Latest / 1 to 100 / E58: Building AI Employees

Transcript

E58: Building AI Employees · 1 to 100

• 0:00
  You should just get a share mic. It's like 150 right?
• 0:03
  It's something. This is what?
• 0:05
  Sure, SM-57 or something. I have a couple of them at home.
• 0:09
  I have a couple of them. It's just that I, I usually
• 0:12
  travel a lot and I do not have a setup that's mobile enough for
• 0:14
  this. So I haven't figured that out
• 0:17
  yet. It's very rare that I can
• 0:19
  actually take those because this they, they help a lot when
• 0:22
  you're recording and stuff as well.
• 0:23
  Like I do bit of music and it's I have a couple of those, it's
• 0:27
  just that I never have the chance to use them properly for
• 0:29
  this, you know, So apologies ahead of that.
• 0:33
  No, it's all good. It's all good.
• 0:35
  On the topic of music, I'm also if I'm not working, I'm playing
• 0:39
  around with music, not making music.
• 0:41
  That's a. Lie, lie.
• 0:44
  It is No, it's true. There's no way is there any He
• 0:48
  just. Became a dad.
• 0:49
  Lie, lie. Lie, lie, lie.
• 0:52
  Lie. OK, Congrats.
• 0:55
  First of all, Congrats on on that.
• 0:57
  If there's anything that we can ever listen to if if if you
• 0:59
  share them anywhere you're links or it didn't happen not at all.
• 1:04
  I. Didn't make music.
• 1:04
  No, no, no, no. I'm about to buy a Roland TD716
• 1:10
  if you guys know what I'm saying.
• 1:11
  It's like a electric drum kit as well.
• 1:13
  OK, very nice. It's hard to get into Peru, but
• 1:16
  yeah, I found, I found a official distributor in Peru and
• 1:21
  I'm pumped, so excited. But I'm not going to release any
• 1:24
  music. I'm just going to jam, you know?
• 1:27
  Yeah. That's nice.
• 1:27
  So you so you play the drums? Yeah.
• 1:29
  Play anything. Else piano as well.
• 1:31
  Yeah. OK, OK, very nice.
• 1:33
  Beautiful. Before he was a father, he
• 1:34
  played the drums. Not anymore.
• 1:38
  Drums is one that I haven't, you know, I played drums for 10
• 1:41
  years and I got pretty good, you know, over the course of ten
• 1:44
  years of lessons and teaching myself.
• 1:48
  But then you can't travel with a drum kit.
• 1:50
  I left England when I was 18 because I started my first, Oh
• 1:53
  no, not 20/18/20. So I started my first business.
• 1:56
  So from that point on, the normal drums can't travel with a
• 1:59
  drum kit, so. You always have your, you always
• 2:03
  have the air. Drums, man, come on.
• 2:05
  Be very autistic. I'm tapping on everything.
• 2:07
  If you don't know a drummer in your life, they're like
• 2:11
  incessantly tapping on things. That's yeah.
• 2:14
  Yeah, you don't have that with pianists, you know, like you
• 2:16
  don't see this around, but like with drummers, it's always like
• 2:20
  this. Is this is drummers?
• 2:21
  Drummers do this with their drummers.
• 2:23
  Drummers with the tappings. Yeah, yeah.
• 2:24
  Yeah, yeah, fortunately I'm not, you know, air drumming.
• 2:28
  That would be like, yeah full blown Aspergers but.
• 2:34
  Jack, I expect AI expect a recording of something with you
• 2:38
  and your baby in the next couple of weeks here.
• 2:41
  But This is why I want to get the kit.
• 2:43
  I want to teach him when he's, you know, old enough.
• 2:47
  I saw this video and my girlfriend sent it to me of like
• 2:50
  a 1 1/2 year old child playing like Green Day on a drum kit.
• 2:55
  And I was like, yes. This is cool.
• 2:58
  Honestly, it's probably a lot like something they can pick up
• 3:00
  a lot faster than most other instruments.
• 3:03
  Yeah, percussion is the easiest for kids, but.
• 3:07
  And then harmonica, actually, surprisingly too.
• 3:09
  Is it as I imagine? I imagine harmonica is not the
• 3:12
  easiest for the parents. At least you know they're like
• 3:15
  if you have the kid constantly playing around.
• 3:18
  Dude, I used to have like a acoustic drum kit and man,
• 3:21
  that's got to be the hardest one for neighbors as well.
• 3:24
  We have like quite a lot of land and like I had a shed solar
• 3:28
  structure with my drum kit in it.
• 3:30
  And man, every neighbor, you know, they let us know whether
• 3:32
  it was a compliment or it was like a, you know, you know, a
• 3:36
  suggestive compliment. It was a, you know, it was very
• 3:40
  loud. Some time to talk about
• 3:43
  hardware, software, right? This is something that, you
• 3:45
  know, I'm becoming increasingly exposed to.
• 3:48
  You know, everyone who has a Tesla is exposed to it.
• 3:51
  But I just bought an Owlet. I saw that they raised over like
• 3:54
  $133 million. I was like, wow, that's
• 3:57
  incredible. You know, there's a lot of this
• 4:01
  software out there that's, you know, subscription based, not
• 4:04
  subscription based. I mean, Woop bands, Google's
• 4:07
  new, you know, Fitbit bands and stuff like that.
• 4:09
  It's a whole market I've never really thought very much about,
• 4:13
  but. Seems to be Jack you got to you
• 4:14
  got to ask me for the baby recommendations.
• 4:17
  I got one. I have an outlet and another one
• 4:19
  and I like the other one better. No.
• 4:21
  Why do you guys do any wait? Why are you doing any baby
• 4:24
  monitoring? You guys are what are you doing
• 4:27
  the cameras? This is psychosis.
• 4:29
  Wait, wait, you don't, you don't have like a camera with the crib
• 4:33
  or anything like that? No, I use a $30 security camera.
• 4:37
  I mean, there's also that option.
• 4:38
  You go to the target 1 is actually probably the best one
• 4:41
  that we got. We have three different ones.
• 4:43
  Each child we got a different one, Yeah.
• 4:46
  Dude I got a $30 U fee. It's like a shitty security
• 4:50
  camera that. That's our last one and it's
• 4:52
  great it it does exactly what we need.
• 4:55
  OK. I don't call that a baby
• 4:56
  monitor. It's a security camera that
• 4:59
  happens to double as a baby. And the reason, no, no, this is,
• 5:02
  this is important, Jeremiah, because the baby industry is set
• 5:07
  up to screw you. Like everything is like 3X the
• 5:12
  price of what it should be. And it plays on like parental
• 5:17
  psychosis of fear that something will happen to their kid.
• 5:22
  And, and like, between those two things, I hate it.
• 5:25
  I hate all of it. OK.
• 5:27
  But it is a real thing like the like, Jack, did you get like the
• 5:31
  sock and stuff that out of the outlet has like they've got the
• 5:34
  camera, they've got the sock. Yeah, this is what I'm saying.
• 5:36
  I was like. What is that?
• 5:38
  What is that? It's like a baby whoop band.
• 5:41
  Yeah, baby whoop. You have you tried it?
• 5:45
  No, I can't set it up. OK, so on the topic of like baby
• 5:48
  brands want to screw you over, I think the regulation wants to
• 5:51
  screw you over. I have to buy this little dongle
• 5:54
  to spoof it into the US because if I use in any other country,
• 5:57
  they'll nerf it because of regulation, right?
• 6:00
  They'll say, hey, no, you're not allowed to use the full
• 6:02
  capabilities of the device. So yeah, I'm waiting on Amazon
• 6:06
  sending a little dongle so I can, you know, use it
• 6:09
  internationally. This is the government babying
• 6:11
  you the same way you're babying your baby.
• 6:13
  Now you know how it feels. Stop it dude.
• 6:19
  I have no problem with people disagreeing with me on this
• 6:23
  because I think everybody's. I think as a parent, everybody
• 6:24
  has a different level of risk they're open to, right?
• 6:28
  Like there's being a parent is really hard because at the end
• 6:31
  of the day, you've got like this weird competition of priorities
• 6:34
  here where you need to protect your kid and you also need to
• 6:37
  give them independence and the ability to learn and grow and
• 6:39
  become whoever they're going to be.
