Latest / 1 to 100 / E56: Building AI from Bangalore with Yash Chavan

Transcript

E56: Building AI from Bangalore with Yash Chavan · 1 to 100

0:00
I was like looking at my like schedule and sleep patterns for
0:05
this week. And sometimes being ACEO or
0:09
being a founder is hard. Sometimes we're we now and again
0:17
I'm like like, like I woke up this morning.
0:23
This is the I think this is the one of the few times where I
0:27
like really felt it. I had to go to Cincinnati for a
0:29
conference yesterday. And so I like, I like touched
0:35
Denver for 20 minutes on the way there.
0:39
So I like got out of one plane and then got on to the other.
0:42
I was like on Denver airport for literally 20 minutes from plane
0:48
to plane. Ended up with 320 hour days this
0:53
week. 20 hour days are not easy for what it's worth.
0:57
Like I work pretty hard, but a 20 hour day is a little bit
1:01
harder than what I usually do. What's the what's the typical
1:07
day like? I mean 1214 or.
1:11
Yeah, yeah, yeah. It's closer to like, yeah, up to
1:13
14, which is like, that's like a normal fucking cakewalk.
1:18
Yeah. You're including like like time
1:21
and stuff like that in there or just like pure work hours
1:25
worked. So I wake up at 5:00 and then
1:30
I'm usually doing stuff like while getting, I mean AI has
1:34
changed how much you can do parallel processing.
1:36
So I'm usually doing stuff till like 6:30.
1:39
Then I help my kids get ready, drop them off.
1:42
That's usually done by 8 and then I'm like straight working
1:48
8:00 to 5:00 or more like 6 and then with the kids till 8:00 and
1:54
then working again till like 930.
1:57
And so the average day basically is like 5:00 to 9:30 with 3 1/2
2:02
hours taken out for kids stuff. So if you like roughly do the
2:07
math on that, you end up with between 12 to 14, right?
2:10
Because like 5 to 5 is 12. Add another 4 1/2 hours,
2:15
subtract 3 1/2 like time, right? So that's like roughly how the
2:19
math works. And so but yeah, this week I
2:23
because of travel, I had like a few 20 hour days and those are
2:27
hard. Those are hard.
2:28
I can imagine. Is this this is 4 hours of sleep
2:31
a day. You're just like into work and.
2:33
Then yeah, so on two of the days I had four hours of sleep.
2:37
So yesterday was more than so yesterday I had 4-4 hours of
2:41
sleep after a 20 hour day. And yeah, the intensity is also
2:45
high. This is the other things.
2:46
When I was in Cincinnati, I think I was like talking to
2:48
people nonstop, which I mean, the conference was super
2:52
productive, but it's like not low intensity.
2:56
Do you know what I mean? It's not like I was there and
2:58
hanging out and listening to. I didn't go to a single session
3:02
at this conference. I was like talking to people the
3:05
whole time. In fact, I don't even know what
3:07
the sessions were. You know what I mean?
3:11
Like, and so and so I like I, I got home last, I reached home
3:15
last night at 12:30 and then I was like, and I finally went to
3:22
bed at 1:00 and then because I was like catching up on emails
3:26
and stuff and then I, I woke up at a little before 6.
3:30
So anyway, so I was like, man, this is this week really.
3:35
And then right after this I have two sales calls.
3:38
Away you're in the sales call too.
3:40
What's your drug? Drug of choice.
3:42
What? Well, I was saying, what's your
3:44
drug of choice? So you're just doing this
3:45
undrugged? Or is it coffee or Zinz or
3:48
coffee? OK.
3:49
No, well, I only do coffee and Diet Coke.
3:52
I'm a wuss, man. I can't like anything.
3:56
Coffee and Diet Coke. I'm like, I'm too much of A wuss
3:59
to do it. Unfortunately, prioritizing
4:03
life, staying alive and not being addicted to things, I
4:06
think it's a good idea. No, sometimes.
4:08
At which some of my Wall Street friends, You know what I mean.
4:14
I did a 20 hour day on, well, I'm on vacation last week, but
4:19
there was this thing that TikTok wanted us to do.
4:22
So I did a day trip to New York on Tuesday.
4:26
So it was 20 hours, but it was like I, I intentionally said I'm
4:30
just going to relax. So I listened to a book and just
4:33
like slept a little bit on the plane.
4:35
And then it was literally like 12 hour, not quite 12.
4:37
Yeah, close to that. I was in like in Manhattan for
4:41
like 8 hours. But it was intentionally A
4:45
relaxed experience. So it's not the same as what
4:48
you're experiencing, but but yeah, I was like, woke up at
4:51
4:00 in the morning, was in Manhattan by noon local time.
4:55
And then, yeah, but it was, it was supposed to be relaxing, and
5:01
it actually was relatively relaxing.
5:03
That's amazing. That's amazing.
5:06
I'm always think that I have a newborn baby, but I have a more
5:09
relaxed schedule than you guys and I'm getting more sleep.
5:14
Interesting thing. I mean, you know what, the one
5:17
thing I would say from that I'm learning with having a newborn
5:21
baby is context switching is, is pretty hard.
5:24
You know, when you go from like holding the baby and relaxing
5:27
and then to like back into it, there's a little bit of
5:30
friction. It's like, you know, I'm used to
5:31
uninterrupted, like, you know, I'm in the screen from morning
5:36
till night. But but yeah, there's a being a
5:40
dad thing. It breaks up the day a little
5:43
bit. Yeah, it's not.
5:44
It's not your time anymore. I've also had sick kids this
5:46
week, so it's actually been kind of nice being on vacation
5:47
because normally that would just be extremely disruptive.
5:50
But yeah, you know, when you have a kid vomit all over the
5:53
place at 5:00 in the morning, it's like you start today and
5:57
then you can't go back to sleep. Really.
5:59
Just like everything is just chaos at that point.
6:02
Yeah. Yes, what is the, by the way,
6:05
one of my biggest learnings being in Cincinnati and
6:08
Jeremiah, I'm curious your take on this.
6:10
I'm going to say something that's going to sound
6:12
controversial, but there was this tweet or ex post where this
6:18
guy a lot Gil, he's like a famous ambassador said like the
6:23
labs are like a month ahead of SF.
6:25
That's three months ahead of New York.
6:27
That's like 6 months ahead of the rest of the country,
6:30
something like that. By the way, like I have been
6:33
saying this to my team for like months at this point and trying
6:37
to get them to spend some time in SF as a consequence.
6:39
Because I'm like, I'm not sure how does how to explain this to
6:43
you, but when you're in New York, you just are not at the
6:46
same speed or like you're just not at the same point as like
6:51
the majority of SF. Now, I think that tweet got
6:56
interpreted poorly because people understand like all these
7:00
things are normal distributions. There's always going to be
7:02
people on the edges and all this other stuff.
