's not necessarily like moving the
• 25:25
  needle as much as it possibly could.
• 25:28
  I was talking to somebody the other day
• 25:29
  and I said, it's sort
• 25:30
  of akin to like, I mean,
• 25:32
  a little microcosm of like when like
• 25:34
  online messaging became
• 25:35
  available in the works.
• 25:37
  Like, you know, like you could tell this
• 25:40
  person you're going to be five minutes
• 25:41
  late or like, you know,
• 25:42
  all these little things.
• 25:43
  But then it like brought it
• 25:45
  to a whole different place.
• 25:46
  And then you're like,
• 25:47
  wait, where was that?
• 25:48
  They told me this over here.
• 25:49
  And so it's like, you have to make sure
• 25:51
  that you are understanding what the true
• 25:54
  incremental impact is in the business.
• 25:56
  And like over the incremental impact that
• 25:58
  you want to make and that all of these
• 26:00
  things are aligning towards that goal.
• 26:02
  And like, that's what we're trying to do.
• 26:04
  And also, I mean, we're at the point of
• 26:05
  it too, where having free thought is
• 26:07
  useful and messing
• 26:08
  around with it is useful.
• 26:09
  And that's good too.
• 26:10
  And some of that can be, we want to be a
• 26:13
  leader in AI as it relates to our agency.
• 26:15
  And by doing that, or in order to achieve
• 26:18
  that goal, we have to just be messing
• 26:19
  around a whole bunch.
• 26:21
  Right?
• 26:22
  Like, how do you enable your employees to
• 26:24
  sort of think through that will?
• 26:25
  Like, are you like, look, we want to be
• 26:27
  putting out 20 new proposals, net new.
• 26:30
  How are we achieving that?
• 26:32
  And also, are you taking like every
• 26:33
  employee takes an hour and a
• 26:34
  half every day to fuck around?
• 26:36
  Like, how do you sort
• 26:37
  of think about this?
• 26:38
  Yeah, it's a really tough one.
• 26:41
  And it's a hard solve because I think
• 26:44
  anyone that's doing this
• 26:45
  will immediately say like,
• 26:47
  well, I can just do it myself.
• 26:48
  Like, it'll be much faster.
• 26:50
  Like, I've built these
• 26:50
  systems around myself.
• 26:52
  Like, I know how my brain works.
• 26:54
  And that's where I was for a while.
• 26:56
  Like, I was the one, you know, generating
• 26:59
  all of the static ads for our
• 27:02
  clients for a bit because like,
• 27:03
  I'm neurotic and I can be a bit of a
• 27:05
  control freak when it
• 27:06
  comes to our standards.
• 27:07
  And then I just like found someone that I
• 27:10
  was able to sort of just like download
• 27:12
  all of my brain onto.
• 27:15
  And then she just like took that and like
• 27:17
  sort of going in her own way.
• 27:18
  And so like now what I've started to do,
• 27:20
  and, you know, we just brought in another
• 27:22
  person to sort of help
• 27:23
  generate statics as well.
• 27:26
  Is I sort of say like, hey, here's my
• 27:28
  framework of how I go
• 27:29
  about generating ads.
• 27:31
  This is just like one
• 27:31
  example, you know, ads as an example.
• 27:33
  Here's like the step-by-step process.
• 27:35
  Like, here are all the
• 27:36
  markdown files I utilize.
• 27:38
  We're going to send you a website that
• 27:41
  sort of goes through our whole process.
• 27:42
  But like, I don't want
• 27:44
  you to follow this exactly.
• 27:45
  I need people at the end of the day that
• 27:47
  are going to, this is like the most
• 27:49
  cliche line in the world,
• 27:50
  but sort of like think outside the box,
• 27:51
  color outside the lines, right?
• 27:53
  I think I read something the other day,
• 27:55
  like, I can't remember who it was.
• 27:56
  Maybe it was like Marc Andreessen was to
• 27:58
  like people that are going to be most
• 28:00
  successful in AI are just the people that
• 28:02
  are just like super unorganized and just
• 28:05
  like want to be testing a lot of
• 28:06
  different things, right?
• 28:06
  Like, and they're just like
• 28:07
  Renaissance people, right?
• 28:09
  Jack of all trades,
• 28:10
  maybe even a master of none.