• 6:41
  And it's like those two things are constantly in conflict with
• 6:44
  each other. And then you have this whole
• 6:46
  aspect of like the world around them might be out to get them
• 6:49
  and you don't know who, who is part of that, who is not.
• 6:52
  It's a it's a very, it is like maybe like 1.
• 6:55
  The most complex aspect of life is, is being a parent if you end
• 6:59
  up being a parent. So the sock was one of those
• 7:02
  things. My wife and I looked at that and
• 7:03
  we were like, that is just too much information.
• 7:07
  It's. Just.
• 7:09
  Like, I don't want something to get waking me up in the middle
• 7:12
  of the night telling me that my kid is probably dying when my
• 7:14
  kid probably is not dying, right.
• 7:16
  Like, it's just like a like the odds of that are so low.
• 7:19
  So you have this constant gave up of like, what is probably
• 7:23
  going to happen? And what do you fear is going to
• 7:26
  happen? And those things are like, yeah,
• 7:29
  they're they're at odds with the story quite a bit.
• 7:33
  I think it's like distribution to everything.
• 7:34
  This for me is when he's sick so we can get the information to
• 7:38
  the doctor. I don't care about any of it.
• 7:40
  I just want the doctor to have all of the information that he
• 7:43
  needs to, but it's not. Actually medically useful
• 7:46
  information like right? Obviously if you're in the US,
• 7:48
  so I have to buy this dongle, you know, so that, yeah.
• 7:54
  Canada is going to say, Jack, I'm not allowed to look at this
• 7:58
  because it's illegally retrieved information.
• 8:01
  Right, I mean the very good finding all of the complexities
• 8:04
  yeah, we have you know, fortunately we have, you know, a
• 8:08
  good a system there. It's just the they're getting
• 8:11
  the tech to work is is tricky. So so yeah, I mean, I say it's
• 8:17
  tricky, but I just have to buy a dongle like 20 bucks on.
• 8:20
  It here's my other thing with stuff like this was like the
• 8:22
  software is always terrible. Like all of these, their
• 8:25
  software is so bad. This is where AI should come in.
• 8:29
  OK, right. This is like the the craziest
• 8:32
  thing is of all of the places where we are saying like oh AI,
• 8:37
  like I can just vibe code it myself actually hard.
• 8:43
  I mean not Tesla's, but these types of things like my UFI
• 8:47
  camera software. What instead they should do is
• 8:50
  just give me an endpoint to the effing video and say now you do
• 8:55
  whatever the hell you want and I will build my own app to watch
• 8:59
  the videos in the way that I want to and things like this.
• 9:02
  And that is an insane opportunity, which I count for
• 9:06
  the life of me understand why it's not possible.
• 9:08
  Yeah. Bear in mind, guys, my my fiance
• 9:11
  is like a she's incredible. So now that she's a full time
• 9:14
  mom, she has to have an outlet. So we have this app called
• 9:18
  Huckleberry. This the couple based in San
• 9:20
  Fran and it's the world's like the UI is so ugly, but she
• 9:24
  tracks everything. You can see.
• 9:26
  That's like every waking hours, sleeping hours, tracked sleep,
• 9:30
  you know, feeds, diapers. But does he track them?
• 9:34
  Does she track them manually? Yes.
• 9:36
  Yeah. This is a This is a first baby
• 9:38
  special. That's exactly baby number one.
• 9:42
  Haven't touched a sense. And these guys.
• 9:46
  Raised they. Raised 12 million I think for
• 9:48
  this app and they can't get the UI right.
• 9:50
  And I'm like, oh, I just want this in I just want this in
• 9:52
  chat. I want it in Claude.
• 9:54
  I want to be able to just like query it and not have to you
• 9:56
  know? You know, I maybe dictate as
• 9:59
  well. I just tap 1 button and say OK
• 10:00
  this just happened as opposed to tapping buttons that suck you
• 10:04
  know? Also, I love how you know how
• 10:05
  much everybody's raised for all these different things.
• 10:08
  Yeah, that's true. Yeah, oh man, I'm like, oh,
• 10:12
  there's an app for this. Cool.
• 10:13
  How much did they raise, you know, 12 million for
• 10:17
  Huckleberry. Yeah, I get it. 133 millionth
• 10:20
  outlet. They're crushing.
• 10:22
  I don't know what you know, what their revenue looks like, but
• 10:25
  yeah, they're a lot of people believe in them apparently.
• 10:28
  It's an insurance company, man. That's why it's so it's so
• 10:32
  financially viable. What do you mean insurance
• 10:35
  company? Why?
• 10:35
  Do you buy an outlet sock? Yeah, good point.
• 10:38
  Yeah, yeah, yeah. I.
• 10:38
  Mean this is The thing is like the the the business models are
• 10:43
  so important to truly understand because that's how you value the
• 10:48
  company. Once you understand what the
• 10:50
  business is, then you can value it, right?
• 10:52
  Otherwise there's nothing. I mean, no one's going to value
• 10:55
  a sock. Maybe right, but.
• 10:58
  Why aren't you making baby products?
• 11:00
  Why are you making Baby I? Mean I have this is like you
• 11:02
  guys got me in shock like I'm apparently I'm the only one
• 11:05
  without a baby. I'm not a parent yet.
• 11:07
  So it's I have absolutely no clue what's happening in this
• 11:10
  whole space. I haven't interacted with those
• 11:12
  software. You know, I barely have any idea
• 11:15
  about like those kinds of problems.
• 11:17
  I haven't even thought of any of these.
• 11:18
  But I feel like at the end of the school we're going to sort
• 11:20
  of start building a baby company, sort of like a baby
• 11:23
  products company, software, kind of a kind of a thing.
• 11:26
  All of us together, you know, I'm down, don't get me wrong.
• 11:28
  Like, let's just go over it, you know?
• 11:31
  It's not even just baby software.
• 11:32
  I mean, we got these cloth diapers, smart diapers or
• 11:34
  something. We got 32 came to like what is.
• 11:37
  A smart diaper? What is it?
• 11:38
  I'm sorry, it's like freaking my what is this?
• 11:41
  What is a smart diaper man? Like high tech diapers?
• 11:44
  No, it's like I'm, it's like a cloth diaper, a washable diaper.
• 11:48
  Apparently. According to my my fiance, it
• 11:51
  has no tech. There's no tech at all.
• 11:53
  But like they cost like 30 bucks a pop.
• 11:55
  OK, then why are they called smart though?
• 12:00
  Like there must be something that we're not getting 100
• 12:03
  years. Ago.
• 12:04
  Exactly. So it turns out that was smart.
• 12:09
  That was smart. Brock, how, how big is the
• 12:11
  company now? I guess maybe you should just
• 12:13
  share what it does, how big it is right now.
• 12:15
  Yeah, just do do quickly of the land.
• 12:17
  And then we, we just sort of like talk about what it's like
• 12:19
  to build a software company. So we are, we're building an AI
• 12:23
  data team. It's basically in a bunch of AI
• 12:26
  agents that have a very deep understanding of your of your
• 12:30
  company data and can answer any questions through Slack, through
• 12:34
  WhatsApp, through Discord teams or our web UI and essentially
• 12:41
  reduces the the time to insight. We're a small company where 15
• 12:44
  people distribute team across the Europe.
• 12:47
  And yeah, been been building this for 2 1/2 years.
• 12:53
  Incredible. Yeah.
• 12:54
  I was really excited to talk to you because we just spoke to Jim
• 12:57
  Benson from ADAPT and we also spoke to Jacob Postal from HQ.
• 13:02
  And these products, I'm sure they're more different than they
• 13:05
  are similar, but on face value to somebody who might be buying
• 13:09
  and can seem very, very similar. So like, I wonder, like it
• 13:13
  clearly is a good space to be building and if so many serial
• 13:16
  players are building in it, but like, what makes you guys
• 13:19
  different? What is it that you guys are
• 13:22
  trying to really solve? When I go to the website, I can
• 13:24
  get an impression of the problem that you're trying to solve.
• 13:27
  But yeah, how would you to stand out against competitors?
