Latest / 1 to 100 / E42: SaaS is changing with Yash Chavan

Transcript

E42: SaaS is changing with Yash Chavan · 1 to 100

• 0:00
  So this guy, I meet him in Bombay 3 weeks ago, makes me
• 0:03
  feel like the hottest girl in the room because you know what
• 0:06
  he says? He says, you know, I'm like, I
• 0:09
  watch the SAS operators all the time.
• 0:12
  And then at the end of our coffee, he's like, that felt
• 0:14
  like a private SAS Operators recording.
• 0:19
  And I was just like, take me now, you know, like, warms my
• 0:25
  heart. Warms my heart.
• 0:26
  It was awesome. Yeah.
• 0:28
  It was lovely meeting you, Rishab.
• 0:30
  Yeah. You guys met up in Bombay, Yes.
• 0:33
  Yeah, That's cool. And.
• 0:34
  Then I was going to say I actually quietly, I'm not trying
• 0:37
  to flatter you, Rishab. Like I actually got like a real
• 0:41
  SAS operators episode. So like for myself, you know,
• 0:45
  one-on-one, maybe this could be a thing.
• 0:47
  Maybe this is like a consulting line you guys could launch.
• 0:50
  I don't know, but but yeah, it's quite.
• 0:52
  Fun dude. I have.
• 0:54
  No, I would don't depress me like that Consult.
• 0:58
  I'm not a OK, I'm not going to say what I was about to say, but
• 1:02
  I'm I'm I'm, I'm never going to charge someone to spend time
• 1:07
  one-on-one with me. I'm that that's I'm like never
• 1:09
  going to do that. Yeah.
• 1:11
  You're just going to be pick selective about who you give
• 1:13
  one-on-one time with because otherwise, you know, that's a
• 1:16
  lot of one-on-one time. I was going to say I live for
• 1:19
  those moments where people like warm your heart.
• 1:21
  We're my first brand. I used to get these essays from
• 1:24
  people who'd liked our products about how much it changed their
• 1:27
  life, Like Facebook, a long Facebook message.
• 1:29
  My first brand was when, you know, people were building
• 1:32
  brands on Facebook primarily. So we had, like, Facebook groups
• 1:35
  and stuff. Man, when a customer would send
• 1:37
  me a long message, I'd be like, OK, This is why I'm doing it.
• 1:40
  But the negative comments would also cut me when people were
• 1:44
  like, I don't know, you have such a croaky voice and what?
• 1:46
  Oh man, this hurts. Really.
• 1:50
  Yeah, because I like with with my first brand, I had a YouTube
• 1:52
  channel and we had like a few 100,000 subscribers and you
• 1:57
  know, I did tutorial videos on how to use our products, but I
• 2:01
  was nervous. It was like my first time for
• 2:03
  for a lot of the videos, I wasn't even on camera.
• 2:05
  It was just a screen recording. And yeah, so I was a little bit
• 2:08
  croaky. And some people comment and be
• 2:10
  like, oh, you're so croaky. I put fuck I know I need to fix
• 2:13
  this. Dude, I never read the negative.
• 2:17
  I mean like oh sorry, I do read them but I'm like I'm surprised
• 2:21
  that they bother you. I'm genuinely surprised.
• 2:24
  Or maybe I see why it would bother you at the time, but and
• 2:27
  probably doesn't now. Right?
• 2:28
  Like is, is that a fair representation?
• 2:30
  Yeah, yeah, I was. I was 20 at the time.
• 2:33
  Yeah. OK.
• 2:33
  OK. That makes sense.
• 2:34
  Yeah. Yeah, I feel like, I feel like
• 2:39
  you either have to start out with really thick skin or you
• 2:42
  have to have a lot of people say negative things about you to
• 2:45
  kind of get over it. So actually, I wouldn't say I'm
• 2:49
  over it. Just to be clear.
• 2:50
  I, it still bothers me sometimes when people say things to me,
• 2:52
  but I'll kind of sit on sit on it for a few minutes or say
• 2:55
  things about really. Yeah, not.
• 2:57
  I mean, it doesn't bother me as much now as it used to, but like
• 3:02
  not nearly as much. But usually what I will do is I
• 3:06
  sit there, I kind of think about it for a little bit.
• 3:08
  I put myself in their shoes. I'm like, OK, I could see why
• 3:12
  they, well, some things just flat out ignore, but some things
• 3:14
  are actually like the most hurtful ones are the ones that
• 3:18
  are the most descriptive by the people who are closest to you,
• 3:20
  right? Like the closer that person is,
• 3:22
  the more it hurts. And so in that context, I'll try
• 3:24
  to put myself in their shoes and be like, OK, well, I see where
• 3:27
  they're coming from. I also like usually it's not
• 3:31
  something I had control over. It's not something I could have
• 3:33
  done differently or I can look at that and be like, OK, well, I
• 3:36
  maybe I could have done something differently, but I did
• 3:38
  the best that I could in the moment, knowing what I knew.
• 3:41
  So I just kind of like processes a little bit, move on.
• 3:43
  So I've had a lot more of those things in the last year that I
• 3:47
  was used to in the past. That's OK.
• 3:50
  Right. When I watched the old videos
• 3:52
  that I used to make on YouTube, I see the way that I behaved and
• 3:55
  it was very much like when you're young, you're trying to
• 3:57
  find your feet in the world, right?
• 3:59
  You're trying to like act appropriately.
• 4:01
  You don't really know what appropriately is like my
• 4:04
  business got grew very fast and you know, I went from being a
• 4:08
  broke university student to suddenly having a business and
• 4:11
  hiring all my friends very quickly.
• 4:13
  So I I really didn't know what I was doing at all.
• 4:16
  I was just like making up as I was going along.
• 4:18
  But I think, yeah, nowadays you just like, I don't know, you
• 4:21
  contextualize it, right? If it comes from someone close
• 4:23
  to you, you think, OK, this is really good information for me
• 4:27
  to course correct. Maybe, right, if it is actually
• 4:30
  a representation of reality, and if it's not, then it's just
• 4:32
  like, whatever, right? Yeah.
• 4:34
  And I think people also interpret.
• 4:36
  People also, like, interpret things just like they're in
• 4:39
  their own head because I got cancelled on LinkedIn a few
• 4:42
  weeks ago for saying, like, what do you do the book on the
• 4:45
  weekend? Yeah.
• 4:46
  Yeah. Like I said, like I said, like
• 4:47
  the most basic thing that you could say, like, hey, I said
• 4:50
  something about team culture, like, hey, etc.
• 4:52
  Like everyone's kind of spending a few hours on the weekend
• 4:55
  reading a book or doing a course or, you know, pursuing a fitness
• 4:59
  milestone or something like that.
• 5:00
  And people got mad at me. They're like, how can you
• 5:03
  control what your employees do on the weekend?
• 5:05
  I'm like, that's not what I said, but then people just got
• 5:08
  so mad at me in the comments and then I just leaned into it then.
• 5:11
  So I like to, I don't know, maybe I'm the opposite of like
• 5:14
  you, Jack, but I just like, feed them.
• 5:16
  Like, yeah. That's just how it is.
• 5:18
  And then it's just like it got like I think or like 100,000
• 5:21
  Russians or something that that post in 24 hours.
• 5:24
  It was quite crazy. So incredible.
• 5:26
  Oh, my God, this is what my shop is going into accidentally,
• 5:29
  right? Oh my God.
• 5:30
  Oh, my God. Yeah.
• 5:32
  I mean, I, I, I, I'm going to say something that is going to
• 5:34
  be like close to reality. But I do think because I just
• 5:39
  came off of like a coaching, I have like an exec coach and I
• 5:42
  just came off of a coaching session with him.
• 5:44
  And so I'm curious what you guys think of this.
• 5:47
  There's this thing that is changing, like the market is
• 5:50
  changing ridiculously fast right now.
• 5:53
  I mean like, I mean ridiculously fast.