• 28:11
  And so like, I think giving people,
• 28:15
  leading the horse to water in a sense,
• 28:17
  but then letting people have that like
• 28:19
  aha moment themselves, for me at least
• 28:22
  has been the sort of the critical unlock.
• 28:26
  Because before we were just like saying
• 28:28
  like, run these steps in a
• 28:29
  row and you'll get good ads.
• 28:30
  And people were doing that and they'd get
• 28:31
  good ads and they didn't really
• 28:32
  understand why they
• 28:33
  were getting good ads.
• 28:34
  Like, I don't necessarily need someone to
• 28:36
  just like run an energetic flow.
• 28:38
  Like, I can have AI
• 28:38
  run an energetic flow.
• 28:40
  But if I have someone that is building
• 28:42
  their own flow based on our foundation
• 28:44
  and comes up with like, oh my God, we
• 28:47
  should create this agent, right?
• 28:48
  This would fit really well after agent
• 28:50
  three and five or whatever it is.
• 28:52
  Like, that is what I want.
• 28:54
  And, you know, people talk about critical
• 28:55
  thinking all the time.
• 28:56
  Like, there are really
• 28:58
  two ends of the barbell.
• 28:59
  Like, and I'm sure everyone has seen it.
• 29:00
  There are people that are like respond in
• 29:02
  Slack with ChatGPT or respond in emails
• 29:04
  with Claude, which is fine.
• 29:06
  And then there are copywriters who are
• 29:10
  writing their copy with Claude or
• 29:11
  ChatGPT, which is fine.
• 29:14
  And those are the people that are on one
• 29:15
  end of the dumbbell.
• 29:17
  You know, they're sort of effectively
• 29:18
  replacing their work with
• 29:19
  AI, which is like useless.
• 29:21
  Because like, who wants to be in the
• 29:22
  middle of the bell curve?
• 29:23
  Then the other people are the people that
• 29:25
  are like actually critically thinking and
• 29:26
  pushing back every second on
• 29:28
  what the AI is giving them.
• 29:29
  And saying like, this
• 29:30
  is how you do it better.
• 29:31
  Like effectively
• 29:32
  training a junior employee.
• 29:33
  You know, this is how
• 29:34
  you write better copy.
• 29:35
  This is how, you know,
• 29:36
  you generate better ads.
• 29:37
  This is how you think
• 29:38
  about these things better.
• 29:40
  So, yeah, I think, you know, sort of like
• 29:42
  the token maxing era
• 29:43
  is like slowly dying.
• 29:44
  And now it's like, how can we actually
• 29:46
  utilize these things to create ROI?
• 29:48
  Like, we've had our fun.
• 29:50
  We've been in the playground.
• 29:51
  We've fucked around.
• 29:52
  We've made a lot of cool things.
• 29:53
  But now exactly to your point, Andrew,
• 29:55
  like, where's the ROI?
• 29:57
  And the ROI for us has been being able to
• 29:59
  build landing pages for CRO, being able
• 30:01
  to bulk create a lot of static ads with
• 30:04
  editors and strategists in the loop, and
• 30:06
  generating really,
• 30:07
  really strong proposals.
• 30:09
  And those have been the three pieces that
• 30:12
  I've utilized for AI that have really,
• 30:15
  really leveled up our organization.
• 30:17
  Everything else is pretty much the exact
• 30:18
  same as it was before.
• 30:19
  Yeah.
• 30:20
  I mean, I think, you know, this makes me
• 30:22
  think about, you know, there's going to
• 30:23
  be these phases of it
• 30:25
  that will impact all of it.
• 30:26
  If you look across our businesses or you
• 30:28
  look across D2C businesses that we work
• 30:30
  on, you know, I think there's going to
• 30:31
  be, we're going to have to establish a
• 30:34
  bunch of different
• 30:35
  buckets as it relates to AI.
• 30:37
  One is there's a governance bucket.
• 30:39
  And the governance bucket is putting out
• 30:41
  a thing of what do we believe?
• 30:42
  How are we securing this data within our
• 30:45
  brand or our agency?
• 30:46
  And how are we, you know, sort of
• 30:50
  utilizing AI in our organization?
• 30:54
  And this is what it
• 30:55
  means for us to utilize it.
• 30:58
  Hey, employees, this is what it is.