7:04
And so like, it's a comment on the displacement of the
7:07
distribution, obviously. So anyway, let's just assume we
7:12
can all be pragmatic about that and, and, and talk about it in
7:16
that way. The so you know, I was in
7:19
Cincinnati and I was like, dude, I, I think the rest of, I think
7:24
the difference between where thinks the world is and where
7:28
someone in Cincinnati thinks the world is, it's like 2 years
7:32
apart, not nine months apart. I think they're a year into tech
7:37
GPT not and I'm not saying it to be like negative about that.
7:43
I'm just saying that the adoption speed is not that fast.
7:48
I'm surprised that I've been to a couple local things in Denver
7:53
lately, tech things, commerce things, stuff like that.
8:00
There's like a Venn diagram, I guess here of like commerce,
8:04
like D to C leading influencer types and tech leading
8:10
influencer types that are overlapping.
8:13
And you're seeing that, that middle layer is what I see it in
8:18
my day-to-day. And in terms of like, you know,
8:21
my engagement on, on X and, and the people that I talked to.
8:25
And that's like the, it's like the 1% of or maybe not 1%, the
8:30
top 5% of each industry's merging.
8:33
And you've got like this little bubble basically that's way
8:36
ahead of everybody else. And then I talked to other
8:38
people who are in software technology commerce that are
8:42
kind of behind that. And they're, they're just
8:45
starting to use some of these tools and not nearly at the
8:48
level. I mean, I mean, they're at the
8:49
level of many of them just now starting to use things, things
8:53
like ChatGPT and, you know, I'm going to rewrite the copy on my
8:57
website with, with AI. It's like, OK, let's yeah,
9:01
that's like 2 years ago. But I completely agree with you
9:06
or should like there's a just a massive there were so like we're
9:12
especially in the circles that you run in are so far ahead of
9:17
the the the bell curve. But I actually think it's a good
9:21
thing. Ultimately.
9:21
I, I actually think if everything was moving as fast as
9:24
the, the early adopters on the front end are moving, it would
9:29
be a negative for society to move at that speed with these
9:34
kinds of tools just because there's so much potential for
9:37
disruption. And so actually the fact that
9:38
the curve is longer than that is actually a good thing.
9:40
And there's a lot of things that can be learned by the early
9:43
adopters that can then be applied later on in the curve.
9:47
Slowest speed of change, Yeah. What do you guys think about
9:50
this? I mean, one of the problems I
9:52
had with my first brand was that because they were downloadable
9:54
products, like after we had a few 100 sales, we'd be lucky if
9:59
we got to like 1000. Before this happened.
10:01
All these like Russian piracy websites would have my product
10:05
freely downloadable. And then our competitors would
10:08
use those and basically resell them.
10:10
They would change them slightly and, you know, make it look a
10:13
little different and then resell it.
10:15
And I remember thinking like, I did all the hard work.
10:18
I didn't, we were kind of a copycat.
10:20
But I remember thinking, man, you're on the shoulders of all
10:22
this work that I put into this And you just get to like change
10:25
it up a little bit, take it to market and you've got your,
10:27
yourself a brand new product. And I, I kind of see the same
10:30
thing happening with the LMS. You know, like Claude makes an
10:34
advancement, accidentally leaks their entire thing online and
10:37
everyone is caught up to Claude very quickly.
10:40
I, I just, I wonder how much advantage there is.
10:42
And then if we go to the micro, which is adoption of these
10:44
tools, I wonder how much advantage that actually is in
10:48
being a first mover here. If the time to catching up is
10:52
always so much faster, right? Like, so if we zoom out and we
10:54
say, OK, typical software business SAS, right, your Moat
10:59
was the actual ability to build. You needed developers, you need
11:02
people who could actually build the thing.
11:05
And then, you know, obviously that Moat today is not so much
11:07
of A Moat. It's not a Moat at all.
11:10
And people can build something that might have taken you 6
11:12
months a year. It might have been completely
11:14
defensible, like maybe in an afternoon, right?
11:18
This like time to catching up is so fast.
11:20
I remember I started building my app in like September, you know,
11:24
August, September last year right in cursor and it was
11:27
painful, right And now Claude is so good, Codex is so good, right
11:32
That like it would be much, much quicker.
11:33
It wouldn't be quite so much of A slog right as it was then.
11:37
And even then it was incredible, like how much you could do
11:40
right. But today, you know, it's it's
11:41
very different. So I wonder yeah, what what are
11:43
your guys thoughts on this? Like I guess the the first mover
11:45
advantage and the time to catching up really just closing
11:48
down and eventually, you know, people being able to catch up
11:52
very, very quickly and do some incredible things with these
11:55
tools despite not doing it since, you know, Q4 last year.
11:58
My perspective on this is you can only answer that question if
12:04
you know what you're trying to do with the business.
12:06
So let's just take a venture funded business like ours then
12:13
speed and being at the very beginning matters because the
12:18
opportunity cost of not capturing the market share and
12:22
making trying to make the switch cost higher and high.
12:24
So let's just take Harvey and Lagora because it's like the
12:28
easiest way to think about this exact question because what they
12:33
did is basically spend immense amounts of capital to quicken
12:37
the speed of adoption legal at a rate that is unnatural to the
12:43
legal world. And the way they did that is
12:47
they spend money to hire lawyers and then just like literally
12:50
deploy and have like huge COC in the deployment on the first year
12:56
to like acquire every customer. But there's like whatever, some
13:00
like finite number of large firms.
13:02
And so you want to just go and land grab as quickly as
13:04
possible. And then it's always harder to
13:07
switch later. Like no matter how shallow it
13:10
is, it's still annoying to like switch your workflows.
13:14
And so, and the reason is those businesses were set up to like
13:21
the strategy was, hey, let's go and transform this industry, the
13:25
legal industry by giving them agentic tools to run all of
13:31
their workflows. And then you have to go super
13:36
fast right now. A different way of asking your
13:38
question, a different strategy, Jack.
13:41
And I think that's sort of what you're saying is like say you're
13:44
running a business whose job it is to cash flow, right then,
13:50
then the way you would want to do it is you want to want to
13:52
optimize for how much cash flow you're generating.
13:55
And then actually maybe it doesn't matter depending on how
13:58
you're playing that game for that period of time.
14:01
But I think that the the actual thing here is you got to be
14:05
really clear headed about what your business strategy is.
14:09
And where it basically Nets out is if you have shareholders who
14:14
are not the operators of the business, what it means is you
14:17
have a duty to improve the value of your stock over time versus
14:21
generate cash flow. And the consequence of that is,
14:27
which is like, again, a strategy thing, you can decide not to do
14:30
that, right? But as long as you have that and
14:32
you have taken on the responsibility of improving
14:34
enterprise value over time or improving stock price over time,
14:38
you have this terminal value question.
14:39
And then what that leads to is a question of like the land grab
14:43
that's happening right now. And so like, I, I think it has
14:46
to work backward from how you're, how you're set up.