• 5:56
  OK. So I was I think about just our
• 5:59
  company and five months ago we were a single product company,
• 6:04
  OK. We used to sell like funnels
• 6:07
  that you would use that were generated against your paid ads,
• 6:10
  OK, fast for only five months later and and we were primarily
• 6:14
  selling to the mid market. Fine, you fast four or five
• 6:18
  months. We are selling to large
• 6:20
  enterprises. We are selling a platform that
• 6:22
  is based off of like consumer behavior and there are like 5
• 6:26
  different product surfaces that are live with customers.
• 6:29
  OK. In five months, the go to market
• 6:33
  shifted, the number of products shifted, the platform shift.
• 6:36
  I mean it's just like the the speed with which you can do
• 6:38
  things has changed fundamentally.
• 6:41
  So my own experience is saying that this is true.
• 6:44
  But then you like look at Twitter and it's like cloud
• 6:47
  code. Something happened over the
• 6:49
  holidays where people like all of a sudden turned on to it in a
• 6:52
  new way. I guess it's called Open Claw
• 6:56
  now, but the artist formerly known as Claudebot happened in
• 7:00
  the last like 2 weeks and I think I was like, I was, I was,
• 7:04
  I was sharing this and and this gets to now the team question.
• 7:10
  My realization as I was sharing this was your team used to be
• 7:14
  able to operate on a time scale that's found quarters.
• 7:18
  OK. So you would do annual planning,
• 7:20
  you would have goals that were set up quarterly.
• 7:23
  You would like as a salesperson, for example, or as a marketer,
• 7:27
  you would be like, OK, when is the product coming in?
• 7:29
  OK, then I'm going to do this and then I'm going to go sell.
• 7:32
  By the time you do those operations, if you operate on
• 7:35
  that time scale, the world has changed.
• 7:39
  So like you used to be able to like think about a product,
• 7:42
  deploy it and then sell it for three years.
• 7:45
  And now it's like, I mean, I just shared my own story, which
• 7:48
  is like, you know, five months later we have 4 new services
• 7:51
  being used by large enterprises, whatever.
• 7:54
  So clearly a year is too long. OK, a week is like ridiculous.
• 7:59
  I think like expecting people to be able to like operate on a
• 8:02
  week is way too short. But it is something in the
• 8:04
  middle and it looks very different than it used to.
• 8:08
  And, and I wonder do you guys, how do you guys think about this
• 8:12
  And, and do you think it's possible to get your companies
• 8:16
  and everybody in the company to operate at this new pace that is
• 8:21
  clearly getting pulled by the market, right.
• 8:24
  Like I, I don't even think you have a choice anymore.
• 8:26
  Like either you operate at this new pace or you you like fade
• 8:31
  into irrelevancy. Yeah.
• 8:33
  I mean, one thing to support your point there is I don't know
• 8:36
  how we're going to adapt, but we were.
• 8:39
  And I think you and I touched on this Rishabh a little bit.
• 8:41
  I'm like super EOS heavy in how we operate the company.
• 8:45
  So we have like OK, ours, we have the L10 meetings, we do all
• 8:49
  of that. We have biweekly check
• 8:50
  insurance, we have a standards report like it's all, it's all
• 8:52
  pretty streamlined. And then for the last like year,
• 8:56
  it's just like that whole way of operating, like we set a goal of
• 9:00
  some sort. And then like you said, the
• 9:03
  market change like OK, you're building the product hoping for
• 9:06
  PMF, but then the market changes or somebody launches something
• 9:10
  that's faster. So that quarterly pace of
• 9:13
  working and that OK, Rs which are typically by quarter are
• 9:17
  just breaking for us. So I, I don't know what the new
• 9:20
  sort of maybe just have like a few org level goals and
• 9:22
  everybody runs towards those and we hit them and we hit them.
• 9:26
  I don't know what the new sort of AI world EOS looks like, but
• 9:31
  I'm also curious how y'all are operating and changing the way
• 9:34
  you operate to adjust for this. One of the challenges with
• 9:38
  trying to move faster on things is some things just take some
• 9:42
  time to develop too. So like for example, in a new
• 9:44
  product launch, like we. One thing I've always seen with
• 9:47
  new product launches is that it takes some time to educate your
• 9:51
  customers and the market about what you're doing.
• 9:53
  And if you move too fast, you might bail on something before
• 9:59
  you've actually validated it properly.
• 10:01
  But you also don't want to move too slow on these things, right?
• 10:03
  So it's AI find it very challenging.
• 10:05
  Like in the past, I've always thought of, OK, if we're going
• 10:07
  to launch something new, we kind of have to give us six months to
• 10:09
  see, is the market actually going to adopt this thing?
• 10:15
  So I think I, I, I agree that an annual plan is like we've been
• 10:20
  working on annual plans, but I look at them like, OK, this is
• 10:23
  like the revenue objective and like the, the path we think we
• 10:27
  can take to get there, but like the actual activities to make
• 10:30
  that happen that's getting reset literally every quarter, right?
• 10:33
  There's no, and we're, so we're kind of operating quarterly at
• 10:36
  the moment with the idea of like, here's what we're trying
• 10:37
  to do this quarter and then moving on from there.
• 10:41
  So I don't, I don't know what the balance is to me right now
• 10:43
  quarterly planning feels like the right approach in terms of
• 10:46
  like product oriented things, but it might not be fast enough,
• 10:50
  I'm not sure. Yeah, I don't, I don't have any
• 10:52
  experience hiring in the software space, but I feel like
• 10:56
  looking at service based businesses agencies, most full
• 11:00
  time hires are made really inefficiently, right.
• 11:05
  It's sort of like I have a suite of tasks and projects that I
• 11:10
  need regularly repeated and I need a warm body to to do that.
• 11:15
  And if the goal posts move, you know, it's not really the
• 11:21
  employees responsibility to build the plane as they fly it,
• 11:26
  right? It's the the service businesses.
• 11:27
  I don't know what that's like in a software business, but I
• 11:31
  imagine it's pretty similar. But yeah, I don't know.
• 11:35
  When you buy something, you're indirectly employing a bunch of
• 11:39
  people and a bunch of systems to do something for you.
• 11:42
  And when you order an Uber or you hire through Upwork, you're
• 11:45
  directly hiring someone for something to get done with like
• 11:48
  varying levels of complexity. And I feel like those
• 11:51
  employments are extremely efficient, right?
• 11:53
  When you employ through up work or on a project basis or an
• 11:57
  Uber, it's, it can get more efficient, right?
• 12:01
  And I, I don't know, I think that that allows you to be
• 12:05
  agile, but you, you are still the person who's in charge of
• 12:10
  direction at every point, right? Or whoever is that's employed to
• 12:15
  employ people on a project basis is in charge of direction as
• 12:19
  opposed to that employee, right? But yeah, there's some level of
• 12:22
  abstraction. I think it's really interesting
• 12:24
  with full time hires. And it's like, I don't know,
• 12:27
  take the the role of, I don't know, head of go to market or
• 12:29
  head of sales or whatever happens to be, you know, you're
• 12:33
  saying, OK, I need you to do this thing and maybe whatever
• 12:36
  the future of this thing looks like and stay up to date with
• 12:38
  that. But it's not a one and done
• 12:41
  task, right? You're committing to that
• 12:43
  transaction on a regular basis. This is the thing that I don't
• 12:46
  know how it's going to evolve, but I think the way we operate.
• 12:51
  Yeah, Jack, I think that's a really interesting analogy
• 12:54
  around like what is your agreement with the person on the
• 12:58
  team? Like I'll give a good example.
• 13:01
  So inside of inside of my company that what I am asking
• 13:05
  people to like, it used to be like, hey, I need you to do this
• 13:08
  task for this quarter with this goal.
• 13:10
  Like now kind of what I'm asking people to do is like, I need you
• 13:13
  to keep up with the market at this pace.