• 13:31
  And I should say that when I say competitors, I mean just like
• 13:34
  somebody raw dogging it and clawed or setting up their own
• 13:37
  flat pot, getting clawed connected to everything, and
• 13:40
  then having that in the channel for Jim Benson adapt.
• 13:43
  It was like they're a supercomputer or they're an AI
• 13:46
  supercomputer in the background. But yeah, I'd love to hear what
• 13:49
  you guys are. How you seeing yourself as less
• 13:52
  replaceable right? A lot of the products out there
• 13:56
  essentially focus on the idea of like, it could give me a file
• 13:59
  that I'm going to analyze this and get give it back to give the
• 14:02
  results back to you. It's like, I don't know, you
• 14:04
  throw me an Excel file APDF and you get to data analysis.
• 14:08
  That's not really how actual data analysis work when when
• 14:12
  you're operating about a certain, certain scale where
• 14:16
  your data doesn't actually fit into an Excel sheet, you know,
• 14:19
  and most of the time today it is not actually coming from a
• 14:23
  single source. It's not like just your, you
• 14:25
  just have that data and Shopify, you actually have a lot of
• 14:28
  questions that span a lot of different data sources, like
• 14:31
  order data comes from Shopify, you have data that's coming in
• 14:34
  from, I don't know, Stripe and then you have your marketing
• 14:36
  data across six different platforms.
• 14:38
  So the reality of answering data for real life business questions
• 14:45
  is a lot less about the AI part of things.
• 14:48
  It's a lot more about the underlying infrastructure.
• 14:50
  Otherwise, it's like you just hire this very confident in turn
• 14:54
  that can just throw out answers to any question that that you
• 14:56
  ask them. But the underlying work is, is,
• 15:01
  is is what makes this a lot more, a lot more successful than
• 15:05
  than just having like Claude Rodok in your, your Shopify
• 15:08
  store. So essentially we have started
• 15:11
  our life as as building like the whole data integration layer
• 15:16
  that was we were building that before we were building anything
• 15:18
  around AI or agents and and over time it became a very natural
• 15:22
  fit. So the reason that it works very
• 15:24
  well today is because we actually understand that the
• 15:28
  full journey of that data for all the way from hundreds of
• 15:31
  different sources that we connect down to all the
• 15:34
  dashboards, all the machine learning models, all the reverse
• 15:38
  ETL usage that you do where you push this data back into
• 15:41
  different platforms. You know, and when at that
• 15:44
  point, when you ask a question, AI actually has a very good
• 15:47
  understanding about all the types of data that you have
• 15:50
  available, all the metadata, all of their relationships,
• 15:52
  etcetera. This means that we're
• 15:54
  essentially talking about like hundreds, if not thousands of
• 15:56
  tables, 10s, if not hundreds of of different sources of data for
• 16:01
  any, any sizable business. And you don't, you don't get to
• 16:05
  solve that problem without a proper data infrastructure.
• 16:08
  At that point our competition becomes not, not Claude.
• 16:12
  It becomes essentially while you hire 10 people and you try to
• 16:15
  build this internally versus well here's a ready made product
• 16:19
  built for this. So.
• 16:22
  That's the. That's the argument.
• 16:24
  Are you are you using Anthropic or Open AI to power?
• 16:28
  We're using both. We're using both, yeah.
• 16:30
  How has that impacted the building?
• 16:33
  Of this over 2 1/2 years because the capabilities 2 1/2 years ago
• 16:37
  were so different than they are. So they like I what my big
• 16:40
  challenge with AI is it's like I feel like as soon as I build
• 16:43
  something on top of the infrastructure that exists, the
• 16:44
  infrastructure changes usually in a beneficial wave, but maybe
• 16:49
  not always. I'm curious how you had to deal
• 16:52
  with. It so we we struggled a lot with
• 16:55
  that initially when we were building that.
• 16:57
  So I'll take a step back around a year or so ago.
• 17:00
  Anyone that has tried this product has essentially failed
• 17:06
  because the models were not that good.
• 17:08
  We did not have as good of an understanding of the problem as
• 17:11
  we do now and it just the ROI was not really there, you know.
• 17:16
  So I think the first hurdle that we had to overcome in terms of
• 17:21
  like these models and their evolution was we have to 1st
• 17:24
  become believers that this will actually work.
• 17:27
  You know, because we had to, we had to overcome this, this
• 17:30
  feeling that like, man, we tried this a month ago, we tried this
• 17:33
  two months ago, we tried this three months ago and it doesn't
• 17:36
  work. It's not as good, you know, And
• 17:38
  then like we had to go around this and, and, and try it again
• 17:42
  and again and again and, and eventually we found that that
• 17:44
  while that works beautifully now on the on the dependency side of
• 17:50
  things, we have initially been very cloth pilled, I would say,
• 17:55
  like we, I think there was a time where they are anthropics
• 17:59
  models were a lot better than than open AI.
• 18:02
  Although over the past three months that has changed very
• 18:06
  significantly for us. We on production, we're running
• 18:08
  an ensemble. So we're running a harness that
• 18:12
  has access to both open AI S models and an entropic.
• 18:16
  Primarily we're running GPT 5.5 for, for the bulk of the work.
• 18:21
  And then we have Opus to, to help with, with some other
• 18:24
  parts. But a lot of the dependency is
• 18:28
  now on the open on the open AI side.
• 18:30
  But because we have to switch things so often and keep
• 18:33
  benchmarking things so often, we had to build a lot of layers of
• 18:37
  abstractions around the, the model providers and the, and the
• 18:40
  things that they supplied to make this as easy for us to
• 18:44
  adapt as possible. And, and right now it's, I can
• 18:48
  confidently say we're not dependent on a particular model.
• 18:51
  We are generally dependent on the, on the high level paradigm
• 18:53
  of large language models. And, and we keep experimenting
• 18:58
  on the ideas on the basis of what, what we want to build.
• 19:03
  If these things were 10 times better, 100 times better, you
• 19:06
  know, and then try to build for that and hope that the models
• 19:10
  are going to are going to catch up to that, which is sort of,
• 19:13
  it's already happening. Like things we thought that
• 19:15
  wouldn't be possible six months ago are are becoming possible
• 19:18
  already. So some of the things that we
• 19:19
  tried that that sounded very stupid at the time are becoming
• 19:22
  quite obvious and and quite reasonable to do today.
• 19:26
  There's a few things on your side.
• 19:27
  I mean, the, the, the product and platform makes sense to me.
• 19:31
  And you have some comparisons with like 5 Tran Air Byte,
• 19:34
  things like this. We actually recently turned off
• 19:37
  A5 Tran and built something on our own.
• 19:41
  Interestingly, I'm curious, do you think about your go to
• 19:44
  market from like a hey, all we need to do is go and attack 5
• 19:49
  Tran customers whose bills are getting exorbitant because like
• 19:53
  by the way, that was like me nine months ago bill.
• 19:59
  They switched their billing model and I was like, I know.
• 20:02
  This is insanity. It is insanity.
• 20:06
  The the whole space is insanity, like data platform and data
• 20:09
  infrastructure. This industry is met like it's
• 20:12
  just pure madness that has been like that for for many years,
• 20:16
  but that we're not. We have tried that.
• 20:19
  We have played with that like it worked to a certain extent, but
• 20:23
  that's not really a core part of our go to market simply because
• 20:27
  we already open sourced all of this, like all of our connectors
• 20:31
  are open source. You could today replace
• 20:33
  everything you do with five tram with our open source tooling
• 20:36
  without paying us a dime. You know, like you just, you
• 20:39
  already have all that ability. But you and I both know that's
• 20:42
  not the problem. The the reason why people don't
• 20:45
  turn off of 510 is because it's two to three months of work of
• 20:50
  like 2 of your best engineers to do the changeovers.
• 20:54
  So I guess what I'm actually asking is why not have that be
• 20:57
  your go to market is like I'm going to supply you the two
• 20:59
  engineers to do the migration you.
• 21:01
  Could do that, but I would much rather.
• 21:03
  I don't think so. On a high level, there's
• 21:06
  companies that would, that would try to cut costs and at that, at
• 21:11
  their size and scale, the cost would be so significant that
• 21:14
  they would actually care buying and paying someone else to, to,
• 21:18
  to reduce that cost. And, and on the other hand,
• 21:22
  there's, there's companies that, that have the money, that have
• 21:25
  the budget, that are more interested in the upside of, of
• 21:28
  like new value that you can bring into them.