14:51
I also think like the industry and the geography matter a lot.
14:55
And I've been curious to get Yash's view in for a while,
15:00
unlike the speed at which this is happening in India because
15:03
like I can tell you that in the UI mean first of all, NASAF, the
15:07
speed is just insane. Like you would think that AI
15:11
happened a year ago, right? Versus is happening just because
15:17
literally every billboard is AI, right?
15:20
And so it's like, oh, of course this is like already a consensus
15:24
thing. But I'm curious like, yeah, in
15:27
India, how it's playing out And then yeah, how do you, how do
15:32
you think about how much it should impact your operation,
15:34
how much it's impacting the speed with what you do things?
15:37
I think yeah, we're in a unique place because we are operating
15:41
out of India, but then like are sort of also like operating in
15:45
the US, like we're up against somebody like someone of our
15:47
biggest competitors actually is based in San Francisco, right?
15:52
So very interesting part. I think like geography wise,
15:55
India in general, like nowhere near SFI think I'll probably put
16:00
it somewhere near, but like not super lagging either.
16:02
Like there's tons of great AI companies and models.
16:05
I don't know, Rishab, you've probably heard of Serum, right?
16:08
Serum dot AI yeah. Like so there's like a lot of
16:11
these model companies in Bangalore, a lot of new like
16:14
application layer AI companies that are that are like on the
16:17
cutting edge. Some of my friends are founders
16:20
of some of these companies. So yeah, I think like there's a
16:23
lot of movement happening here as well, but yeah, nowhere near
16:26
the nowhere near SF level. I'm sure like again, it's like
16:30
normal distribution. There's probably some people at
16:33
the edges who are who are doing things like that.
16:35
But then yeah, I think like the normal average also like I'm
16:39
also strictly talking about like tech people, right?
16:42
If you look at like the average person, like in working and
16:44
consulting or something, they're still prob like they're using
16:47
ChatGPT to write emails or like a negotiator is with their boss
16:51
or some like very basic two years ago use case of like just
16:55
chatting with the thing. But then in tech, I think it's
16:58
definitely like it's picked up, especially like in the last few
17:02
quarters, I would say. I think that's under appreciated
17:05
and why people got tweaked on acts when a lot made that
17:10
comment. People forget that the average
17:14
person in SF is a tech person. Yeah.
17:17
There's no distinction between an average person in SF and like
17:21
I walk, If I walk on the street right, right now outside and I
17:25
turn left and I say, hey, Matt, how did you set up that Codex
17:29
agentic workflow? And I don't know the person they
17:31
would respond with an answer. Awesome as I mean, as long as
17:36
they're not using the street as a toilet, right?
17:40
That's luckily not yet the average person.
17:43
We're not yet at that. I think that's like.
17:44
Two different sets of people I would say, right?
17:47
Yeah, yeah. But like a lot of these
17:50
companies, I think Anthropic now has a office here in in
17:54
Bangalore. Cursor has an office in
17:56
Bangalore now. So I think a lot of these like
17:58
model companies are also opening up, which is which is a signal,
18:01
right to interpret, yeah. Finally, maybe we're going to
18:04
see the, the truly global workforce, right?
18:08
Well, I mean, I don't know if that's a, that's a good thing
18:10
for the US workforce or, you know, especially in tech.
18:14
But hey, you know, it's a, it's a, it's a good thing for hard
18:17
working people, you know, location agnostic, right?
18:22
Yeah. But I think a lot of the whole
18:24
transformation with of India also has been just like there is
18:27
a push to go from services to software.
18:30
I won't say like a push, but I think like a natural step in
18:33
that direction. Cause a lot of the whole India
18:36
tech from the 90s, early 2000s was just like services
18:40
ecosystem, right? Infosys, TCS, like all of these
18:44
companies. And essentially it was just like
18:46
cheap labour, right? Is like the most crude way to
18:48
put that. But then now I think it's moving
18:51
more towards like software more so than I mean, services is
18:56
still a large chunk of the economy.
18:58
I think like a large chunk with the whole like our startup
19:02
revolution really like started maybe in like early 20 tens.
19:06
So maybe like 15 years ago from now or some of like the biggest
19:10
companies in India who are now like public companies started
19:13
and were startups. I think like since then, a lot
19:15
of it has moved from like people who would typically be at one of
19:19
these services companies now work at a startup, right?
19:22
Like the makeup of that person. So like the fresh college grad,
19:26
like when they would like 15 years ago go join Infosys and
19:30
that was like their dream job. Now they want to work at a
19:33
startup in Bangalore or in Mumbai, right?
19:35
So like I think a lot of this changing and moving.
19:38
Incredible. Amazing.
19:39
Yeah. And the sentiment, like
19:41
excitement, like from the people that you speak to.
19:43
And not just like, obviously your friends that are founders,
19:46
but like the people who are, you know, getting started, excited
19:49
to speak to you, excited to hear your thoughts.
19:51
Are they excited or? Yeah, there's so much optimism.
19:55
Yeah, I think like everyone's as bullish as we've ever been.
20:01
So yeah, about the space, about the ecosystem, about the
20:04
country, I think you name it. I mean, obviously like there's
20:07
problems, but people are way more optimistic than even like
20:11
five years ago, I would say. So, yeah, especially with AI
20:15
now, because a lot of the things like if you can, because a lot
20:19
of the delta used to be how good is your code?
20:22
But then if now the agents are just coding.
20:25
Now it's really about like strategy and all the other
20:28
things. So I think like it's a, it's a
20:30
great equalizer in some some ways.
20:32
How do you? Think about growth rate and like
20:36
the product launch that you just did and launch schedule and and
20:39
things like that. Like one of the biggest shifts
20:43
to tech marketing in the US is product velocity is the number
20:50
one reason why, at least in the enterprise, why a company works
20:54
with another company. It's not even actual like #2 is
20:58
so far behind that it's almost, I don't even know what it is
21:02
anymore, actually. And I think it's because we are
21:05
in a shifting landscape. And so when you're in a shifting
21:08
landscape, your buyer has to have confidence in your speed.
21:13
Yeah, otherwise it's like I'm going to buy your tool and I'm
21:16
going to wake up in six months and I have to go buy something
21:18
else because like the thing has just changed like that what I
21:23
need has changed and what's possible has changed.
21:27
How do you think about it for you and yeah, what what is the
21:30
general and amongst your friends in India Like I, I, I know here
21:35
we that's all we think about is, is shipping velocity.
21:39
Yeah, I I won't say that's all we think about.
21:41
Also like, I think a lot of like the ecosystem that I'm in still
21:45
is not enterprisey. It's still like mid market SMB
21:51
and they still care about velocity, but maybe not as much.
21:55
So we do get asked questions about how quickly are you
21:57
shipping? What was, what did you ship last
21:59
month? What does we sometimes have to
22:00
like show our change log on sales calls saying that we like
22:03
here, here's all the things that we shipped last month, here's
22:06
what's on the road map, here's what's coming, blah, blah, blah.