• 13:16
  So that way when someone interacts with you with this
• 13:18
  type of question, you are informed enough to respond in
• 13:20
  this way, which is a, which is like a totally different thing.
• 13:24
  It has more to do with like what I would describe as like an
• 13:27
  athletic stance. Like you're kind of ready to
• 13:29
  move with the market. And I think that every vendor is
• 13:32
  being asked to do this. So I'll we just turned a bunch
• 13:37
  of SAS tools, OK, Which as a SAS company, I'm like, my God, like
• 13:44
  I like I can actually, sorry, I don't even describe us as a SAS
• 13:47
  company anymore. I describe this as an AI
• 13:49
  platform and I'll I'll get to that in a second, but nice.
• 13:53
  I'm like, we cannot be a SAS company because look at what we
• 13:57
  just did. We turned off of gong.
• 13:59
  We turned off of five Tran. We turned off of zoom info which
• 14:04
  like we're we're turning like at a ridiculous rate.
• 14:09
  You're breaking my heart. Don't check in Stripe.
• 14:12
  And the vendors that we grow with, I expect them to be AI
• 14:17
  plot. Like I'm remembering a
• 14:19
  conversation I had now with my head of finance.
• 14:21
  She also like runs our HR and people OPS and, but like, I
• 14:25
  mean, she just like runs most of our OPS basically.
• 14:28
  And she was asking me about buying green.
• 14:31
  You guys know what greenhouse is?
• 14:32
  It's like a applicant tracking system.
• 14:35
  It's like it's like Ashby or. It's an applicant tracking,
• 14:39
  yeah. Yeah, exactly.
• 14:42
  And I remember her asking me like, hey, which tool do you
• 14:46
  think we should buy? And I was like, buy the one
• 14:49
  that's an agentic platform is like literally how I responded.
• 14:53
  And I finally reflected on my own buying behavior enough to
• 14:59
  realize if I'm telling my team not to buy SAS, so is everyone
• 15:05
  else. Like you like buyers want to buy
• 15:09
  agentic platforms now. And like I was like, I literally
• 15:13
  said, is there an agent? Is it getting better and are
• 15:17
  they investing in it? Like like that was my buying
• 15:20
  criteria. And, and I, I wonder how much of
• 15:25
  that is going to like especially over 2026.
• 15:29
  Like is this the year that really comes to fruition?
• 15:32
  Because I'm like observing my own buying behavior to, to
• 15:36
  become like that. And so like, I feel like, I
• 15:40
  mean, this is like the SAS Operators podcast.
• 15:42
  At some point, we're going to have to rename it because
• 15:45
  there's not going to be any SAS. Yeah.
• 15:47
  And I think how we operate our teams is going to change.
• 15:49
  And I think how we think about our product is going to change.
• 15:52
  You know, I think there's an interesting overlap there with
• 15:56
  the sort of project based gig economy style of Upwork and Uber
• 16:02
  and your full time hires. Because with, you know, agents,
• 16:05
  it's sort of point and shoot. And to some extent maybe it's a
• 16:08
  little bit more complicated than point and shoot if it's planning
• 16:12
  and there's, you know, lots of different steps to achieve the
• 16:15
  goal of whatever it is that the agent is trying to achieve.
• 16:18
  But yeah, most, you know, AI agentic tools right now are text
• 16:23
  or voice based on inputs from the person who is setting
• 16:28
  direction, you know, around whatever idea that they have and
• 16:30
  whatever thing that they want to achieve.
• 16:32
  And it, yeah, it's sort of, it's sort of point and shoot, it's
• 16:36
  sort of like project based, get this thing done right, but
• 16:40
  obviously infinitely more scalable than humans because AI
• 16:43
  doesn't need to sleep and it's a lot smaller.
• 16:45
  Rishab, do you think that buying behaviour translates beyond
• 16:49
  deck? Like do you think you're you're
• 16:50
  looking to? I'm just trying to push back a
• 16:52
  little bit. So if you think you want to buy
• 16:54
  agentic software because you're building agentic software
• 16:57
  yourself versus do you think like like the if you care about
• 17:00
  agentic software, it's like a level 9 out of 10.
• 17:03
  You think like AE COM brand owner like cares the same way
• 17:07
  when they're buying like support software for instance.
• 17:09
  Not yet, but let me ask you back a different question.
• 17:14
  Do you think the answer to that question in 20-30 is yes?
• 17:17
  Yeah, probably. OK.
• 17:18
  So at some point between today and five years from now, the
• 17:22
  answer becomes yes. And so now all we're debating is
• 17:26
  the time scale, right? OK.
• 17:29
  If all we're debating is timescale, I would rather be
• 17:32
  early than late, yeah. Makes sense, yeah.
• 17:36
  I think it's interesting to think about what AI represents
• 17:39
  to the customer or the prosumer today.
• 17:41
  And I, I think that's profit. I think that the, the consumer
• 17:46
  right now used to see software as well, minimized inputs can
• 17:50
  save me time. It's going to do with a certain
• 17:51
  thing, it's going to unlock a certain capacity, it's going to
• 17:54
  be able to do it better or more. And I think that that's the same
• 17:57
  idea that they have with AI now. It's going to minimize inputs
• 18:00
  going to maximize outputs. I'm going to be able to do more
• 18:04
  with less. And that delta represents
• 18:08
  profitability to me or profit. And also in a way that software
• 18:13
  doesn't, but AI does. It represents the cutting edge,
• 18:17
  right? This is the bleeding edge.
• 18:18
  This is early adopting. If I get AI, like what that sort
• 18:23
  of represents is something that's newer and more capable of
• 18:27
  that sort of scale of minimizing inputs and maximizing outputs
• 18:31
  rather than just software, right?
• 18:33
  Because basically these are the same thing, right?
• 18:34
  AI is software, right? But it's a different kind of
• 18:37
  software. And I think to the consumer, it
• 18:39
  represents software at its peak, right?
• 18:42
  Would you guys agree? Well, yeah, I think there's
• 18:46
  what's interesting too. I think that the agentic stuff
• 18:49
  has changed the way in which I think about building things in
• 18:54
  terms of like like I've been playing a lot with replit lately
• 18:58
  just for fun, because I the reality isn't my day job.
• 19:02
  I don't do a ton of it because like it's hard to actually take
• 19:05
  something that I built and translate it to something that's
• 19:07
  functional for my team. So it's been probably a year
• 19:10
  since I maybe not a year, nine months since I use replit.
• 19:13
  It's way better now, which a that's also really interesting.
• 19:17
  It's taken out a lot of the planning and I think what like
• 19:20
  part of what's happening now in if you are the creator of
• 19:24
  software, I think a lot of what's important here is
• 19:27
  actually just like incremental improvements, not so much big
• 19:30
  feature launches, big things that you're you're doing.
• 19:33
  And, and I look at agents in sort of a similar way.
• 19:37
  You basically ask an agent a question and the agent goes off
• 19:40
  and does something and it comes back to you.
• 19:41
  And then that creates a new set of questions and you ask those
• 19:44
  questions. And so there's, there's this
• 19:46
  change, I think that's happening where like there's actually less
• 19:48
  planning. I don't know if that's a good or
• 19:49
  a bad thing, to be honest. It's probably a bad thing in
• 19:53
  some ways, but it's also really good because at the end of the
• 19:56
  day, like part of the, the problem that you end up having
• 19:59
  with planning is that you, you have a lack of information.
• 20:02
  And so I, I kind of been thinking about in the way that
• 20:05
  we build software like we, it's very easy to look at the
• 20:08
  software we have and say there's a very obvious problem in front
• 20:10
  of us that we need to fix. As soon as possible, let's fix
• 20:14
  that thing and then get to the next thing and then fix that
• 20:17
  thing or, or make it better, whatever that is, right?
• 20:20
  So there's kind of this big picture, here's where we want to
• 20:22
  go, but there's also just like that, all the things between now
• 20:26
  and then don't really matter. It's really about how do we get
• 20:29
  to the next step as quickly as possible.