• 21:31
  We are trying to focus on the on the latter because for the
• 21:34
  companies that we tried this with where they want to reduce
• 21:37
  the costs, essentially this becomes like a, a bidding war
• 21:41
  with regards to who can, who can do all this work with much less.
• 21:44
  If you do the servicing part of this, you essentially become
• 21:47
  like a hand like hands on consultancy business for which I
• 21:52
  don't, I don't want, I don't want to build that, you know,
• 21:54
  that's not what that's not what gets me up, up from the bad
• 21:57
  every morning. I'm not excited about that idea.
• 22:00
  So and, and when we think about like, OK, could we just put out
• 22:04
  a product that does already that?
• 22:05
  Yes. But if you, if you are not
• 22:08
  willing to migrate either you don't care about this as much as
• 22:12
  we thought we would think you would or the ROI, ROI is not
• 22:16
  there for you. And if that's not there for you,
• 22:18
  there's probably not much for us to earn from there as well.
• 22:21
  On the other side, I can actually eliminate the work of
• 22:24
  like 10/15/20 people team and help you avoid hiring another
• 22:29
  1020 over the over the course of a year.
• 22:33
  That becomes a much much better value proposition for me to to
• 22:36
  go for than than try and undercut 5 tram.
• 22:38
  Like if you're paying five tram 10 KA month, you know, how much
• 22:41
  am I going to make off of you? I'm going to reduce your bill to
• 22:44
  1K but there's charge you but. The problem is that there's, I
• 22:49
  mean, we were paying them. We I mean, we were going to have
• 22:52
  to pay them 80 KA month. So I mean, like I this my, my,
• 22:55
  my hypothesis is, is more that and, and this is a question.
• 23:00
  I not my business, right? So I have no idea.
• 23:03
  I find it hard to believe that you're not going to end up in a
• 23:09
  world where there are 8 enormous number of companies with an
• 23:14
  enormous amount of data. Like look, the number one thing
• 23:17
  about AI is we now have more data, not less data, not and not
• 23:20
  by like 20%, by like 100. XI am sure.
• 23:25
  By the way, This is why 5 trend changed their pricing.
• 23:28
  They felt like they had leverage in this context.
• 23:30
  Exactly. And so now that I have 100 X the
• 23:34
  data that I need to move around and deal with, like, I mean,
• 23:38
  like truly insane scale for a company our size, I've got to
• 23:42
  believe that the number of people for whom it is worth it.
• 23:47
  And then, and then to your point on like, I don't want to build a
• 23:49
  services business. I mean, right, like every other
• 23:52
  podcast on tech and at least in the US is like every business
• 23:56
  has to become a service business for the next two years.
• 23:59
  And so like, I separately am curious why you believe the
• 24:02
  opposite? Because I I just don't believe
• 24:04
  that there is such a thing as a as a software company that
• 24:07
  cannot provide services, that gets very big.
• 24:09
  I literally don't believe it. Think it's that there can be
• 24:13
  software companies that that won't provide services, but the
• 24:17
  decision for me is more like we can and we will provide
• 24:22
  services. What are the types of services
• 24:24
  that we want to provide and what is the upside?
• 24:26
  What I need to optimize for today is every service work that
• 24:30
  I do, I am doing it not to make money.
• 24:33
  I'm doing it to cover some of my costs, but more importantly, I
• 24:36
  need to learn something out of that that work that I do, you
• 24:39
  know, so I have two paths in front of me.
• 24:42
  I could either do some services to help people move off of 5
• 24:46
  Chan and make some money on that process, or I can help
• 24:49
  organizations adopt agents for their analytical workloads and I
• 24:52
  do some some servicing work. I'm always going to prefer that
• 24:55
  ladder because there's so much more to explore, so much more to
• 24:58
  learn there that would be a lot more applicable to the to the
• 25:01
  rest of the world then moving people off of five Tran.
• 25:04
  You know, eventually I imagine we will do that, you know like
• 25:07
  this is already happening on its own.
• 25:09
  We already have like this inbound flow with regards to
• 25:12
  like every time 5 Tran bumps their prices like we get we get
• 25:15
  inbound leads and stuff, but that's not where I want to put
• 25:20
  services effort. You know, I don't I don't find
• 25:22
  that interesting. Like we already take care of
• 25:24
  that. That's already a done market.
• 25:27
  In my perspective, this is a lot more exciting where if I am
• 25:31
  going to do services, I will be personally involved with all of
• 25:34
  those conversations as well. And I need to make sure me, my
• 25:37
  Co founder, my organization learns something out of that.
• 25:40
  You know, so it's not that I don't want to do the service,
• 25:43
  you know, we do, we get our hands very dirty in a lot of
• 25:47
  these cases. This is that, you know, cutting
• 25:49
  the costs is is not where I want to put that time and effort.
• 25:54
  One thing that I think is really interesting about your business,
• 25:57
  I don't know much, but the the core of this is open source.
• 26:01
  I think is really interesting because I there, I mean, for a
• 26:03
  long time open source drove so much and now with AI, it's
• 26:07
  really changing that landscape. And my, my actual, my first
• 26:10
  start in software actually was building open source software as
• 26:13
  well. The WordPress ecosystem a little
• 26:16
  bit different, not as technical, but the point is like we would
• 26:20
  put everything out for free. And basically we had like we
• 26:23
  made our money by effectively selling services in the form of
• 26:27
  a better package and better customer service.
• 26:31
  And so I'm curious how you're thinking about open source in
• 26:34
  this context, Like what does it do for your Business Today and
• 26:36
  what do you think the future of that is given where you're at
• 26:39
  today? So we initially started building
• 26:42
  everything open source for a couple of reasons.
• 26:45
  First of all, we are selling core data infrastructure like
• 26:49
  companies that we work with build a lot of stuff on us.
• 26:53
  They depend on us, they depend on us a lot and and we wanted to
• 26:57
  make sure that them trusting a, a small company like ours in the
• 27:03
  early days would not put them in a in a very hard position.
• 27:06
  If he were to ever run out of business, if he were to ever
• 27:09
  have any trouble, they would have all of the software
• 27:12
  available for them. They could at worst, like if all
• 27:16
  of us got in the bus and we, we died, they could pay someone a
• 27:19
  couple 1000 bucks to get everything up and running in a
• 27:22
  day or two. You know, so this is.
• 27:25
  Point early on. Absolutely.
• 27:27
  It's the list like we're, we're still like, we're still small,
• 27:30
  we're in a much better position than like 2 years ago.
• 27:33
  But it's still, it gives people a lot of Peace of Mind that all
• 27:37
  of the core tech, everything they depend on is as part of our
• 27:42
  open source tooling. And and we try to do this in a,
• 27:45
  in a sustainable way where intentionally all of our core
• 27:48
  product design starts from the open source and then flows
• 27:51
  through our cloud offering as well.
• 27:53
  So that means that it is incredibly rare other than like
• 27:57
  some enterprise governance weird features that we built into our
• 28:01
  cloud product. Pretty much everything goes
• 28:03
  through our open source setup, you know, so my team
• 28:05
  contributes. If you take a look at our repos,
• 28:07
  there's 10s, if not hundreds of deployments that we do every day
• 28:11
  to our open source software and then and then take it to the
• 28:14
  cloud. I think with AI, this is
• 28:17
  becoming an even better advantage for us because part of
• 28:20
  the reason why we were building this open source was also OK.
• 28:22
  People can contribute and we want to cover a lot of different
• 28:27
  touch points, a lot of different integration points.
• 28:29
  We're going to make mistakes, we're going to fall behind.
• 28:31
  We're going to, you know, there's going to be APIs that we
• 28:33
  don't know how they work. People can always contribute
• 28:36
  With AI, these contributions are becoming much easier.
• 28:39
  You know, people can just jump in and submit APR and, and get
• 28:43
  something deployed in about a day that both solves their
• 28:46
  problem. And now we actually have a new
• 28:48
  integration point that someone else just committed for everyone
• 28:51
  else to use and benefit from. You know, AI reduce that barrier
• 28:54
  quite a bit. It is adding a bit of noise, to
• 28:57
  be honest on the on the maintain maintenance side.