22:09
But then definitely not like I won't I won't call it number
22:12
one. I think it's definitely top, top
22:14
three, top five, though, depending on depending on the
22:17
prospect that we're speaking with.
22:18
And that's, that's definitely changed from even like a year
22:21
ago, people wouldn't really care about that.
22:24
But now they care about like, hey, do you have an MCP?
22:26
Do you connect to like does your app connect to cloud, things
22:29
like that, Which is also part of the reason why we shifted and
22:33
launched like a new product, because the old product was just
22:36
like legacy code written by an agency and less tech that to
22:40
turn that into an MCP would have been like a lot of work, grant
22:44
work. So just like having a new clean
22:47
software product that's just like built in the right way,
22:50
that's almost AI first in some sense also made sense from that
22:54
perspective, which is why we launched, did that whole launch
22:56
and then spun that out. And there's like other business
22:59
reasons for that, but then there's that's definitely one of
23:02
them. Jeremiah, what's your sense on
23:05
the importance of like shipping velocity right now?
23:09
Yeah, it's interesting because I think the IT kind of goes back
23:12
to what you were talking about with earlier with what is the
23:17
business objective and goals. So I think you're talking about
23:21
like AVC backed company. It's all about momentum.
23:23
Momentum matters so much because you're trying to get to the next
23:26
step effectively. You've got like this series of
23:28
steps that you're trying to, to get to, as you know, in the
23:33
context that we operate in a holding company, it's really
23:37
just about like sustained profitability and profitable
23:39
growth over times. Growing EBITDA matters more than
23:42
anything else. And so that's a.
23:45
So anyway, all that to say, I think like the, the, you have to
23:48
understand the goals of the business and what the
23:52
shareholders want. And then in that context, then
23:55
you have a better understanding of, of what to do.
23:58
So I think, I do think shipping velocity is really important,
24:02
but I think there's a couple of things with that too. 1 witching
24:05
cost is still a real thing. So you don't necessarily have
24:07
to, to be the absolute leader in features.
24:13
Somebody else could ship faster, but there's still like a cost of
24:16
getting them to switch off of whatever the the tool is that
24:19
they're using. I mean, we experienced that
24:21
early on actually, we were way ahead of the incumbents.
24:25
We kind of came in a little bit later than a couple of
24:26
businesses. We were way ahead of the head of
24:27
the incumbents for like a year. But it actually was really hard
24:32
to get people to switch off of those tools that that took time.
24:36
That was a process. And so now we've kind of, I
24:40
would say we're, you know, probably the first company that
24:42
most people think of in this space on the the no side for
24:45
surveys, but that took years to get there, even though product
24:51
wise we were actually there long before the, the reputation
24:54
caught up. And then I think the other thing
24:56
with software that's really important is I think this is
24:59
going to become more important over the next couple of years is
25:01
also the stability aspect of it, however you want to phrase that.
25:04
But basically, how secure is your product?
25:07
How well does it work? Can somebody actually trust that
25:10
it's going to deliver on, on what's promised?
25:13
And I, and the reason why I think that's going to become
25:15
more relevant is that there's just, we're starting to see more
25:20
hacking happening. We're seeing like, you know,
25:23
there, it is really easy to launch something new.
25:27
You have to be very careful about the, the underlying
25:30
performance, stability, all those kinds of things and, and
25:32
taking a piece of software and scaling it rapidly, you end up
25:35
running into issues that you weren't expecting to, to pop up.
25:38
So I think there's a balance there, I guess is what I'm
25:40
trying to say. And ultimately then it comes
25:42
down to what it, what risk profile do you want to have in
25:45
your business? What are your objectives in your
25:46
business? And then shipping speed, I think
25:49
depends a little bit on on how you're thinking about those
25:51
things. How much, how much are you
25:52
willing to invest in shipping faster, all those kinds of
25:55
things. I'm so to be fair, I'm not
25:58
either. I'm actually very curious.
25:59
That's how this turns out. The in the past that was big
26:05
thing, but there's kind of a shift that's happening in
26:09
expectations. I mean clot is down almost every
26:11
day. It feels like yeah, there's like
26:15
there's all these things where I do wonder if there's a certain
26:20
rate of shipping at which people don't care about stability.
26:23
I just wonder if we get tired of that after a certain point in
26:26
time. Like if you if you have a piece
26:28
of software that you feel like matters and that you need to
26:32
rely on, do you get to a certain point where six months of having
26:36
this thing be out for, you know, half a day every week?
26:41
Do you sit down and say, I just can't take this anymore?
26:44
I just need something that works.
26:46
So that's that's where I'm coming from there.
26:47
I think that that may be everything is kind of a pendulum
26:51
and swings right. We've swung really far toward
26:54
you don't think so. You don't think there are
26:56
software where so so a lot of software?
27:04
Cloudflare Cloudflare can't go down OK because Cloudflare
27:07
Cloudflare the tool availability.
27:09
A good. So I'll just use surveys.
27:11
So when we are we are showing a survey on an order confirmation
27:15
screen after somebody places a purchase that is a singular
27:21
moment in time that you have the ability to actually have that
27:24
connection with the consumer. And so and there's there's some
27:28
exceptions or some percentage of people that come back to a thank
27:31
you page in the future or whatever.
27:32
But largely speaking you have one chance.
27:35
And so we care a lot about making sure that we do not have
27:38
downtime in that experience. And of course that happens a
27:41
little bit. Shopify goes offline, right.
27:42
But we actually had better availability of our post
27:45
purchase surveys than Shopify check out does.
27:48
And that's kind of a thing that we we look at and I think
27:51
matters. So you could go vibe code your
27:53
own piece of software and if it stops working, you might not
27:57
know for a week. And so in that time period, you
27:59
missed all of that data, right? And for us being the builders of
28:03
this thing, if we go and just ship things without any, I
28:07
shouldn't say without any consideration, but there's a
28:08
speed of shipping in which we are risking that experience
28:12
breaking. And if we break it too many
28:14
times, I believe our customers would lose trust in our ability
28:19
to actually do the core thing that we say that we're doing.
28:20
I don't know if that makes sense, but that's I think
28:22
there's like certain not every software maybe has that problem,
28:25
but I think a lot of pieces of software, the whole point of
28:27
software is that something is happening automatically for you
28:30
that you don't have to consider. And so if it's I need this thing
28:33
to do my workflow and I'm participating in that, I think
28:36
that's a different story than I need this thing to just work.
28:39
And in that context, I need availability to matter as well.
28:45
Yeah, I think it's a simple question is, is tool
28:47
availability directly related to output?
28:50
But the answer is not yes for all agentic tools.
28:55
And that's exactly the insight, the the reason why it doesn't
28:58
matter for Claude is it kind of doesn't matter like the
29:04
availability of cloud does not actually have a direct one to
29:09
one relationship with your output of you using cloud.