• 20:31
  And that's kind of the way I've been thinking about it.
• 20:34
  And I feel like that kind of mirrors the actual interactions
• 20:38
  that we have with agents too. So maybe it's changing the way
• 20:40
  that I think about planning. And so Rashad, I don't know that
• 20:42
  that fully the guy, I still kind of look at it like I think of
• 20:46
  things on a quarterly basis, like here's what we're trying to
• 20:47
  accomplish this quarter, but the exact steps to get there don't
• 20:50
  really matter. We just take it one step at a
• 20:54
  time and get there as quickly as we can.
• 20:56
  Yeah, I, I, I think that, by the way, the path dependence is
• 21:01
  everything, right? Like it turns out that once you
• 21:05
  do a thing, it will inform the next thing.
• 21:08
  And so I mean, I think we're agreeing because, and I think
• 21:12
  probably I just view the path dependence as significantly more
• 21:16
  important than the statement of the objectives because I'm like,
• 21:21
  so it's probably just a bias of my concord upbringing.
• 21:24
  Like as a scientist, you're like, I don't know what's going
• 21:28
  to happen, So I'm just going to like do these experiments and
• 21:31
  then I'm going to like do the next thing, do the next thing.
• 21:32
  And like, all of a sudden I discover something, right?
• 21:34
  Like that's that's like what the scientific process looks like.
• 21:38
  And so maybe it's just a framing thing because of like how I was
• 21:41
  brought up. But yeah, I totally agree with
• 21:44
  you. Yes.
• 21:44
  I want to make sure you describe you.
• 21:47
  When we caught up in person, you were sharing like what your
• 21:52
  product does and how people use it and where you see gaps and
• 21:59
  it. I want to challenge some of what
• 22:01
  we talked about then in this context.
• 22:05
  Yeah. Let's do it.
• 22:07
  So you want me to tell a little bit about Sarah and what we do?
• 22:11
  Yeah, yeah. Perfect.
• 22:12
  Yeah. So, Sarah, we're a influencer
• 22:15
  operating system, as I like to call it.
• 22:17
  We have brands find influencers, contact them, manage
• 22:21
  relationships, track Bay. Everything you would want to do
• 22:24
  with the creator program, you can do on our platform.
• 22:28
  And then we've sort of differentiated right now in a
• 22:30
  way where we enable them to build programs in a certain way
• 22:34
  that reduce their risk of failure by walking them.
• 22:39
  We've built software in a way that biases for less and less
• 22:43
  risk over time the more they use it and that's what we do.
• 22:46
  So Long story short, that's that On the personal side, it's been
• 22:50
  three years that I've been running.
• 22:53
  I know Jeremiah and I spoke way back one-on-one when I was like
• 22:57
  just launching. So yeah, three years
• 23:01
  bootstrapped, cash flow positive mostly signing to E com.
• 23:06
  So yeah, that's that. But I've been, I've been getting
• 23:09
  the updates for three years. So it's been really cool to see
• 23:11
  you in the progress. I actually love your.
• 23:13
  It's a side note for anybody listening.
• 23:16
  I love the updates because it's, it's cool to see because of the
• 23:20
  way a business like this works. You, you have a, if you're on a
• 23:25
  rocket ship, it's like it has to refuel sometimes too.
• 23:28
  I don't know if that's the best analogy, but it's like you go
• 23:31
  up, go up and then you hit some sort of like little plateau,
• 23:33
  like things don't go, the thing that was working over the last
• 23:36
  six months stops working. And so you got to like find the
• 23:38
  next thing and then go on from there.
• 23:40
  So it's been cool to kind of see that in your updates too.
• 23:45
  Yeah, we and you know this Jeremiah, Yeah, sorry to cut you
• 23:48
  off. And like that just the refuel
• 23:50
  thing is really good because we literally had to refuel
• 23:53
  customers like we we went from zero to 1,000,000 in 14 months.
• 23:57
  And then it was all really small, like really, really small
• 24:02
  SMB brands, like less than a million GMB sort of brands.
• 24:07
  And then we had a self-serve motion.
• 24:10
  We completely changed the business model.
• 24:12
  So we went from self-serve to sales LED.
• 24:15
  So previously you could just start a trial and anybody could
• 24:17
  start using it. Now you have to book a call.
• 24:20
  We tripled our prices, went upmarket.
• 24:23
  We almost had to like over the last 12 months replace the 1st
• 24:27
  $1,000,000 of high churn SMB revenue with low churn mid
• 24:34
  market, more high highly committed customers revenue and
• 24:37
  like entirely changed the business model.
• 24:39
  So like last year was like a refueling year because we
• 24:42
  literally like depleted all the old customers and rebuilt,
• 24:46
  refueled with better customers. So yeah, completely on point.
• 24:50
  Something you mentioned, and I'm curious if this is still true Is
• 24:53
  like the value of the platform is correlated to how effectively
• 25:01
  it's used, which is like very typical of SAS, right?
• 25:04
  It's like we're giving you the tool, but you're the Workman.
• 25:07
  And so here's the tool, Madam or Mr. Workman, please use the
• 25:11
  tool. And and we were sort of debating
• 25:14
  some of like how to think about that.
• 25:16
  And you know, like Taylor has this view of you've got to be
• 25:21
  like a software enabled agencies.
• 25:22
  You have to have like a yeah, it's like they're different
• 25:25
  versions of how to interact with this question.
• 25:28
  I'm curious what Jack thinks too, because I my impression is
• 25:31
  like Jack and Zach are like, you got to build the best UI and UX
• 25:35
  and like people will know how to use like it is a tool that they
• 25:38
  will want to use. That's the impression that I
• 25:40
  got. But I'm curious, Jack, what you
• 25:42
  think? But Yash, when you when you
• 25:46
  think about that problem statement and now that agents
• 25:49
  can do crazy things, does part of you feel like, why don't I
• 25:55
  just give them an agent to operate the platform for them?
• 25:58
  Yeah. We, we're experiment early,
• 26:01
  expect early experiments are ongoing on that.
• 26:04
  We've seen we've used other people's agents, they're not as
• 26:07
  good. We're trying to see if we can do
• 26:09
  something different in that regard.
• 26:11
  But yeah, totally like, so I'll tell you like the pre AI stuff
• 26:15
  that we saw, right? It was like, 'cause we are, I
• 26:17
  think it's, it's also kind of like a spectrum with software.
• 26:20
  We're probably on the worst end of that spectrum or towards the
• 26:23
  worst end because there's so much you could do with an
• 26:27
  influencer platform and there's so many things that could go
• 26:30
  wrong. It's kind of like the best
• 26:32
  analogy really is HubSpot, for instance.
• 26:34
  Like if you buy hubs, HubSpot, that doesn't mean you buy a
• 26:39
  sales team or a sales process or product market fit like there
• 26:42
  could be. There's so many other things
• 26:44
  that have to go right in a business for HubSpot or
• 26:47
  Salesforce to be useful versus if you think about like
• 26:50
  intercom, like customer support software, like it's so to the
• 26:54
  point that there isn't much that you can like do wrong with the
• 26:59
  software. So at that point, I think the UI
• 27:02
  really matters. But for us, it does matter and
• 27:05
  we take our design really seriously.
• 27:08
  But it's like there's so many things you could do wrong that
• 27:12
  you could literally like waste three months doing the wrong
• 27:15
  thing, which is why we have a really serious customer success
• 27:18
  and onboarding process versus somebody like, let's just say an
• 27:21
  intercom board, because it's, it's really like a point
• 27:23
  solution. So yeah.
• 27:25
  And then like totally the working on replacing a lot of
• 27:29
  the CS work with agents right now.
• 27:31
  It does kind of bring up an interesting point though, where
• 27:34
  there's certain types of products, software solutions
• 27:39
  where it's really easy to implement AI and others where
• 27:44
  it's actually really hard. And I think that's, that's been
• 27:48
  one of the things I've been thinking about a lot is how we
• 27:50
  actually, I guess CS function in a software like yours that's
• 27:54
  actually much harder than us than an agent that's answering
• 27:58
  customer support tickets because the, the inputs are so well
• 28:02
  defined, the outcomes are so well defined.