• 31:00
  Okay.
• 31:00
  And who is going to
• 31:02
  be the captain of this?
• 31:03
  Right.
• 31:03
  There needs to be a person or a group of
• 31:05
  people that are a counsel in an agency or
• 31:08
  whatever, you know, that sort of says,
• 31:10
  because this stuff's
• 31:11
  going to keep evolving.
• 31:12
  So it can't just be on the founder.
• 31:14
  And it can't, it
• 31:15
  shouldn't just be on the founder.
• 31:17
  Right.
• 31:17
  Like it has to be on a group of employees
• 31:19
  that care and contractors.
• 31:22
  I mean, for what, you know what I mean?
• 31:24
  That's like a governance thing.
• 31:25
  Like, this is what we do.
• 31:26
  This is where we stand and how we use it.
• 31:28
  And then this is, you know, kind of the
• 31:31
  way that we think about it.
• 31:32
  And that, then the next piece of it is
• 31:34
  revenue enablement, like we were just
• 31:38
  talking about, which is what are the
• 31:40
  challenges we face in business?
• 31:42
  And whether that's unlocking new personas
• 31:44
  for scale in any D2C business, unlocking
• 31:46
  new landers for scale.
• 31:48
  I mean, you could, you know, go specific,
• 31:49
  like I'll keep it high level.
• 31:51
  Unlocking and
• 31:52
  understanding unit economics.
• 31:55
  Great thing that AI can help you with.
• 31:56
  Understanding and going through, you
• 32:00
  know, inventory planning.
• 32:02
  As inventory relates to how much money
• 32:05
  you have in the bank any given day that
• 32:07
  you could spend on ads.
• 32:08
  I mean, and if you're a subscription
• 32:10
  brand, there's even more of this.
• 32:11
  Okay.
• 32:12
  So like, it's sort of what is the revenue
• 32:14
  enablement that we can utilize.
• 32:16
  Right.
• 32:17
  Right.
• 32:17
  And it's, we're going to go from, from
• 32:19
  piecemeal project based stuff
• 32:20
  of like, add this, add this.
• 32:22
  Oh yeah.
• 32:22
  And then I made this
• 32:22
  landing page or whatever.
• 32:23
  And it's going to be more like, it's
• 32:25
  going to have to go under an umbrella of,
• 32:28
  is this making us more money?
• 32:30
  And then the third bucket or a third
• 32:32
  bucket on this is employee enablement.
• 32:35
  So what are the things that you are
• 32:37
  already doing that you could replace?
• 32:39
  That's one.
• 32:40
  And that's where a lot of people sit now.
• 32:41
  The next one is, what are things you've
• 32:43
  always wanted to do that you believe
• 32:45
  would have an impact on the second bucket
• 32:48
  or on your own time that would give you
• 32:51
  time back that will enable you to be
• 32:54
  better at your job and have you as one
• 32:57
  employee be three employees.
• 32:58
  Right.
• 32:59
  And if, and under that
• 33:01
  bucket is incentivization.
• 33:02
  How are you incentivized to do this?
• 33:04
  If you do better with this and you, and
• 33:07
  you save four hours of time every day,
• 33:09
  that's great for you.
• 33:11
  And you save the company 20%, you should
• 33:13
  get a 10% bonus or whatever.
• 33:15
  Right.
• 33:15
  Like, how are you incentivizing that
• 33:17
  underemployee AI engagement and whatever.
• 33:20
  And I think those are kind of like some
• 33:22
  of the main pieces of
• 33:25
  how it will do that.
• 33:27
  I mean, I would say the fourth bucket is
• 33:28
  probably like a partner bucket of like,
• 33:29
  how are you communicating and working
• 33:31
  with partners in your,
• 33:32
  is you in your agency?
• 33:33
  Or in your e-com business?
• 33:36
  Right.
• 33:36
  Like, what are the things that you know
• 33:38
  from previous conversations that you
• 33:39
  don't have to repeat yourself?
• 33:41
  What do you know from previous things
• 33:42
  that, you know, are in Slack that you can
• 33:43
  say, like, I know I talked to Julie about
• 33:47
  this three years ago or three months ago.
• 33:50
  What should you say about pricing per
• 33:52
  pound on tea then or whatever?
• 33:53
  You know what I mean?