• 29:01
  But regardless, it's definitely a net net win for us.
• 29:04
  Like I just enjoy putting all of this this out in the open.
• 29:10
  Part of that is also like emotional.
• 29:11
  I feel like we're giving it given back to the community that
• 29:13
  we have also benefited from. I want you to tell me why I'm
• 29:16
  wrong and Rashad, if I mischaracterize from and this,
• 29:22
  please tell me why I'm wrong as well.
• 29:24
  But you know, I really like this.
• 29:26
  Focus on solving time to insight, trying to allow
• 29:31
  operators to be informed and make decisions.
• 29:33
  Move time to action a little bit faster and but more informed as
• 29:37
  well. I think in in a lot of the
• 29:40
  spaces that I've been in around incredible people, they're
• 29:44
  overconfident and that's incredible, OK.
• 29:46
  They just move forward, right? I used to call it Raging Bull,
• 29:49
  right? All of the most incredible
• 29:50
  people I knew would make these decisions based on like two data
• 29:53
  points and then an condition. And they'd be like, yes, let's
• 29:55
  go, right. And they didn't actually need
• 29:57
  very, you know, or maybe it was, you know, this aggregate got
• 30:01
  instinct of a lot of incredible information or it was just
• 30:03
  enough to be persuasive for them to take action.
• 30:07
  But, you know, actually getting really good data into those
• 30:10
  people's hands and letting them feel even more confident that
• 30:13
  they haven't had to scribble things together in convoluted
• 30:17
  spreadsheets or trust someone else for that information.
• 30:20
  So I think it's great that you're solving that for those
• 30:23
  those people, right, those operators.
• 30:24
  But I think there's like a there is an end, right, to getting
• 30:28
  good information. You want it to inform good
• 30:31
  decision making, good action, right?
• 30:34
  Are there any plans for you to move from time to insight to
• 30:37
  time to action and actually percent empower the actions?
• 30:41
  Because I look at from right from art is interesting.
• 30:43
  I don't know too, too much. I keep seeing for Mart in all of
• 30:47
  these really big stores and I'm like, oh, cool, what they're
• 30:49
  using for Mart. That's nice, but it's a lot of
• 30:51
  action, right? It's not actually a lot of like
• 30:54
  trying to get information into these people's hands.
• 30:56
  Maybe it is that this way. Correct me if I'm wrong, Rashad,
• 30:59
  but a lot of it is actually like, let's take the best
• 31:01
  information and let's act on your behalf to achieve eve like
• 31:04
  a business outcome, right? And I imagine this is, you know,
• 31:07
  maybe no falls in the first camp, right?
• 31:09
  Time to insight, get that information that lets you act,
• 31:12
  whereas Vermont maybe falls in that second camp, like get the
• 31:15
  information, but then actually act with it, right and solve
• 31:18
  that business case. So, yeah, what are the plans for
• 31:20
  the for the product? Are you ever going to move into
• 31:22
  time to action? I, I mean, I don't think you're
• 31:25
  wrong at all. I think we are, we already moved
• 31:28
  into that in the sense that I consider us as like a decision
• 31:34
  making business. You know, in the end we're
• 31:35
  trying to make help people make better decisions faster and
• 31:39
  cheaper, you know, and ideally we're going to help them do that
• 31:43
  by eliminating some of those decisions and automating quite a
• 31:46
  lot of them. So using our agents today, we
• 31:49
  have quite a few customers that are automating their marketing
• 31:52
  campaigns, customers that are automating their their CRM,
• 31:55
  customers that are operating up the running AP tests on their
• 31:59
  play store listings or on their like lending pages.
• 32:03
  We even use it for our own internal lending pages and
• 32:07
  marketing material that constantly updates itself, you
• 32:10
  know, so we already are there in terms of taking those actions.
• 32:15
  However, this is not like a, this is not a problem that a
• 32:20
  company like ours can console it by itself because a lot of these
• 32:24
  actions are dependent on other products, other companies, other
• 32:27
  people chat to be able to actually take those actions.
• 32:30
  So until that is there, it is going to be very hard to
• 32:35
  automate all of these. But my sort of internal feeling
• 32:40
  about how we're going to talk, yeah.
• 32:43
  I'm sorry to cut you off, but as you're saying this, does that
• 32:47
  mean at some point you're going to start competing with like a
• 32:49
  high touch basically? We already are like it's not
• 32:53
  just like it's not even high touch kind of like reverse it.
• 32:56
  We already do that. We've been doing that for three
• 32:58
  years the way that it was. Evolved right like high touch
• 33:01
  doesn't do reverse ETL anymore. High touch is squarely now in
• 33:07
  like agent data pipelines to drive marketing.
• 33:10
  So they have chosen to narrow their area that they're going to
• 33:16
  operate in to marketing actions. And and so like actually what
• 33:21
  their focus now is is like like driving systems of action where
• 33:27
  like they don't have the system about like they don't have the
• 33:30
  e-mail provider, they don't have the SMS provider, they don't
• 33:33
  have the ad provider. But the marketer can log into
• 33:36
  their system and then say, hey, understand this data and make
• 33:41
  act take action in this way. So that way they can drive.
• 33:44
  In particular, when you're doing it at that scale, you can drive,
• 33:47
  for example, much better personalization or much better
• 33:50
  like nuanced rules for when to drive a campaign and when not
• 33:53
  to. But if that's what you're
• 33:55
  saying, then yeah, like the Jack is right.
• 33:59
  Like the opportunity size just becomes much bigger because the
• 34:04
  the set of opportunities for companies like high touch and
• 34:07
  some again, high touch happens to be focused on marketing is
• 34:11
  pretty large, right? There's like another a
• 34:14
  competitor called Oxia, There's a bunch of them.
• 34:17
  Yes. So we are we're moving in that
• 34:19
  direction. I don't think in the short term
• 34:22
  we're going to be we're going to be building any, any industry
• 34:25
  specific sort of touch points. It's likely going to be a lot
• 34:29
  more like we are building that platform for people to build
• 34:32
  these sorts of automations on top of, on top of our, our tech.
• 34:36
  Which means that we do a lot of work around using, making our
• 34:40
  agents use both the data warehouse as well as the browser
• 34:43
  very well. So that customers can actually
• 34:45
  build browser based automations onto the what they want to do
• 34:48
  based on the data that they have in your in their warehouse.
• 34:51
  And then getting these agents to, to take these actions.
• 34:54
  I do not see us building tooling to make marketing people to get
• 35:01
  better insights from the from their marketing campaigns.
• 35:04
  To be quite frank. I think we're moving towards a,
• 35:08
  a future where you don't have marketing people or like you
• 35:13
  have 99% less, you need a lot less with these people because
• 35:19
  the peoples that people that you have can can manage a lot more
• 35:22
  today or you don't even need them because you can manage all
• 35:25
  of these things because the number of decisions that you
• 35:27
  need to make is 90% less now. So yes, we are moving in a,
• 35:33
  there is a universe where we are competing with high touch or
• 35:39
  kind of like marketing automation tools in one place,
• 35:42
  CRM automation platforms in another, data integration tools
• 35:47
  with with another. And it might seem like we're
• 35:50
  we're taking fights, wars with everyone with everywhere.
• 35:54
  I don't find that scary. I think it's a it's where things
• 35:57
  are evolving. And no, I don't think it's yeah.
• 36:01
  I don't. I don't think it's scary, I just
• 36:03
  think you need to convince a few friends to build some service
• 36:06
  businesses. I guess, you know, you seem to
• 36:11
  be very into building services. Why don't you help us out?
• 36:15
  Let's build it together, I. Am oversubscribed.
• 36:20
  I'm I'm oversubscribed man. I'm like I, I here's what I'll
• 36:24
  say. Jeremiah knows this about me.
• 36:27
  I, I think, sorry, I should have asked this much earlier.
• 36:30
  Is is your focus today SMB or enterprise?
• 36:33
  SMB. OK.
• 36:36
  We have like quite a few enterprise customers, but again
• 36:39
  like up we're trying to optimize the learning.
• 36:41
  So I need people and companies that can move faster.