• 28:05
  And then in your context, that's not the case.
• 28:08
  So I think that's a kind of an underrated aspect.
• 28:10
  And actually, Rashad, I'm curious how you think about that
• 28:12
  because I think you like you all are dealing with that too, I'm
• 28:16
  sure. And I'm I'm sure you have a mix
• 28:18
  of like people and agents. I think we're we're like back to
• 28:22
  a time scale question, just like you described your interaction
• 28:26
  with Replit and nine months ago like it was mad like kind of
• 28:33
  built like you could build like a dashboard app and now you can
• 28:36
  build like a real app. And I don't know, man, like I
• 28:44
  think this and and this goes back to how I buy software now,
• 28:49
  which I think Yasha is your point.
• 28:51
  Like I don't know how many people are buying like me yet,
• 28:53
  but at some point everyone will buy like I do.
• 28:58
  I expect every tool that I buy to have an agent that can
• 29:00
  operate it like you said, like HubSpot.
• 29:04
  No, like unacceptable. It is unacceptable that HubSpot
• 29:08
  does not have an agent that can operate HubSpot Salesforce.
• 29:12
  What why is like Benioff hammering agent force and
• 29:15
  hammering his team? He, he knows that there are
• 29:18
  people who expect agents to do the sales and to operate that
• 29:23
  CRM and they're not there yet. But we're going to wake up in
• 29:30
  like 3 years and it's going to be the expectation.
• 29:33
  And because there are teams that are already operating as if it's
• 29:37
  the expected, like my marketing team, if I find out that they
• 29:41
  did anything manually, it's like no bueno conversation.
• 29:48
  I'm like, can you describe the like, I don't want to know about
• 29:51
  the asset or I don't want to know about the outcome.
• 29:53
  I want to hear about the agentic workflow that you implemented to
• 29:57
  create the thing. And I don't care what The thing
• 29:59
  is. And so because what that does is
• 30:03
  now we're on a learning flywheel that is like fundamentally
• 30:07
  different and like our production rate changes and our
• 30:10
  output rate like everything changes.
• 30:12
  And so like, yeah, like, like if you do something manually, at
• 30:18
  least at my company, it's like a yeah, it's like nothing good
• 30:23
  happens basically. Oh, I have so many things that
• 30:26
  have come to mind. The first one, this is a burning
• 30:30
  question. I think all of you could
• 30:32
  probably help me figure this one out.
• 30:34
  Do you guys think there's going to be a collapse in the
• 30:36
  competitive advantage of OpEx? And what I mean by that is like
• 30:40
  before when software was a lot more expensive to build, teams
• 30:43
  almost had a Moat, which is just the ability to build, you know,
• 30:46
  it was a capital requirement. So there was and there still is
• 30:49
  an intuitive software that is massively time costly to use
• 30:52
  where you can arrive at the solution that you're looking for
• 30:56
  when you bought the software. And it's on the other side of
• 30:58
  hundreds of hours spent becoming an expert of the software.
• 31:01
  And maybe this is part of the difference between software and
• 31:04
  AI and the consumer's mind that software can make things
• 31:08
  possible. And it's at its best, maybe it
• 31:10
  automates the solution where AI is autonomous.
• 31:14
  And at its simplest, it solves, you know, it's point and shoot,
• 31:17
  it's our ChatGPT. What should I have for dinner
• 31:18
  tonight? But at its best, it races ahead
• 31:21
  of you. And maybe instead of just
• 31:23
  building a cool thing that solves for a business like like
• 31:27
  a business goal, like it has a business case, it might be
• 31:30
  building something that races ahead and everything else also
• 31:34
  that's required for a successful business like distribution and
• 31:36
  marketing. But yeah, what do you guys
• 31:38
  think? Do you think there's going to be
• 31:39
  a collapse in that piece and maybe other things like
• 31:42
  distribution and marketing and actually building something that
• 31:44
  races ahead become more important?
• 31:46
  Jeremiah, I'm curious what you think because.
• 31:50
  Yeah, I'm not sure exactly. I think, I think I don't know if
• 31:54
  collapse is the right word. I mean, like I was talking about
• 31:58
  using Revlit. I dropped $400.00 in Revlit
• 32:01
  credits in the last week. Like it's not like it's like to
• 32:05
  actually like good agents and it's not good enough actually.
• 32:08
  Like the only reason I can make the things work that I'm trying
• 32:11
  to do is because I'm technical enough and I know enough about
• 32:14
  software to actually get it to where I needed to go.
• 32:17
  And there is no way I would trust this little project.
• 32:19
  So that I have been playing with in the hands of a bunch of
• 32:22
  users. So there's like there's, we're
• 32:24
  still like a couple of steps away from even that.
• 32:27
  But I look at that process and Rep what has, you know, it's
• 32:30
  effectively, I think it's like 3 different agents that are doing
• 32:33
  this thing for you, right? You've got the coder, you've got
• 32:34
  the architect, you've got like the QAI think that that stuff
• 32:38
  and that all of this is subsidized currently by VC
• 32:42
  dollars, right? So when you talk about like the,
• 32:44
  the cost of building these things, I actually don't think
• 32:47
  that stuff's going to be super cheap in the future.
• 32:49
  I think it will be cheaper than humans.
• 32:52
  I think you move faster. I just don't know that like it's
• 32:54
  not going to 0. So if the, if the question is
• 32:58
  like are we, is the cost going to become so low that like it,
• 33:01
  it's basically like infinite production of the stuff.
• 33:04
  I don't know if that's the case. I also think you're dealing with
• 33:08
  the challenge of once somebody learns how to do something by a
• 33:12
  certain way, it's hard to change that way of doing things.
• 33:15
  So like there's a, there's a great, at least currently market
• 33:19
  cap wise, great software business called Constellation
• 33:21
  out of Canada. They literally just buy like
• 33:25
  small Tam software products that have high levels of stickiness.
• 33:31
  They raise the prices, they scale back the teams, they make
• 33:35
  it as profitable as possible to run it.
• 33:37
  And they, they don't and they, they do it in a a way that that
• 33:42
  still keeps the customers happy with the products, right.
• 33:43
  So they're not like trying to run into the ground, but those
• 33:46
  guys, that's their entire business model.
• 33:49
  And, and so there is a legitimate question like, OK, if
• 33:51
  you are a, I don't know, like I'm trying to think of something
• 33:55
  small. You have some sort of software
• 33:58
  for operating a fire station or something.
• 33:59
  And there's only, you know, X number of fire stations in the
• 34:02
  country. The, those people will pay $100
• 34:05
  a month for the software. The Tam is $10 million.
• 34:09
  And this company owns the entire Tam.
• 34:11
  Like is it actually worth trying to go rebuild something and
• 34:14
  compete against that? Maybe, maybe not, depends on how
• 34:17
  complex it is, but probably not. And so I think that's my longer
• 34:22
  answer. Like I don't, I don't think it's
• 34:23
  going to 0, but I think the rate of change will be massive.
• 34:31
  And I think in certain markets, it will feel like it's going to
• 34:34
  0, if that makes sense. Like there's some markets where
• 34:36
  the market, because the Tam is big enough, I think it can, it
• 34:40
  can effectively feel like somebody can come in and just
• 34:44
  like spend enough to to get to the point where they can just
• 34:47
  blow everybody else out of the water.
• 34:50
  I'll give you 2 reactions such. I mean, it's a great question.