• 33:54
  It's like, and like bringing that
• 33:55
  together and how can you make partner
• 33:57
  relationships 20, sorry,
• 33:59
  2x, like 200% more effective.
• 34:02
  Um, because that is going to make your
• 34:04
  life easier as well and will
• 34:06
  allow you to scale a lot faster.
• 34:08
  Um, and I think, you know, it's like you
• 34:12
  can talk about media buying, of course,
• 34:13
  which all of us are
• 34:14
  media buyers ultimately.
• 34:15
  And like, I mean, that
• 34:16
  goes under bucket two, right?
• 34:17
  Which is like, how do we
• 34:18
  make more money from this baby?
• 34:20
  Um, as well.
• 34:21
  So anyway, that's kind of like, I've been
• 34:22
  thinking about how we go through this and
• 34:24
  what those buckets look like.
• 34:25
  I don't know what you think, but that's,
• 34:27
  that's been my take.
• 34:27
  Yeah, I think that's a really sound take.
• 34:29
  Um, I think it's a
• 34:30
  really, really sound take.
• 34:32
  It's, uh, uh, which company was it?
• 34:35
  Maybe Uber release, uh, you know, this is
• 34:38
  how much we're spending on tokens.
• 34:39
  And like, this has been our ROI and it's
• 34:41
  like, you know, 80% versus 20%.
• 34:44
  Uh, so I think, you know, at the end of
• 34:47
  the day, it really comes back to like,
• 34:48
  what is revenue generating for you?
• 34:50
  And I think, and this
• 34:51
  was a reality check for me.
• 34:52
  Like, I just needed
• 34:53
  to look in the mirror.
• 34:54
  I was spending way too much time
• 34:55
  generating shit that we just didn't need,
• 34:57
  but I thought was like, cool.
• 34:59
  Uh, and I was like, I'm, you know, I
• 35:01
  should be spending this time with my kid.
• 35:03
  Like, what am I doing?
• 35:03
  Like, you know, uh, and so figure out
• 35:07
  what lanes you need to be focused on.
• 35:09
  It's going to be
• 35:09
  probably different for everyone.
• 35:10
  And just go in that direction.
• 35:13
  Like, try to correlate the projects that
• 35:16
  you've created with cloud code, codex,
• 35:18
  whatever it is, to revenue generated.
• 35:22
  And what was your costs?
• 35:23
  And like, what, how much revenue?
• 35:25
  And like, what was your margin?
• 35:26
  Like, I'm looking at the amount of status
• 35:29
  that we're creating right now.
• 35:30
  And like, we probably spent like seven or
• 35:32
  eight grand this month,
• 35:33
  just like using AI tooling.
• 35:35
  Like, that's a lot of money.
• 35:36
  And I'm like, oof, you know, that feels
• 35:38
  like a lot, you know?
• 35:39
  And, you know, we're
• 35:40
  still have the same headcount.
• 35:41
  We actually added people, but like an
• 35:43
  extra 7k a month to generate ads that we
• 35:46
  were using editors for before.
• 35:47
  Obviously, we're doing a lot more.
• 35:48
  So, you know, it's, it's sort of six, one
• 35:50
  half a dozen, but like, just like be
• 35:52
  super conscious about that stuff.
• 35:54
  Because it's so easy to get taken away,
• 35:56
  you know, just, yeah, be
• 35:57
  taken away by all of this.
• 35:59
  And be like, this is amazing, especially
• 36:00
  if you're using an API.
• 36:01
  Because like, you don't see the costs
• 36:02
  like upfront in cloud code.
• 36:04
  Or using an MCP, you
• 36:05
  don't see the costs upfront.
• 36:07
  So, yeah, I appreciate it.
• 36:09
  Thank you for listening, everyone.
• 36:11
  Hopefully, this is helpful to you of our
• 36:13
  ranting and talking about
• 36:14
  all this different stuff.
• 36:15
  And we appreciate you being here.
• 36:17
  And if you have any thoughts or anything,
• 36:19
  always email me, andrew at
• 36:20
  flexibledigital.com.
• 36:21
  And we look forward to
• 36:22
  hearing from you next time.
• 36:23
  Thank you, guys.
• 36:24
  Thank you.
• 36:24
  Thank you.
• 36:25
  I appreciate it.