• 36:43
  So it's like companies that are smaller than two hundred 250
• 36:47
  people is sort of our our target.
• 36:51
  Yeah, Yep, Yep. I got it.
• 36:52
  I got it. Yeah, I think so.
• 36:55
  I mean, so first of all, as long as you're focused on SMB and
• 36:59
  like you're driving the platform capabilities, everything,
• 37:01
  everything that I set up until now, you can like entirely
• 37:04
  ignore because obviously the business model does not support
• 37:07
  it. Just to clarify, right, If and
• 37:11
  when you decide to go to enterprise, the thing that is I
• 37:15
  am 100% sure underappreciated is enterprises the way that the way
• 37:23
  you need to like chat with the CX of an enterprise.
• 37:27
  Like these are the most sophisticated people on the
• 37:30
  planet. Like you don't get to be the CDO
• 37:32
  or CMO of a multi billion dollar company unless you're really
• 37:35
  good at what you do. And these people are very
• 37:39
  sophisticated and they when you say to them, I'm going to help
• 37:44
  you move your organization forward in this tangible way.
• 37:49
  They already also have a hundred person engineering team.
• 37:52
  And So what, what, what actually becomes the problem is, given
• 37:57
  the level of sophistication with which I run my business, if you
• 38:00
  don't to come here and spend 3 months working alongside my
• 38:04
  team, the probability of you actually moving the needle from
• 38:07
  my business devolves to 0. That's, that's actually like
• 38:11
  what ends up happening. And, and they're correct, by the
• 38:14
  way. And the consequence of that, at
• 38:17
  least in the US, is there's this massive dearth of what is now
• 38:23
  being called forward deployed engineers services, right,
• 38:27
  Formerly known as services. And it's so massive that
• 38:32
  everyone is asking anyone who is competent with AI to please come
• 38:38
  and sit next to their team to enable something that will
• 38:42
  actually move the business forward.
• 38:44
  It is really, it is really hard. Like I can't understate how hard
• 38:49
  it is. And, and I think that underneath
• 38:53
  really hard is massive effing opportunity, just like massive
• 38:58
  opportunity because the amount that it's worth to them is 10s,
• 39:03
  if not hundreds of millions of dollars.
• 39:05
  And therefore the amount that it is worth to vendors like us is
• 39:09
  millions of dollars, right? And so it's just that I think
• 39:13
  that the enterprise opportunity is 100% understated.
• 39:16
  I mean, like it is almost comical.
• 39:19
  Did you read the SpaceX S1? I'll just like shift topics a
• 39:22
  little bit to make it a little fun.
• 39:24
  Did you guys read the SpaceX S1? I haven't there's.
• 39:26
  This chart in the SpaceX S1, which shows the total
• 39:30
  addressable market for all of their different markets and it's
• 39:33
  like half a trillion launch business, 2 trillion of like
• 39:38
  Star Link, and then it's like 23 trillion enterprise AI.
• 39:44
  OK, So all of their previous businesses look like little like
• 39:48
  dots. And then it's like enterprise AI
• 39:50
  is like 90% of their future tab or 95% of their future tab.
• 39:57
  I mean it. There we go.
• 39:59
  Look at that beauty. Look at that beauty.
• 40:05
  But. I mean, this makes sense.
• 40:06
  I mean, if you think about it, we're talking about this
• 40:10
  technology is disrupting the way that we work and the way that we
• 40:12
  work was built over decades, especially when you talk about
• 40:16
  like a Fortune 500 company that's been around for 50 years.
• 40:19
  There's 50 years of doing things in a certain way and building on
• 40:23
  top of this foundation that now has to be undone.
• 40:25
  That is not easy. It's going to be extremely
• 40:29
  expensive, extremely disruptive, and it will take a lot of time
• 40:32
  and money to make those changes. This is not it.
• 40:34
  Like it's a very obvious reason to me why this is such a massive
• 40:38
  opportunity. Is there anyone that says this
• 40:41
  is not a massive opportunity? I also, I'm not trying to
• 40:44
  convince you by the way. I'm just no, no, no.
• 40:46
  No, no. But I'm just.
• 40:47
  Wondering, I'm just wondering. Today you're learning by
• 40:51
  offering to SM BS and we're all agreeing.
• 40:54
  We're doing the podcasting, we're saying the same thing and
• 40:56
  we're all agreeing, which is wonderful.
• 40:58
  So when is it that you will switch to offering to the
• 41:01
  enterprise? What's the target The.
• 41:05
  Services part like us, us offering a product to enterprise
• 41:10
  versus US offering the services part to enterprise is I find it
• 41:14
  very different. I would probably say like at
• 41:16
  least a good couple more years for us to ever get into that.
• 41:20
  It's not because I don't think there's an opportunity, it's
• 41:23
  just because I that opportunity does not excite me.
• 41:28
  You know, there might be like massive opportunities there.
• 41:31
  There's massive opportunities everywhere.
• 41:34
  I don't care. You know, I think there is a,
• 41:37
  we're operating in AI would much rather be Stripe than being
• 41:45
  McKinsey. You know, I would much rather
• 41:47
  bet that some of the SMB, some of the smaller companies, some
• 41:50
  of the future looking companies that we're working today with
• 41:53
  are going to become the enterprises of tomorrow.
• 41:55
  And I try to build for them from Solit foundations.
• 41:58
  I find that idea much more exciting than going in at a
• 42:01
  Fortune, Fortune 100 company and trying to like change the way
• 42:07
  completely they work. I 100% there's a huge
• 42:12
  opportunity, huge, huge. You feel like no disagreement at
• 42:15
  all. I just don't find it exciting,
• 42:18
  so I guess yeah. What's what's your goal?
• 42:21
  For your business, for yourself over the next 5-10 years because
• 42:24
  that I think I know we get into these conversations a lot on
• 42:27
  this podcast, but that's that definitely helps guide where you
• 42:30
  adopt because I'm curious what you know what your what your
• 42:32
  ideal outcome is personally. So I have this, I have this
• 42:37
  internal sort of thinking that that the time to incite or time
• 42:45
  to action, like Jack put it is, is basically supposed to
• 42:51
  approximate the zero over time. So if we're successful, the
• 42:55
  companies that end up working with us, the companies that
• 42:58
  we've been helping along the way are going to be the ones that
• 43:01
  are going to be making the the most amount of decisions and
• 43:05
  they're going to be the ones that can iterate the fastest.
• 43:08
  You know, So the goal is you take a look at the Fortune 505
• 43:14
  years down the line and half of them are less than 10 years old
• 43:22
  and half of those are our customers.
• 43:26
  You know, maybe there is a future that that will happen.
• 43:29
  And I would like to be on, on that part of, of, of powering
• 43:33
  those kinds of organizations where they're actually much
• 43:35
  smaller in headcount, much bigger in operations, much
• 43:38
  bigger efficiency gains. And, and we're challenged in
• 43:41
  that regard, sort of making that a reality.
• 43:44
  I do not see us winning over business from Oracle.
• 43:51
  You know, I do not see us going behind like companies that that,
• 43:56
  you know, Microsoft holds very close and very dear large
• 44:00
  enterprises, you know. So that will be my, my dream
• 44:04
  scenario. That will be the goal that we're
• 44:05
  building towards, to have a few years down the line.
• 44:09
  That's what I would like to see. That's what we're building
• 44:12
  towards. How do you guys rank order
• 44:16
  information right? I'm sure you've all seen it.
• 44:19
  And you know, everyone's begging for your attention.
• 44:21
  Rashad Jeremiah at your companies, right?
• 44:25
  It's how do you take information that's coming from your team or
• 44:29
  just very true information and decide like what order it should
• 44:33
  come at you? What I mean is that, like I've
• 44:35
  seen decisions made before based on accurate data and sometimes
• 44:39
  enormous amounts of data, and they have the exact opposite
• 44:42
  conclusion about the action that should be taken, right?
• 44:45
  This is true, and this is true, right?
• 44:47
  But both of them are arriving at different conclusions, right?
• 44:51
  I hate to say that maybe it comes down to opinion at the end
• 44:53
  of the day. What excites you or what is the
• 44:55
  most persuasive to the decision maker at that moment of time and
• 44:58
  like links to the most amount of information that they have when
• 45:01
  they read it? But yeah, I'd love to hear how
• 45:03
  you guys think about that. I see this all the time.