• 34:54
  I think it will be less people, but I'm not sure less expensive
• 34:59
  because of this compute issue. And so and I think our hunger
• 35:05
  for compute is going to go up, not down because we're going to
• 35:08
  be able to do even crazier things and, and we're just going
• 35:11
  to want to do that now. I think the place where it's
• 35:16
  not, it's not a difference of opinion to Jeremiah, but it is a
• 35:21
  different opinion. So like I agree on like the fire
• 35:25
  station example, right, But the different opinion, which is also
• 35:31
  true as a consequence of the lower human need, higher agent
• 35:35
  power, things like that. And again, we're experiencing
• 35:39
  this. There are way better known
• 35:41
  companies than us who are experiencing this like Decagon,
• 35:43
  Sierra, Lagora, Harvey. What's happening is the customer
• 35:50
  realizes that the rate at which you can produce product has gone
• 35:54
  way up because all of a sudden they start to get product really
• 35:59
  fast, right? And so businesses like ours, but
• 36:04
  let's use like the ones that are like, it's very clear that this
• 36:08
  is winning like like the Harvey Lagora, whatever, like those
• 36:11
  kinds of companies, a Sierra Decon.
• 36:13
  What's happening is you get pulled in to do a single thing
• 36:17
  like solve customer support tickets.
• 36:20
  But then they're like, actually, can you do this?
• 36:21
  And can you do this? And can you do this?
• 36:23
  Can you do this? Can you do?
• 36:25
  And you just say yes, because it takes you 2 weeks to do.
• 36:28
  Like as an exaggeration, maybe it's more like 6 weeks or 8
• 36:31
  weeks, but it's not a year, OK to, to produce the next product
• 36:35
  because it's on the same foundation and you're just
• 36:39
  doing, it's basically like a implementation work.
• 36:42
  I mean, this real engineering, but it is engineering
• 36:45
  implementation work. And the consequence of that is
• 36:51
  like the fire station example is true.
• 36:53
  And there are like Acvs are going way up.
• 36:58
  Like, I mean, because you're just getting asked to do so many
• 37:01
  things and add so much value on the other side of the spectrum.
• 37:06
  And, and I'm, I can tell you this because we're experiencing
• 37:09
  it. And again, I think it's
• 37:12
  underappreciated. What like exactly what is
• 37:15
  happening. I think like parts of the market
• 37:18
  are obviously getting it right. That's why Decagon raised at 3X
• 37:21
  the value that it raised at six months ago.
• 37:24
  Because people are just like, Oh my God, like every one of these
• 37:28
  100 enterprises. I don't know.
• 37:29
  Did you guys read the fundraising announcement for
• 37:31
  Decagon by any chance? OK.
• 37:35
  The single most interesting thing about the fundraising
• 37:37
  announcement by Decrecon is they signed up 100 enterprises last
• 37:42
  year and that is the only data point that they gave.
• 37:45
  They did not say how much their revenue grew.
• 37:47
  They did not say how much. Why does that matter?
• 37:51
  It matters because all of a sudden people have turned on to
• 37:56
  the fact that if you land the logo and you believe that the
• 38:00
  product velocity is ridiculously high, every one of those 100
• 38:04
  customers in three years or in four year, again, period of time
• 38:09
  doesn't matter. We can debate period of time.
• 38:11
  In X time, we'll be a multimillion dollar customer
• 38:15
  because now that you're in there building agents for the company,
• 38:19
  you will add that much value. And that's just like a total
• 38:24
  reframe, total reframe of of how to think about like markets,
• 38:29
  market opportunity, things like that.
• 38:30
  Do you, Rishabh, do you think this is just like a faster
• 38:33
  version of going multi product in like traditional SaaS?
• 38:37
  Cause traditional SaaS you would hit like 10/15/20 million err,
• 38:41
  you tend to launch another product.
• 38:43
  Do you think this is like a much much faster accelerated way of
• 38:47
  doing the same thing? Increase ACV.
• 38:49
  No, because in traditional SAS you would sell the next product.
• 38:53
  What's happening is that the customer is asking and it's and
• 38:57
  that that is different. It is a different dynamic.
• 39:01
  Like in traditional SAS, the customer expected new products
• 39:06
  and new features to take time. And so they like didn't ask in
• 39:10
  the same way, right. Like, like I, I like, it would
• 39:14
  be crazy for me to go to Rippling and be like, hey, guys,
• 39:17
  could you build this thing? But now I regularly go to like
• 39:22
  startups who I work with and say build this, build this, build it
• 39:24
  because I expect it's easy. And so like, yes, it is a
• 39:28
  version of it, but it is demand driven instead of like, yeah.
• 39:33
  And so it I understand it's a different dynamic.
• 39:36
  Yeah. It's a different dynamic.
• 39:38
  Yeah. You are in the fun category of a
• 39:41
  brand with a brand name that doesn't tell anybody about the
• 39:45
  product at all or similar to format.
• 39:49
  Why Sorrell? Why did you pick Sorrell?
• 39:51
  Because so Sorrell in Sanskrit means simple.
• 39:55
  And when we started, a lot of the space was crowded with a
• 39:59
  very enterprise, boring, hard to use, complicated tools.
• 40:04
  So you were like him? We should call it something
• 40:07
  that's that simple. Often people use.
• 40:09
  Greek names or you know, some, some fancy Latin phrases and so
• 40:12
  on. So I was like, hey, we are
• 40:13
  building from India, so might as well give it like a nice Indian
• 40:16
  Sanskrit name. So that is why Saral and also
• 40:19
  intentionally so because I didn't know if it was going to
• 40:21
  work. So I wanted like a name that
• 40:23
  could be anything. Saral could be ADDC brand,
• 40:26
  right? So that's why that's that's what
• 40:29
  it means. Nice.
• 40:31
  Yeah, Awesome. I like it.
• 40:33
  I think it's nice. It's simple.
• 40:36
  Exactly. Yeah, Very, very meta.
• 40:39
  Guys, I got I got to show you something.
• 40:40
  Did anyone see I know this is very much in the world of ECOM
• 40:44
  SAS, but did anyone see the Parker launch from Alex Cooper?
• 40:50
  Yeah, I'd be really. Yeah.
• 40:51
  I'd be really surprised if you didn't see it where I got to
• 40:53
  show you this. I think he's put because.
• 40:55
  People people were dunking on him because they thought it was
• 40:58
  an icon copycat. Yeah, they were right.
• 41:00
  Isn't that the? Isn't that the primary like like
• 41:04
  debate? I was just like, I don't get it,
• 41:06
  guys. Like I'm I don't even know what
• 41:08
  to say anymore. But we're customers.
• 41:10
  I signed up right away. It's kind of cool.
• 41:12
  I mean, it's really interesting. They personified their AI agent
• 41:17
  so that, you know, it's obviously, you know, inescapably
• 41:20
  a labor replacement. It's supposed to be the, you
• 41:23
  know, an AI creative director that lives in the place that you
• 41:25
  work in Slack. So my understanding, I mean,
• 41:28
  there's a dashboard. I haven't played around with it
• 41:30
  too much. It required a few more inputs
• 41:33
  than I was expecting. I had to upload ACSV of reviews
• 41:36
  and the CSV had to be formatted a particular way.
• 41:38
  And I was like, oh man, we're not there yet.
• 41:40
  But yeah, my understanding is that most of it's in Slack and
• 41:44
  you just kind of chat with it like like it was an employee of
• 41:48
  yours. Very interesting though.
• 41:50
  I think he did a great job with this launch video.
• 41:52
  I mean, 27th of January, it's 3 three days and it's 1.4 million
• 41:56
  views. I'm not sure if you guys saw
• 41:58
  much of what happened with the launch, but like pretty much
• 42:00
  every big AI page on X was posting it.
• 42:04
  I saw one that said imagine going to marketing, going to
• 42:08
  university for marketing for four years just to get replaced
• 42:10
  by a bot. I was like, ah, look, it's it's
• 42:12
  out. So yeah, interesting stuff.
• 42:16
  You know, Alex has been on top of it from the very beginning.
• 42:21
  And I think that's like super underappreciated.