29:13
No, it's true. Yeah.
29:14
And I guess what I was saying the, I think the stability,
29:18
availability, whatever does matter in software and shipping
29:20
speed and those things are inversely correlated.
29:25
I mean, there's a, there's a for some type of.
29:27
Software, but not for a genetic workflow software, right?
29:31
No, Yeah, yeah, I agree with that.
29:33
Yeah, yeah. Yeah.
29:34
Oh, I wish I knew what you guys were talking about.
29:35
I don't know. Claude went down yesterday and I
29:38
was like fuck, Codex is kind of performing better than Claude
29:41
recently and now this happens. So I switched.
29:44
I was like, OK, everything. So when I start a new project to
29:46
start a new thing, new terminal tab, it launches Claude in my
29:50
project folder. Now it doesn't, it launches
29:52
Codex because Codex has been outperforming for a little bit.
29:55
So I'm like, OK, and then Claude went down.
29:57
I'm like, how do I, I can't use Claude right now.
30:00
So don't get me wrong, I haven't cancelled my Claude
30:02
subscription. I'm still paying for Claude.
30:03
I'll probably switch back to Claude if they pull ahead again.
30:06
But like right now everything's Codex like from default and if
30:09
they if they were offline for enough time I would cancel my
30:13
quote subscription. I think once you start to get
30:16
into like the bigger enterprise stuff down the the road, like
30:19
let's say five years from now, I think that stability,
30:21
reliability will matter more for those tools than it does today.
30:24
Because you're going to get into that world of like, I'm picking
30:26
this tool, I'm building all of my infrastructure around this
30:29
tool. And if that stops working, our
30:31
business stops working. But I should say that I'm like,
30:33
I'm one, one person, you know, Claude goes down for 5 minutes.
30:36
Not disruption. Yeah.
30:38
Think about human availability. I'm like so poorly available.
30:43
I don't answer my slacks. I don't I'm.
30:45
Only 20 hours a day. Fucking tired.
30:48
I'm like. And you guys are talking about
30:50
software availability like it matters.
30:52
We pay like so much more money for humans who are poorly
30:55
available than software. But, but isn't that point part
30:58
of the point of software, right? Like if your Shopify checkout
31:01
goes down for four hours and you're a brand that does
31:04
$600,000 a day, that's $100,000 you just lost because your
31:08
software went down for four. Yeah, I'm not disagreeing with
31:10
you on on that type of I'm I'm simply remind like I don't want
31:13
to lose sight of this idea that there is now a new form of
31:18
software which is a genetic workflows that does not have the
31:21
same property as the previous generation of software.
31:24
Where I can give it a task and it can do it when it does it and
31:29
it'll report back to me when it reports back to me, just like a
31:32
human does. And in the middle, if it was
31:34
down for a couple of hours, it makes no difference to my life.
31:39
Yeah, I hate it. Yeah, that makes sense.
31:41
It's the same thing with the token cost thing.
31:43
It's like, well, this is infinitely cheaper than a
31:44
developer or XY, any individual person.
31:47
I was like, well, I don't know how much I care about token cost
31:50
going down. I, I really don't at this point.
31:53
But I, again, I don't have 100 people on my team using it.
31:56
It's just me with like two $200 subscriptions.
31:59
I'm like yeah fuck it, use as many tokens as you can.
32:02
Wait guys, actually can I take this opportunity to shift the
32:05
convo to this new thing that I want to talk about because I've
32:07
like fucking gone crazy with it. Have any of you guys used the
32:11
codecs like cloud hosted agents like that you can like create
32:15
your own agents and it has it like it's a very nice UI.
32:19
So like it'll automatically set up a memory file for you.
32:21
It'll like it, it shows you where all the connectors are.
32:25
And so like it's an it's a real agent workflow.
32:27
So it will like improve as you give it feedback basically.
32:31
OK, so anyways, talking about availability, which is why I
32:34
fucking love this my tea. I've become what is the number
32:38
one problem with the CEO? You can't reach them when you
32:41
need something. Exactly.
32:42
OK, so I made a Risha bot that responds to my team's questions
32:48
the way that I would. So I gave it a bunch of training
32:51
material, which is like my previous ways I respond things
32:54
like that, blah, blah, blah. Because I have big downtime.
32:57
You think Clot has downtime? I'm just like not available 99%
33:01
of the time, right? And so I create this bot and I
33:06
tell it to use this memory file such that when I respond to it
33:12
and I invoke it and I tell it here's how I would have
33:14
responded, it continues to learn.
33:17
So every interaction it has, it gets better, right?
33:21
Because I'm giving, I'm giving it constant feedback, but it is
33:25
learning better than a human because it just directly takes
33:28
it and puts it in the memory file.
33:29
Like I don't have to remind multiple times, right?
33:31
It puts it in the memory file. Now it's in the memory file and
33:34
that's it. To distinct is astonishing,
33:38
astonishing. You guys should build one of
33:41
these spots, go to the codecs like whatever, whatever they
33:44
call these like cloud hosted agents thing and create a bot
33:49
that is basically a version of I created a public channel.
33:53
So the company, everybody in the company can ask my bot for
33:57
feedback and everyone else can see how my bot gives feedback
34:02
and then can see how I coached the bot and the person on how to
34:06
think about it differently. And the rate at which context
34:09
spread has started to happen has gone up like 100 X.
34:14
It is astonishing. It is astonishing.
34:16
And I was talking to someone yesterday in Cincinnati who gave
34:22
me this idea to do the same thing for each function.
34:26
So because like there's this, there's this whole thing right
34:28
now of like you want to flatten the organization.
34:31
How do you manage like 80 people, blah, blah, blah.
34:34
You create robot managers. And what those robot managers do
34:39
is you like create like an account.
34:40
Let's just take account management for example.
34:42
You take an account management channel.
34:44
In that Channel, the account managers can say, hey, here's
34:48
the situation or like here's the customer thread.
34:50
Can you go in and give me a suggestion for how to reply?
34:54
And because it is programmed to use all of the like best
35:00
practices, learn what works and what doesn't with the customer
35:03
and why, and it'll command it directly to memory and there's
35:06
no training time other than the live feedback that it's getting,
35:11
it becomes a highly effective functional coach to every
35:15
function in your company. There's a guy out there who does
35:19
AI customer service. It's going to be I'll meet, I'll
35:21
meet from Rich panel and he's going to be like, I have the
35:24
next thing I have the next thing I'm going to build these AI
35:27
managers and these AICEOS, right?
35:30
Because they have all the infra, right?
35:31
They've done this before. Dude, yeah, I mean, yeah, I
35:35
really recommend. And Jeremiah, there's like a
35:38
couple of convos you and I have had about like, because we
35:41
talked about like, how do we think about the team and
35:43
managing the team and bringing the team along?