• 45:05
  Are people making decisions based on some amount of limited
• 45:09
  data or a lot of data, right? They're taking action, but how
• 45:13
  are they rank ordering it? I guess it's the same question
• 45:15
  as how do you know which question to ask, right?
• 45:18
  Because you might be asking the entirely wrong question to
• 45:20
  inform like a decision, right? Yeah, I have this funny thing
• 45:24
  that I say inside the company, which is either it matters or it
• 45:27
  doesn't matter. And it's like, which sounds
• 45:31
  obvious, but man, people don't think that way.
• 45:35
  Like people come to me with like, oh, like they said, this
• 45:39
  isn't I'm like, doesn't matter to them.
• 45:42
  And they're like, oh, this. And I'm like, no, no, no, no.
• 45:44
  Can you answer if it matters or if it doesn't matter?
• 45:47
  Because we only matter if it matters.
• 45:50
  Like it only matters if it matters.
• 45:52
  And, and like the consistency with which you have to say that
• 45:56
  is astonishing. Like, I'll give you a good
• 46:00
  example. OK, Let's just say someone
• 46:02
  inside your company, like Jeremiah, I'm sure you deal with
• 46:04
  this. Someone is complaining about
• 46:06
  someone else is work output. OK?
• 46:08
  It's like person A complaints about persons B's work output.
• 46:11
  And I'm like, does it matter? OK, but this is my number one
• 46:14
  question every time I this fucking happens.
• 46:17
  OK, I'm like, doesn't matter. And they're like, you know, it
• 46:21
  should be this, this is. And I'm like, it can be whatever
• 46:23
  the hell you want. Let's pretend person B's work
• 46:25
  output was zero. Has your life changed?
• 46:28
  OK, now let's pretend person B's work output was fucking
• 46:31
  exceptional, the best on the planet.
• 46:33
  Has your life changed? Your life doesn't fucking
• 46:36
  change. So stop complaining.
• 46:39
  Like either it matters or doesn't matter.
• 46:42
  Like, and and this is like a crazy thing that people don't
• 46:46
  understand. Like people don't understand
• 46:49
  that companies have to do a small number of things to move
• 46:55
  the needle and therefore it matters or it does not matter.
• 46:58
  And just look at that list and tell me, is this action that you
• 47:02
  are taking an action that matters?
• 47:04
  And if it's not, then just please leave me alone.
• 47:08
  That is it, yeah. I think there was this warm
• 47:10
  fuzzy feeling that everyone got the first time AI told them they
• 47:13
  were stupid and that they were wrong and that maybe they should
• 47:16
  be paying attention to something else.
• 47:17
  And people do this on purpose, right?
• 47:19
  They've finished their prompts with tell me why I'm wrong or
• 47:21
  whatever happens to be. Or they, you know, set up their
• 47:24
  skills to constantly tear them to pieces.
• 47:27
  But yeah, yeah, you know, I wondered to what extent you have
• 47:31
  to be opinionated, right? Like I imagine that you, you
• 47:34
  come into contact with this a lot, Jeremiah with no, right.
• 47:37
  If you leave it to the operator to decide which questions they
• 47:39
  want to ask, sometimes they're going to go through a sequence
• 47:42
  of different things that don't matter right, before they arrive
• 47:45
  at the thing that you already know because you have all the
• 47:47
  data. That is probably probably a good
• 47:49
  place to start, right, In terms of a question on a post purchase
• 47:53
  survey, right. That's actually going to inform
• 47:55
  good decision making, right? But like, you know, your
• 47:57
  customers, I imagine, Barack, and I imagine this applies to
• 48:01
  format as well, right? You have to use what matters and
• 48:04
  you have to be opinionated about what matters and what doesn't
• 48:06
  matter because there's swathes of data that can inform decision
• 48:10
  making, right? But Barack, if the operator is
• 48:12
  in charge of the question that they ask, how do we make sure
• 48:15
  that they get the right thing, right?
• 48:18
  Is that the model coming back to them and saying, actually, no,
• 48:21
  don't pay attention to that, You don't need to know about that.
• 48:23
  Pay attention to this right. I was gonna say something about
• 48:27
  this before I jump into it because he got like, no.
• 48:30
  No, no. I was saying yes.
• 48:32
  I was saying, Jack, you fucking got it.
• 48:33
  That's what I was saying. I was like, this is this is this
• 48:35
  is a great fucking question. No, no, no, I'm, I'm, I'm in,
• 48:39
  I'm in Brock. I want to hear it.
• 48:40
  Yeah. I.
• 48:41
  I, I think, I think the way I think about this is there is
• 48:45
  there's types of organizations that so there's, there's two
• 48:47
  types of pilots that we do OK. And, and I, I was not expecting
• 48:51
  this to happen because it became much more apparent with the,
• 48:53
  with the whole AI stuff that we're doing good.
• 48:55
  Then I find I find it's funny. So in one set of pilot, we are,
• 49:02
  we set things up for your organization.
• 49:04
  We give them two weeks to to use the product for free.
• 49:07
  They can, they can explore every aspect of our agent, our
• 49:10
  integrations, pull data, push data like they can do whatever
• 49:13
  they want and they're going to hammer the product.
• 49:15
  You know, they're going to like bankrupt us.
• 49:18
  They they do like they threw a lot of people on the product, a
• 49:21
  lot of questions, a lot of they're very demanding,
• 49:24
  constantly sharing feedback, etcetera.
• 49:27
  And then there's this other group where they, they sometimes
• 49:32
  like they use a product moderate amount, but most of the time
• 49:35
  it's for very superficial stuff that they want to ask, they want
• 49:40
  to learn. And, and we've tried to make and
• 49:44
  make them better at that. You know, like we're we're sort
• 49:47
  of like, is this really what you should care about?
• 49:49
  Like shouldn't you ask this kind of stuff like you're like an
• 49:51
  e-commerce company? There's like these and these and
• 49:53
  the diary tracking these metrics.
• 49:54
  Do you know what these mean etcetera.
• 49:57
  A lot of them do not know what they want to ask.
• 49:59
  You know, a lot of them do not have an idea about what what
• 50:03
  matters for their what matters for their business.
• 50:06
  I think there's nothing rational behind this.
• 50:09
  You know, on a grander scale, I think it's a lot of emotions and
• 50:13
  gut feeling about what they think matters versus what it
• 50:17
  actually matters. So at this point, I'm of the
• 50:20
  opinion that if a company does not have a a very clear idea
• 50:25
  about tons of stuff they would they would throw at if they had
• 50:29
  like unlimited amount of question answering machines,
• 50:32
  there's probably not much of a chance that we can actually save
• 50:36
  them from from the state that they are in, you know, so I
• 50:39
  think. A lot of that pain driven
• 50:41
  ultimately. So if somebody has some sort of
• 50:44
  like, because I have the same exact problem and it is so hard,
• 50:48
  really easy to get somebody to do one thing.
• 50:50
  It's really hard to get them to do a second thing, third thing,
• 50:53
  fourth thing, and less. You can address a specific pain
• 50:57
  point that they have. And oftentimes, like for me, I
• 51:00
  only get to interact with a small portion of what somebody
• 51:05
  like whatever my job is. If I'm coming in and I'm using
• 51:07
  this piece of software, it's only addressing a portion of my
• 51:09
  job. So I don't know what's going on
• 51:11
  with the other 80% of the things that that person is dealing
• 51:15
  with. And so for me to push something
• 51:17
  to them, I know that I need them to do this thing.
• 51:20
  I know it will be better for them.
• 51:21
  But to actually get them to engage with it is extremely hard
• 51:24
  because they have all these other pains that they're dealing
• 51:26
  with, or maybe not always pain to.
• 51:29
  It could be things they're excited about.
• 51:31
  They're just not actively thinking about the problem I'm
• 51:33
  solving for them. And if they're not thinking
• 51:35
  about it, it's hard. As soon as they're thinking
• 51:37
  about it and they're engaging with it, it's extremely easy to
• 51:40
  get them to do additional things.