• 42:24
  And it, I mean, I any, he deserves all the success in the
• 42:29
  world. He leaned in in ways in which
• 42:32
  other people were unwilling to like back when the only thing
• 42:36
  ChatGPT could do is give you text, he was using it to like
• 42:41
  come up with scripts and things like that for the creators to
• 42:44
  you. I mean, like he has been pushing
• 42:47
  himself and his organization to use it in its maximal way at
• 42:52
  every stage of like maturity over the three years since
• 42:58
  ChatGPT, the since the ChatGPT moment.
• 43:01
  And, and so it like my big feeling from it is like, I hope,
• 43:09
  first of all, I hope it's successful.
• 43:10
  But even more importantly, he deserves every success in the
• 43:13
  world because he has like tried to push it as much as he could
• 43:18
  at every step of the way. And I think that that's awesome.
• 43:22
  And like, more people should be doing that.
• 43:25
  Yeah, I think you brought up something really important
• 43:26
  though. The fact that people are
• 43:28
  criticizing this for looking just like icon means that he
• 43:30
  didn't get across the fact that it's entirely text based in this
• 43:36
  video. So, you know, it's got 1.4
• 43:37
  million views. People saw this, They got some
• 43:39
  impression, like a first impression that, you know,
• 43:43
  fueled their response. They educated what they were
• 43:45
  going to say in reply to this. And it was oh, looks like
• 43:48
  another icon, looks like another scam.
• 43:51
  But it's just text based, right? It doesn't promise to do
• 43:54
  everything under the sun for you.
• 43:56
  So maybe in this launch that wasn't communicated as well as
• 44:01
  it could have been, right? You know, I think the AI
• 44:04
  creative director hook also is is very similar to the icon.
• 44:08
  I think creative strategist is what he AICM more creative
• 44:11
  strategist is what he called it. So probably also that
• 44:15
  positioning, yeah. Right.
• 44:17
  It's, it's not from the the discerning critiques, It's from
• 44:19
  the people who we're going to get it wrong no matter what you
• 44:22
  do, right? Yeah.
• 44:23
  Yeah, so literally, literally, it just gives you advice.
• 44:28
  Is that the the idea and helps you with like planning and all
• 44:31
  that kind of stuff? It doesn't.
• 44:32
  Produce. It's interesting.
• 44:33
  So I I played around with it a little bit in in Slack because
• 44:37
  my understanding is it lives pretty much entirely in Slack.
• 44:40
  It writes messages I don't want to read right now because
• 44:43
  they're too long, but it seems to connect.
• 44:47
  It's so it's connected to our ad account.
• 44:48
  It knows how much we're spending.
• 44:51
  It's giving fairly specific recommendations based on our ad
• 44:54
  spend, which is interesting. And, and also the reviews.
• 44:58
  The reviews, man, that wasn't so fun to set up.
• 45:00
  Yeah. So it's, yeah, it's, it's
• 45:02
  basically very long, verbose messages about what you could do
• 45:08
  with your creative strategy. Some of the recommendations I
• 45:11
  think are incredible, but they're just like you actually
• 45:14
  have to, you don't you have to read if you, you have to value
• 45:19
  the messages to get to some of those insights.
• 45:22
  OK, Actually real quick, this is a very interesting topic with
• 45:26
  agents, I think because that we've been doing a lot with
• 45:30
  trying to unlock reviews better and a couple of things.
• 45:35
  One, like the default is that it's too verbose, so that's
• 45:39
  challenging. And two, I don't know about you
• 45:42
  all, but I don't read it most of the time.
• 45:44
  Like whatever, like doesn't matter if it's the things we're
• 45:47
  working on, doesn't matter if it's ChatGPT, whatever it is
• 45:49
  like I skim and I kind of, and then I asked the next question.
• 45:54
  I'm not actually like paying enough attention to the detail.
• 45:57
  I have a really hard time actually sitting down and
• 45:59
  reading all of the things that an LLM gives me.
• 46:03
  And I, I'm curious if that's like something you all
• 46:06
  experience too. And I wonder if one downside of
• 46:10
  having LLMS and agents doing everything is that there's so
• 46:14
  much information that we kind of just stop actually consuming it.
• 46:18
  Maybe that's ultimately not a downside because the LLMS are
• 46:21
  are the the agents rather are talking to each other and doing
• 46:24
  all the things for us. But that's fine.
• 46:26
  But but the actual like consumption of the output of
• 46:30
  agents and LMS, I find it really challenging to to basically keep
• 46:33
  up with. Like it's like I I'm trying to
• 46:36
  move faster than what I can actually absorb in those
• 46:39
  contexts. Yeah.
• 46:40
  I think that's you defending your time.
• 46:41
  You know that there's there isn't value in the, you know,
• 46:45
  extra words that you're getting back from it.
• 46:47
  But then maybe there's a flip side here.
• 46:49
  And this is the second thing that I was thinking of when you
• 46:51
  were speaking earlier. Every job are we making our
• 46:54
  teams, especially the marketing teams, retarded.
• 46:57
  My thought here is like, people are using AI to take shortcuts
• 47:03
  and it's like the lowest quality everything.
• 47:06
  It's the lowest quality everything.
• 47:08
  Lowest quality X replies, It's the lowest quality writing.
• 47:10
  It's the lowest quality ads where you just see this
• 47:13
  preference for expedience and automation over doing something
• 47:18
  that's actually good, that's actually worthwhile, right?
• 47:21
  You guys seen Wally where you've got the like?
• 47:23
  I, I, I, I don't know, I'm maybe I'm just more optimistic, but I,
• 47:32
  I think the response of the agent is what you would expect,
• 47:35
  which is, it looks like a normal curve.
• 47:37
  There's this much that's good, There's this much that's mad and
• 47:41
  there's this much that's bad. I think I only worry about the
• 47:45
  leading edge and I just like, don't worry about the rest of it
• 47:50
  because it's, it's, it's supposed to give me a
• 47:52
  statistical output. And so like, if you view its
• 47:55
  response as a normal distribution response, it's
• 47:58
  like, OK, what's on the leading edge?
• 47:59
  What's in the middle? What's the bottom?
• 48:01
  And then it's like, fine, you know, and, and I think that
• 48:05
  that's actually a fine way to interact with, with LLM
• 48:08
  responses. And you can tell it to give you
• 48:12
  back the information in whatever way you want to see it.
• 48:15
  So it's like, I'll give you a good example.
• 48:17
  I use Chai GBT integrated to Slack in our product releases
• 48:21
  channel where our engineers and PMS write the products back and
• 48:25
  what it does and it's super detailed.
• 48:27
  And I'm like go in the product releases channel over the last
• 48:30
  week and summarize it like I'm a go to market salesperson and I
• 48:32
  really only want to know the top three bullet points about what
• 48:35
  we've released, what it can do and how it impacts customers.
• 48:37
  And it does it and the quality is great.
• 48:40
  It actually summarizes correctly, but I did give it a
• 48:45
  specific prompt with a specific point of view of how I want to
• 48:48
  receive information. I do think that that that's that
• 48:53
  is for sure. The hard part about like
• 48:54
  interacting with agents is like you need to know how to tell it
• 48:58
  to communicate with you. Now, to be clear, I also have to
• 49:01
  tell humans in my company how to communicate with me because I
• 49:05
  equally don't read what they send because it also looks like
• 49:09
  a normal distribution. So like I tell people in my
• 49:11
  company, don't slack me. If you want me to take action
• 49:14
  and send it in my e-mail, this is what it needs to look like.
• 49:16
  Don't send me more than three, but like, like, just like I tell
• 49:20
  a human, I have to tell the AI agent.
• 49:21
  And so I, I, I personally, and again, and maybe I'm like more,
• 49:26
  I'm just optimistic, but I find it to be like, like, I find it
• 49:31
  to be fine. Now, I think that there are bad
• 49:33
  agents, right? Like the reply bots on LinkedIn
• 49:36
  and Twitter are like, totally trash because they're always in
• 49:42
  that median of the normal distribution.