35:46
Dude, you should have a bot do the things that you're doing
35:48
instead. Could the codex agent could even
35:51
infer in the channels like when it should be invoked if you ask
35:57
it to? I don't want to say too much
36:01
because the people that I'm working on this with are like
36:05
really cool and I don't want to just name drop for for no
36:08
reason, but we're working on this thing inspired by what you
36:13
brought up before Rashad, this self improving feedback loop for
36:18
a specific purpose for like e-commerce businesses.
36:22
And it learns, right? It gets better.
36:24
It doesn't redo tasks and it's it, I think a lot of it, all of
36:29
it, all of that component of not just doing the job, but getting
36:32
better at doing the job over time was inspired by the fact
36:35
that you said you, you mentioned bringing up these feedback loops
36:39
that self educate, right? When it does something wrong,
36:42
it's not just self correcting for that individual.
36:45
It's self correcting for the full function of what that thing
36:48
does. So it can all get better.
36:50
And I think that there's, there's something in that for me
36:53
that's really exciting because you're not just the first person
36:57
to have built something that does something very, very cool.
37:00
But if you're the first person to do it very, very well and
37:03
then it gets adoption and you have 1020100200 users using it
37:09
and contributing to it, you have something that is just so
37:12
incredibly robust and, and always getting better.
37:17
I again, I don't know how defensible that is, but it's,
37:19
it's, it's cool that it can, it can do something cooling and get
37:23
better at it over time, right? How does the team feel about
37:26
that? Are they comfortable with the
37:28
idea of talking to a bot CEO? Dude it's way better than the
37:33
real thing, especially for the team.
37:37
Well, you've solved something interesting as well.
37:39
As there's no procrastination, you can't build well, Rashad
37:43
didn't respond me in to me in time and make that the reason
37:47
why you didn't get your job done right.
37:49
It's now like you got your response, go ahead.
37:51
No blockers, right? And if you can do that in
37:53
multiple layers, right, then that requirement layer of, oh,
37:57
well, I, I, I need to do this, but I have these questions that
38:01
maybe should have already been answered before, right?
38:03
But I have to have them in my bandwidth, in my human context
38:07
in order to continue the task and get the job done.
38:10
You, you solved that. I forget actually having like a
38:13
suitable or better replacement for you to answer these
38:15
questions. These people are now no longer
38:17
blocked by needing a conversation with you, right.
38:20
And that would be my main, that the main thing that I, I would
38:23
benefit from because man, the last thing I want is Adm with a
38:27
question because, you know, I mean, I answer the question once
38:30
for one person and another person messages me the same
38:32
question I have to answer again. So it's a nightmare.
38:36
I don't, I don't respond to any DMS anymore.
38:38
This is just not. It's not scalable.
38:42
Unless they're paying me. Unless they're paying me, then I
38:44
have to respond. Yeah, yeah.
38:46
Do you feel like that's, do you feel like yeah.
38:49
Do you feel like culturally it would be accepted if you said
38:52
like, talk to the spot, don't talk to me?
38:54
I think for the size that we're at, is there like 20 people?
38:58
I think they expect me to be available.
39:00
I think if we were like 50 or 60 people, it may be like more
39:05
acceptable. I think it's, I mean like if you
39:07
take it to the extreme, if you're like 3 people in the
39:10
company and you're like, hey, talk to the bot, feels a little
39:14
bit like, why can't you just answer it right?
39:17
So I think like we're, we're maybe like a little bit small.
39:19
But I think like for trans I, I see Jack's point.
39:21
I think for like more transactional questions where
39:24
it's like, yeah, I've answered this question about Julian Times
39:27
across like 6 different people already.
39:31
I think it makes sense if the bot just has access to all of my
39:35
chats and can just pull different threads.
39:37
But I don't like, what does that mean in terms of privacy?
39:39
Like, does it just like learn from everything I say to
39:42
everyone? And then like, can it just like
39:44
read feedback and people just ask it for like, hey, what was
39:47
the feedback that Yash gave the marketing person?
39:50
And just like just reads it out because it has access to the
39:53
data? I don't know.
39:53
Yeah, I don't know how that works.
39:55
I mean, these things are easy. I think there's a beautiful
39:57
irony in the name in the name Slack, right?
40:01
Because if you're responding to Slack messages all day, you're
40:03
not getting work done. Maybe you're permitting work to
40:05
get done. But man, the amount of dumb
40:08
questions you would have to answer all day long if you were
40:11
just sitting in Slack facilitating DMS, right?
40:13
Like if you actually replied to everyone who shot you as a silly
40:16
message. I remember Rashad when your name
40:18
changed to like. Read this document before you
40:21
say. Like you should say today, if
40:26
you encouraged it by like replying to everyone, man that's
40:28
that'd be a full time job. Just replying to people on slack
40:31
just slacking. I get no work done right.
40:34
Yeah. Yeah, I mean, my current Slack
40:37
name is Risha. Please e-mail action items.
40:39
But I think I, yes, I think I would push you a little bit on
40:43
this. I think I don't know that people
40:48
actually want to talk to me or us.
40:54
Like I think they want to get good work done.
40:59
And I think that if that requires talking to someone.
41:01
In fact, I think I I've noticed people are asking more questions
41:06
with more detail to my bot than they would ever ask me.
41:10
And I think you should. I can see that.
41:12
You should, you should check this because what happens is
41:15
people worry about, oh, it has to be an important enough
41:18
question for me to ask Rishabh. And there's all these other
41:21
shenanigans that show up in people's psychology.
41:24
So technology reduces the barrier by increasing
41:28
availability, right? Roughly, very roughly, right.
41:33
Like you make the thing that was scarce much more available.
41:36
So if CEO time is scarce, you got to increase that massively.
41:41
And once you do that, the Jevons paradox thing happens and it
41:46
just increases the amount that you can you can be connected to.
41:51
And so actually people feel more connected to you, not less
41:54
connected. Yeah, I like that.
41:57
Yeah, I'm going to check it out. I guess there's a world,
41:59
especially in Slack, where your bot can use your handle and use
42:03
your user, right? So there could be like no
42:07
separation. But I mean, I guess that's not,
42:09
it's not super ideal. You just got to let people know,
42:11
hey, every now and then it will be me.
42:13
Yeah, that's what open Claw is. I don't.
42:15
I don't like that everybody can have their own opinion.
42:18
I like the idea that the bot is its own entity.
42:23
But anyway, that's just my view. All right, Yosh, let's talk
42:26
about the new product. You mentioned there was like a
42:30
like a wider business case for the product.
42:31
But dude, this launch video so good.
42:35
Have you guys seen it? Yeah, Jeremiah.
42:37
No, no, Yosh, can you pull it out?
42:40
Yeah, let's launch it. Did You Know 70% of the impact
42:45
of your influencer affiliates is not showing up on your affiliate
42:48
dashboard right now? That's because your platform
42:49
only tracks the last key, which means that one creator who
42:51
introduced someone to your brand, or the other one who
42:53
nurtured someone over 5 fours, or that person whose DMS
42:55
convinced somebody to buy, none of that comes.