• 51:42
  Maybe this relates back to you guys see the new codex release
• 51:46
  like goal slash goal as like a as a command, right, And you can
• 51:50
  set a goal. Maybe this comes up.
• 51:52
  They're like decision makers and bounders and operators, like
• 51:56
  setting, you know, deciding what should be valued and then
• 51:59
  orienting around that. I was, you know, I imagine all
• 52:02
  of us have a friend who pays too much attention to politics and
• 52:06
  talks about it too much and you're like, oh, stop, right.
• 52:09
  I saw my girlfriend was watching some the South Korean cooking
• 52:13
  show the other day and she showed it to me.
• 52:15
  She was like, oh, look at how incredible they are.
• 52:16
  And I said, you know, how weird would it be if they just started
• 52:19
  talking about Trump in geopolitics, right?
• 52:21
  They start having opinion. They're focused on something.
• 52:23
  They have a goal, right, food or whatever happens to be within
• 52:26
  the time constraints. So yeah, maybe that's maybe
• 52:29
  that's just the role of the founder, right?
• 52:31
  And the the operators to set the direction and what the goal
• 52:34
  actually is and how people are orienting themselves, right.
• 52:37
  I'm. Going to say something that's
• 52:38
  going to sound so bad, especially if my tier team hears
• 52:41
  this, but life should be simple, OK?
• 52:46
  There's a very small number of things that matter to a
• 52:48
  business. Increase revenue and reduce
• 52:50
  cost. You should know what your
• 52:53
  business formula is to increase revenue and to reduce cost
• 52:58
  because it's usually pretty simple.
• 53:01
  If you're not working on those things that matter to how you're
• 53:04
  doing that or to that matter to how you can get there, then just
• 53:08
  stop talking. The problem is that people are
• 53:12
  good at certain things and they want the things that they are
• 53:16
  good at to matter because it makes them feel good.
• 53:19
  Now, I don't care how people feel.
• 53:22
  I care about growing my business.
• 53:24
  And therein lies the kerfuffle, right?
• 53:27
  And, and, and that's the like, sad reality here is like people
• 53:31
  want to feel progress and sometimes they're feeling
• 53:35
  progress on something that has no benefit to the end business.
• 53:40
  And, and, and that's like the really tough thing to nail.
• 53:45
  And to nail it in a way in which does not demoralize your team so
• 53:49
  much that they don't even do the work, but instead you need to
• 53:53
  get them to recenter on the things that do matter.
• 53:57
  And doing that in a way in which is like, no, no, listen, like
• 54:02
  Rishabh, you're not a dummy. It's just that you're using your
• 54:07
  strengths in the wrong way. I understand that you're tempted
• 54:10
  to use it in this way, but just stop it because it doesn't
• 54:14
  actually help you And and that that is the central problem,
• 54:19
  right? Yeah.
• 54:20
  And so somehow you got to do that.
• 54:22
  Yeah. Yeah.
• 54:23
  I mean, we had when I took over stamps, like one of the things
• 54:27
  that was a problem there was just optimizing for the wrong
• 54:30
  thing ultimately. And so like the the technology
• 54:33
  there the team had. I like the the biggest signal of
• 54:37
  this to me was that in three years the business had not
• 54:42
  grown. The actual like actual business
• 54:44
  outcomes that we were driving hadn't changed.
• 54:46
  But the we were doing 10 times as much work with the the
• 54:52
  information that was happening there as it had been three years
• 54:55
  before, which is 10 times as much cost work.
• 54:58
  All these things happening to deliver the same outcome like
• 55:01
  that means that you have optimize for the wrong thing.
• 55:04
  The optimize it, the optimization was towards showing
• 55:06
  bigger numbers, not towards driving an actually different
• 55:10
  outcome. And like, that's the that's the
• 55:12
  problem ultimately in that context.
• 55:14
  But how do you, how do you, OK, you guys have more experience
• 55:17
  than this than than I do? How do you guys do this without
• 55:20
  you actually being in every room where every little decision is
• 55:23
  being made? No.
• 55:24
  No, no, dude, this is you want to know the craziest unlock that
• 55:27
  I had? So there are two unlocks that I
• 55:30
  had recently that are potentially interesting.
• 55:34
  And then I'm sorry, I feel like I have to hop every single time.
• 55:38
  Like at the end we got to find a way to like fight, like get
• 55:42
  ourselves on time. But the, the, there are two
• 55:44
  things I did 1 is I made an agent that is like knows what
• 55:51
  the priorities of the business are.
• 55:53
  And every time someone does something, it will like, chime
• 55:57
  in and say like whether it matters to the business or not.
• 56:00
  So now what's happening is I'm depersonalizing the feedback
• 56:05
  because now what the person feels like is that a bot is
• 56:07
  giving them feedback. And they're actually much more
• 56:09
  open to getting feedback, which is a very weird phenomenon.
• 56:13
  They were much more averse to asking me for input when I would
• 56:16
  tell the person to their face, hey, this doesn't matter.
• 56:19
  But when my boss says this doesn't matter, it doesn't hurt
• 56:22
  their feelings. Which is a fascinating dynamic
• 56:25
  and has been a massive unlock. So that's one thing that I did.
• 56:29
  The other thing that I did is I have stand ups.
• 56:33
  I don't have weekly meetings. I have stand ups every other
• 56:35
  day. And I tell people that their
• 56:39
  actions and their behavior is well motivated, but that it's
• 56:43
  incorrectly placed. And I think that people just
• 56:46
  need to hear that their actions are well motivated.
• 56:50
  And if you say those words out loud, then they feel validated
• 56:53
  enough that you can actually then say, but it's wrong.
• 56:57
  Like I understand and it's wrong, right?
• 57:00
  And so just apply that same thing to this other direction
• 57:03
  and they're more much more willing to hear it.
• 57:06
  But those are the two things that I do is because I do it
• 57:08
  every other day. It doesn't feel like I was like
• 57:11
  sitting on it. People feel like it's real time.
• 57:14
  And when people feel like it's real time again, it becomes less
• 57:17
  personal because it becomes about the action, not about the
• 57:21
  person. But does this mean that?
• 57:26
  Does this mean that you're sort of like, with that use of like
• 57:29
  agent, you're essentially mimicking like you being in
• 57:32
  every room? Is there is there a way that
• 57:36
  this could become the, the company's part of the company
• 57:40
  culture where, you know, there's some percentage of the company
• 57:44
  that sort of internalized all of these what what actually matters
• 57:48
  for this business kind of thinking and every part of the
• 57:51
  company without you actually being the one that that drives
• 57:56
  all of that out. Of course you are like the
• 57:58
  strongest pillar to culture, but could it work without you being
• 58:02
  there or without? 100 Because now what's happening
• 58:07
  is people give feedback back to my agent when they think it's
• 58:10
  wrong. And so what's actually happening
• 58:13
  is my agent is not a reflection of just me.
• 58:16
  It's a reflection of the best parts of everybody, but it is
• 58:20
  founded on what are the things that actually matter for the
• 58:22
  business. So, so because, because see, The
• 58:24
  thing is about training an agent is you can write anything to the
• 58:27
  memory file. So I'll see like an idea from
• 58:30
  someone else and I'll say like, hey, and then I'll like tag my
• 58:33
  agent, command that to memory. That's a good idea.
• 58:36
  And so now it actually is not literally just me in every room.
• 58:40
  And so it has actually something that can permeate through the
• 58:43
  company. Does that does that make sense?
• 58:45
  So you you, you rely on agent being sort of like this
• 58:48
  collective set of learnings and like the the bearer of all these
• 58:53
  learnings and to to the rest of the organization which.
• 58:56
  Is what every company is going to look like, yeah.
• 58:59
  It's an interesting version of plan for Insight, right?
• 59:01
  It's an interesting version of get the data or as you get the
• 59:03
  data, of course, get back on goal.
• 59:05
  Right. Get back on target.
• 59:07
  Yeah. Yeah.
• 59:08
  Cool. Yeah.
• 59:08
  This is Jeremiah Rashad. I know you've got to go, so I'll
• 59:12
  wrap this up. Barack, this was awesome.
• 59:13
  It was great to meet you. Thanks a lot for having me.
• 59:16
  Really appreciate it. Thanks guys.
• 59:19
  Cheers.