• 49:44
  And so they're like, it's like completely uninsightful, you
• 49:48
  know what I mean? It's like, it's like, Oh my God,
• 49:52
  like it. Yeah.
• 49:53
  It like makes you want to cringe, basically.
• 49:55
  And so I think you just have to like, not use that.
• 49:58
  I find I actually find myself replying less to things on
• 50:00
  social media because I realize I sound my like my like normal.
• 50:07
  I'm just going to reply to something without thinking about
• 50:08
  it sounds like AI. So I'm just like, I just don't
• 50:10
  even reply unless I, like, actually have something
• 50:13
  thoughtful to say. I'm not going to say it because
• 50:15
  otherwise it just looks like an AI bot that is like my my
• 50:20
  distribution curve is the same. Yeah.
• 50:22
  As long as you don't put M dashes as your replies, I think
• 50:25
  you're good. Yeah, I was going to say
• 50:28
  sometimes I just want to write like this is great or something
• 50:30
  in in the on Twitter or something.
• 50:32
  I'm like, has there someone's going to say it's AI?
• 50:35
  Maybe not. Yeah.
• 50:38
  I yeah, I like what you're saying, Rashad, because it's,
• 50:40
  you know, this mandate to use AI is less of like a mandate to not
• 50:44
  strive for excellence. Excellence is obviously what you
• 50:47
  want to be doing, but you also want scalability.
• 50:50
  You don't want somebody to be doing something manually that
• 50:52
  can only be done once, even if it is excellent, because that's
• 50:56
  not going to scale. We're looking for excellence at
• 50:59
  scales. I like that.
• 51:01
  I think that at the same time, some sometimes you just got to
• 51:04
  do the hard work. You could do the hard thing, not
• 51:07
  the hard work, but the hard thing.
• 51:08
  Zach would probably disagree, but he just did the hard thing,
• 51:11
  right? I'm not sure if you guys saw
• 51:14
  this video that Zach put together recently, but it's
• 51:17
  incredible. I'll show you it quickly.
• 51:33
  You get the draft right? Like.
• 51:35
  Holy shit. You don't get that out of an AI
• 51:38
  prompt, do you know what I mean? But at the same time, you know.
• 51:42
  Yeah, but you cannot compare. OK, see this is the problem.
• 51:45
  You can't compare the bleeding edge of the normal distribution
• 51:49
  curve to the curve. Like Zach is like a, what is he
• 51:52
  like a top .1% designer? Right.
• 51:56
  Yeah, okay. I mean, I don't know, like,
• 51:59
  yeah, like we're not going to get that yet.
• 52:01
  Do you, do you make like so here's a question for you,
• 52:04
  Rishab. Like if everyone you is is Zach
• 52:07
  a top 1% designer, would he be the same if he was like reliant
• 52:13
  heavily on AI tools in like year one of him doing design?
• 52:18
  Like I think to Jack's point totally, I think there's like do
• 52:22
  you get to a top 0.1% if you have like AI kind of making you
• 52:27
  retarded by being like a crutch to your skills?
• 52:31
  100% you do. 100% you do. Having the best equipment makes
• 52:37
  you the best athlete. You know, I, I, I hate you there
• 52:40
  again, I think that I think you've got this right, Rashab.
• 52:44
  It is about the the top 1%. It is about the top 10%.
• 52:48
  It isn't about all of the dribble, right?
• 52:51
  And as soon as you start looking there, it's like it's not.
• 52:53
  This isn't to say that Zach doesn't use AI.
• 52:55
  Zach uses AI all the time. He's not shipping new products
• 52:59
  in foreplay yet, but he is shipping new features, right?
• 53:02
  Totally. Like I'll give you a simple
• 53:04
  example like just like how great this video is.
• 53:06
  Did you guys see that runway video that the guy from runway
• 53:08
  ML made like with the with the bears look up like fuck, OK, we
• 53:13
  have to show it. I don't know Jack, if it's easy
• 53:15
  for you to find. But I'm not as quick as Jamie
• 53:18
  from JRE, but I can I can try. Hold on, let me see Runway God
• 53:24
  Runway ML bear video. OK, because I want to show the
• 53:29
  equivalent of AI, the AI version of what you just showed me for
• 53:33
  Zach. Anyway, it like there's a bunch
• 53:38
  of people there's like this thing went viral on Twitter.
• 53:42
  No, it's like this. It's this beautiful short film
• 53:46
  made entirely in runway ML about like these like kids and like a
• 53:51
  bear caught like a bear cup costume.
• 53:54
  And you like watch this thing and you're like, Oh my God, like
• 53:57
  this could be like an Oscar nominated short film, OK.
• 54:01
  Like it's that good. And so and so you have to
• 54:05
  compare the 1% to the one percent or the .1 to the .1 or
• 54:08
  the point O1 to the point O1. And and that's more my point.
• 54:11
  It's like, yeah, there's going to be a bunch of people
• 54:15
  producing shitty things. And that's true with the tools
• 54:20
  and without the tools, so. So yes, what's next for Sorrell?
• 54:27
  Are you getting AI involved? We're actually going multi
• 54:30
  product this year, so I'm very, very pumped.
• 54:33
  We're launching a new Rishab. Let's go.
• 54:38
  Rishab approves. Rishab approves.
• 54:40
  Yes, I think we finally have the right and obviously like our
• 54:45
  engineering team is like AI coding all the time.
• 54:48
  So we are shipping really, really fast, faster than ever
• 54:52
  before. But I think still not like still
• 54:54
  not like 5 times faster, but I think we're getting there.
• 54:57
  Like I think it's going to be a process.
• 54:59
  So we're we're launching a new product.
• 55:01
  So that's what's next for me coming up in the next couple of
• 55:04
  months. So really just heads down
• 55:06
  focused on on making that real. So yeah, that's that's top of
• 55:09
  mind. Nice.
• 55:11
  And what's the go to market plan?
• 55:13
  What are you? How are you planning to get
• 55:14
  everyone to use this? So we're actually not on the
• 55:18
  Shopify App Store. This new product will be on the
• 55:22
  Shopify App Store. So I was actually quite
• 55:24
  separate. We do integrate with Shopify
• 55:27
  separately through other apps, but we're not on the App Store.
• 55:30
  This one will be on the App Store.
• 55:32
  So the GTM is actually heavily reliant on on standing out on
• 55:36
  the App Store and they're doing interesting things with our
• 55:39
  pricing model and the way it's sort of structured against the
• 55:42
  first page of the App Store that makes it really compelling to
• 55:46
  use. So yeah, that's all I can say
• 55:48
  for now. And then obviously going to look
• 55:50
  into the, I think I saw the SAS operators with, I forget the
• 55:55
  name of the two, the two nice guys who came up the launch, the
• 55:59
  launch people. Matt and Brian, yes.
• 56:03
  Yes, yeah, yeah. So obviously like going to do
• 56:06
  some sort of a launch video around it and try to get as much
• 56:10
  attention as possible on that. So yeah, all of that is
• 56:13
  currently in the works. Nice, nice.
• 56:16
  I think they're genius. For what it's worth, I think
• 56:18
  they're. Totally.
• 56:19
  I agree. I was so inspired.
• 56:21
  Yeah. What they nailed was they built
• 56:24
  great software that that they can commercialize extremely
• 56:27
  well, right. They've they've figured out what
• 56:30
  is it that people want to buy. How do we go from idea to
• 56:33
  product as quickly as possible and then and then deliver it so
• 56:38
  well, I mean, the only thing I think that is holding them back
• 56:41
  is just how much of it it's actually a service and not
• 56:44
  software. But you know, who knows the
• 56:47
  opinions changing on that one. So maybe they've got it right.
• 56:51
  Yeah, it's a while. All right.
• 56:53
  Awesome. I got a drop, but this was fun.
• 56:56
  Let's go build some agents. This is great.
• 56:58
  Yes, All right. Thanks guys.
• 57:01
  Yes, bye.