42:57
They're measuring a fraction of your own success as an
42:58
influencer affiliate manager. Your program is driving impact,
43:01
70% of which is not being seen by leadership on dashboards
43:03
right now. That means, in other words, over
43:05
3 1/2 days of your own work does not show up in dashboards.
43:09
That's pretty bad. Affiliate marketing is riddled
43:11
with these issues. Nobody is talking about your
43:13
links and codes. I don't get it.
43:14
As we speak every time a customer googles the code at
43:16
checkout you end up paying Commission on a sale that was
43:18
anyway going to happen. Your best creator has posted
43:21
about you 5 times last month for a last week affiliate who drops
43:23
the code at the right time get all the credit for the sale.
43:26
Every quarter your CFO asks is this spend incremental?
43:29
Are these sales even real? And I know that your gut says
43:31
yes or your dashboard says no. You're constantly following up
43:33
with influencers, chasing for W nines, building invoices,
43:36
following up for content and doing all of this boring work
43:38
and is not marketing. This is not why you became a
43:41
marketer. Meanwhile, you're a free
43:42
platform. While they should have solved
43:43
this years ago, they're all stuck on last click attribution
43:46
and dashboards that look like they were built in 2015.
43:48
You deserve much better and that's what I've been to you
43:50
today. Over the last three years that's
43:52
all we've had. e-commerce brands build influencer and affiliate
43:54
programs driving over $59,000,000 in revenue and we
43:57
kept seeing the same problems. And that is why we built Sathi,
43:59
the modern ecom affiliate platform for the modern world of
44:02
influence and agentic commerce. It transforms affiliate
44:04
marketing in three magical ways. We have cookieless multi touch
44:07
tracking, which is a breakthrough because it works
44:09
even when people have cookies disabled or ad blockers enabled.
44:12
And it's multi touch, meaning the creator, the growth
44:14
awareness on TikTok still gets credited for the sale when it
44:16
happens weeks later. Next, there is powerful AI fraud
44:19
handling which keeps your program safe and honest.
44:21
Sathi flags the basics like self referrals or leak codes, but it
44:23
goes beyond and tracks if someone's hijacking paid traffic
44:26
and spots fishy auto patterns. And finally, we've created a
44:28
beautiful affiliate experience that's gamified and makes
44:31
affiliates wanting to post more and make sales for you without
44:34
you following up with them. All of that plus there are AI
44:37
automations that take care of all the boring taxes, payments,
44:39
follow-ups, reporting for you so you can focus on being a
44:42
marketer. Safi is your companion that
44:44
takes care of everything else so you can be a marketer.
44:47
Feedback from our early users has been unreal.
44:50
Me and the team can't wait to show you the hidden potential in
44:52
your affiliate program with Safi.
44:54
Welcome to the new world of affiliate marketing.
44:56
I will see you on the other side.
45:02
Wait a second. Wait, did you put a copyright
45:04
copyrighted song right at the end of your launch video?
45:07
I did, yeah. LinkedIn has very loose
45:09
copyright laws I found. So, you know, I figured might as
45:12
well. Very cool song and sounded.
45:16
Yeah, that's awesome. Yeah, very cool.
45:18
And how's it been since the launch?
45:21
I've been crazy. Like yeah, not 20 hours.
45:23
So definitely like 15-16 hour days on this side, lots of demos
45:28
and we've done an interesting like wait list strategy.
45:31
So we almost like if you guys remember the Superhuman launch
45:34
where you had to like get on the wait list to get in and then you
45:37
could be like bumped up if somebody used your referral code
45:41
to like also sign up and things like that.
45:43
So we're executing that. So we're like curating how we're
45:46
on boarding brands and that's actually going well for us.
45:49
So like, yeah, lots of demos and like I'm obviously like
45:52
personally involved in everything because it's like
45:54
brand new and they're like figuring everything out.
45:56
So just in founder mode, but yeah, gone really well.
46:00
I think it has this video and LinkedIn has I think over 12,000
46:03
views or something like that just on LinkedIn.
46:06
So I think everyone in this space pretty much like brands
46:09
have have seen it and commented on it and so on.
46:12
So it, it, it did like we, we put in a lot of work, but I
46:15
think like it just performed really, really well for us.
46:18
So yeah. Did you hire somebody to help
46:22
with that, or did your team do all of it yourselves?
46:24
The no for the for the shoot we did, yeah, we hired some, like
46:29
some friends of mine who went to film school.
46:31
But the the direction, the like the script and everything was
46:37
was all internal. I see.
46:38
That's really good. Yeah, I wanted to.
46:40
Talk to people about this could be a good place to wrap up on.
46:44
You mentioned found a mode and actually being in the sales
46:46
course. Rashad just mentioned jumping
46:49
off to go to a sales school and we were just talking to Jacob
46:52
Postal who's just launched HQ and this is his calendar.
46:55
It's like back-to-back course. I think he posted like a post
46:59
onto his page. Like how do people raise and
47:03
deal with a calendar full of sales calls?
47:06
And I was like, yeah, I read it. I thought sales reps right.
47:10
You hire sales reps and then you go do the thing that only you
47:14
can do. Oh, yeah.
47:15
This is it. The paradox of fundraising.
47:17
Founders with bad businesses run a full process pitching A
47:20
gazillion VC's. Founders with good businesses
47:23
busy running a company. My calendar is 10 hours a day of
47:26
demos, the rest important to customers.
47:29
I thought this was what sales reps before, but like, obviously
47:32
this is founders mode. And like, you know, I'm in.
47:34
I do all of our sales course because we're tiny.
47:37
But yeah, I'd love to hear from you guys.
47:38
You know, Jeremiah, you still in sales course?
47:40
And yeah. What's your what's your thought
47:42
process to this? I mean, I can see this paradox,
47:44
but I can also see like, you know, a few of my friends have
47:47
software businesses and they have sales reps.
47:49
They're not in any of the sales course, right?
47:51
Yeah. I mean, most of our customers,
47:54
we have a couple of different types of revenue streams, but
47:59
majority of it is, you know, software subscription, somebody
48:04
signing up to use the core piece of our our platform.
48:08
And I think our largest customer is like .6% of our revenue in
48:14
that context. And so I just don't do those
48:17
calls. It's just like there's no,
48:19
there's no customer I'm going to talk to that's going to move the
48:21
needle in a way that's worth me getting on a call and spending a
48:27
bunch of time on it. And in that context, now I'll
48:29
communicate with people and you know, some people tips and
48:32
tricks and all that kind of stuff, but I'm not going to do a
48:33
sales call. And then we have a couple of
48:36
other things like, you know, we have some relationships where
48:39
it's like bigger bulk pricing type things or it's like a
48:44
relationship with an agency or relationship with an ad platform
48:47
or whatever those things are. And that's a different story.
48:50
But you know, when you're when most there's only a handful of
48:54
those and there's like, you know, couple opportunities 1/4
48:57
like that, that I engage with and otherwise I just leave it to
49:00
the sales Rep. So it's just not like we're at a
49:02
scale now where that's it's not worth my time to to be on that.
49:08
I feel I hate saying that because it's not that I don't
49:10
care about the customers. Like I literally don't have time
49:12
to spend with thousands of customers.
49:14
It's just not an option for me to give that to everybody.
49:18
But the first like 3 years, I was basically doing all the
49:22
sales calls 22 to three years and then ultimately just kind of
49:25
handle that off over time. Yeah, I like that.
49:28
I think we're, I'm kind of seeing both layout at the same
49:32
time because we have Sarah, which is kind of like our older
49:35
legacy company, which is still running, still active, still
49:38
onboarding customers. And I'm not like involved.
49:41
Like we're now at a stage, like, I mean, there used to be a time
49:44
where every single customer we on board, I knew them by name.
49:47
I knew the city they lived in, like I knew what their hobbies
49:50
were like everything. Now we're at the point where we
49:53
like are ahead of sales Max, like he just makes sales and we
49:56
just know like, and that just works, right.
49:57
So like I'm not involved in that whatsoever.
50:00
Even on the customer success side, there's Chris on the team.
50:03
Like he takes care of it. Like I'm not super involved in
50:05
that. I'm looking at metrics like
50:08
sometimes like I get looped into a sales call, but maybe it takes
50:11
maybe twice a month, right? Which is like maybe an hour of
50:14
sales calls on the serial side on the southeast side because
50:17
it's like like literally today is like one month as of the
50:20
recording. It's like brand spanking new
50:23
everything I'm involved in everything like talking to
50:26
customers on the product side, sales calls on boarding calls,
50:32
setting them up, migrating their affiliates, like I'm in
50:34
everything. So I think like I get to see
50:36
like both worlds where Saul now is almost almost three years
50:40
old. So that's like set up and
50:42
running. But Sati, because it's new, I
50:45
get to do it almost all over again in the right way.
50:49
So almost like I'm like a second time founder in like the same
50:53
company in some sense, you know, so.
50:55
Yeah. And how fast are you trying to
50:57
get out of sales course? And I guess the take away for
51:02
somebody who is in sales calls every day, you know, yeah, when
51:08
should you be getting out of sales calls?
51:10
Obviously at your bandwidth. But like Jacob, to some extent,
51:13
I imagine people want to talk to Jacob and that's why he's doing
51:16
10 hour, you know, 10 hours of sales calls a day.
51:19
But that can't scale, right. Like if you if I imagine like
51:23
even if it's a week out, there's opportunity cost there.
51:26
Or like if if people can't get on your calendar on the same
51:29
week, then are you losing customers just because you're
51:31
not hiring a sales Rep? Yeah.
51:33
I think the way I would think about it for us because we're
51:36
new is like rate of learning right now on every single sales
51:40
call, I'm like learning something new.
51:42
I'm taking notes, I'm passing it on to product versus like answer
51:45
all. We're not necessarily like
51:47
learning those many things on sales calls.
51:49
We learn things when we like talk to customers who are deep
51:52
and they're like so at the edge of the program that they want
51:55
this one thing that the platform doesn't do.
51:57
When we need to build that, that's where we learned where I
51:59
think like it's like how top of funnel are is the learning
52:04
right? I think like for Sati, it's like
52:05
at the sales call level for Sarah, it's like customer
52:08
success. Once they're activated, what do
52:11
they need? It's like, OK, the customer's
52:12
like 6 months old is when I I talk to them.
52:15
I think I look at it as I think Jeremiah is nodding to probably
52:18
agree. I look at it from a rate of
52:20
learning perspective. Like once I once like the rate
52:23
of learning, however you want to measure it, like insights per
52:27
call or whatever. Like once I feel a drop off, I
52:30
think it's I feel more comfortable even giving it or
52:33
delegating it to a sales Rep saying, OK, like, yeah, you just
52:37
do it. And then I, I go like lower and
52:39
lower in the funnel because then like there's more learnings to
52:41
be had. But like I think of myself as
52:43
like context learning machine because I want to be like an
52:47
expert in how the customer thinks and what they want.
52:51
Yeah, I think that's really interesting the and I would
52:55
agree with that. I like the rate of learning
52:57
framing of it. I think of it kind of in terms
52:59
of like product market fit, which everybody has different
53:01
way of defining, but it's similar concept of like, do you
53:05
once you know that the product meets the customer need at
53:07
whatever price point, you then at that point, that's really
53:12
that's the first point that you could really hand it off to
53:14
somebody who isn't a Co founder level person.
53:18
Maybe that's a little unfair. But anyway, you need you need
53:20
the authority to make decisions on the fly and and to be able to
53:23
say, OK, this doesn't work for a person.
53:25
So let's let's try something else.
53:27
And you're basically experimenting, experimenting,
53:30
experimenting. So once you kind of start to get
53:31
out of that experimentation phase and you have something
53:33
that works, that's when I would end it up.
53:37
And that could be among us. It's probably like eager.
53:41
Yeah. You know what's really
53:43
interesting about both those answers, though?
53:45
Neither of you spoke about selling.
53:48
You both spoke about making sure that the product is right for
53:50
the customer, right? And obviously the salesperson
53:52
comes in with that as the main objective, right?
53:55
Like sell, But like this is a different approach.
53:58
It's I want to make sure the product is right for the
54:00
customer who's buying and get the sell that way.
54:03
Right. Yeah.
54:04
Interesting. Yeah, because I mean, at that
54:06
point, ideally you don't have to solve that problem for a sales
54:09
Rep. It's just like a matter of them
54:11
identifying good fit customers and getting them.
54:15
Yeah, because of like our GTM is so new, I get to like learn
54:18
what, what's working. Hey, did you actually see the
54:20
launch video? Did you see like a customer post
54:23
about it or like how did you come?
54:24
So like I, I, I'm learning that if I want to make like pricing
54:27
changes, I can just, I can just do that Like hey, OK, this, this
54:30
like this component here, OK, you get this feature here.
54:33
Like I'm learning all of that versus.
54:34
So if I give it to a sales Rep, then that like so they have to
54:37
like they got to like talk to me about it or like, you know,
54:40
somebody about it and then they change the pricing and then it's
54:43
like it just slows the whole process down versus when I'm in
54:46
it, then I can just take these calls.
54:48
And then so again, this is kind of like pricing, market fit or
54:52
whatever you want to call it, right?
54:53
So yeah. Cool.
54:55
I think that might be a good place to to finish it off.
54:58
It was fun, guys, very good at the end there.
55:02
I mean, I always learn so much in this course, so I appreciate
55:05
you coming back on. Yeah.
55:06
Yeah, thank you. Thank you for having me again.
55:09
Yep.