August 14, 2025 • 35 mins
Philipp Kandal is the chief product officer of Grab, an app that serves several countries across Southeast Asia. Two of Grab’s main businesses are delivery and mobility – like a combination between Instacart and Uber. And maps are at the core of its business.
On today’s show, Philipp talks about improving online maps for places like Southeast Asia, where streets are often winding, narrow, and harder to access than those in the US and other developed countries.
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:15):
Pushkin, what do you understand about maps and the world
because you have been, you know, studying it and working
on it for so long.
Speaker 2 (00:26):
I think the reason I personally love maps is just
because there's a super super complex problem and very very
hard thing to solve. That's why why I'm so passionate
about it.
Speaker 1 (00:37):
This is Philip Kundall. He's the chief product officer at
a company called Grab. Grab is huge in Southeast Asia
and its main businesses are delivery and mobility, kind of
like a combination between Instacart and Uber, and as a result,
maps are at the core of its business.
Speaker 2 (00:53):
But the fascinating thing for me is like, once you
get a digital representation of the real world, you can
basically the vision that I always had for long, long years.
If you haven't solved that yet is imagine when he
went from Yahoo to Google. Right. Yahoo was like the
static index of web that you could say, for something
that just leads you to a website. Right if you
would search for sports and it gets you to a
(01:14):
sports side and then you need to navigate your way.
And then Google you can say you search for a
very specific thing and it brings you exactly to that page. Right, Yeah,
and that's what we haven't solved with with maps yet.
Speaker 1 (01:26):
Right if you say, if we're still in the Yahoo
era of maps, is what you're telling me.
Speaker 2 (01:31):
Maybe slightly beyond the Yahoo era. But let's say like
when you say, where do I find the in Southeast Asia?
Durian is like a super popular fruit? Right say where
I said, where do I find the freshest Durian for
this price? Right now, right now, the real world, right now,
in the real world, right like, which store stocks a
Durian right now? Show me where it is and then
(01:52):
guide me to that store. There's nobody who has solved
that problem. And and that's the fascinating part for me
that I want what Google has done for the web.
I want that for the real world.
Speaker 1 (02:03):
I mean, that's a wild problem. When you formulate it
that way, you would have to know everything all the time.
There's an information problem there that is quite hard.
Speaker 2 (02:13):
I mean, the hot problems are the fun ones and
a yeah.
Speaker 1 (02:23):
I'm Jacob Goldstein, and this is what's your problem? The
show where I talk to people who are trying to
make technological progress. Philip Bilton sold a mapping startup before
he wound up at grab in twenty nineteen. Today, he's
based in Singapore, which is one of several countries where
Grab operates. Others include Indonesia, Malaysia, the Philippines, Thailand, and Vietnam.
(02:44):
And in addition to doing delivery and mobility, they also
do payments. It's what they call a super app. And
I wanted to talk to Philip about maps in particular
because I mean, I guess I just kind of thought
maps were solved or assumed that without really thinking about it,
and then when I started learning about GRAB, I realized
I was very wrong. So there's that dream of real
(03:08):
time mapp apps that Philip mentioned a minute ago. But
there's also something that's really interesting and kind of more
immediately relevant, and that is this The online maps we
use in the US and other developed countries just don't work.
Speaker 3 (03:22):
Very well in a lot of other parts.
Speaker 1 (03:24):
Of the world. And when Grab launched several years ago
in Malaysia, that turned out to be a big problem.
Speaker 2 (03:30):
I mean, from day one, you needed maps, so you
can't run the company without maps, and grabs started using
a third party service. But then what we've realized a
lot of these services are built for like a developed
market like the US, and very built around the mental
model of like cars, and Southeast Asia is really different.
We're operating primarily on motorbikes, even two thirds of our
(03:53):
transport trips on the back of a motorbike. And then
if you know Southeast Asian cities, then you have these
narrow alleys and sideways and so on, and traditional maps
just don't cover them because interesting, basically how maps are
made traditionally is with these big mapping vans that I'm
sure you've seen driving through cities. One hundred and fifty
two hundred thousand dollars basically look like a way more right,
(04:16):
That's kind of like how they look like, and they
don't cover the roads that we need to deliver our
services and really reach our customer because they expect to
be picked up in their home. They expect that the
food get delivered to their front door, and not just
to the nearest street that might be two hundred meters
away in terms of like a big car drivable street.
Speaker 1 (04:35):
So lots of life, lots of people live, lots of businesses,
earned streets that a van couldn't even drive down if
it wanted to.
Speaker 2 (04:42):
Yeah, no chance. I mean, when I do the immersions
like on the back of a motorbike. There's absolutely no
chance that with a man you can get there. It
was a scooter centric world, and Southeast Asia updated so quickly,
like I mean, there's like entire new neighborhoods springing up,
and it was traditional maps are built in a way
that these big vans collect once every one or two years,
(05:05):
and we need to refresh the maps in like days.
Speaker 1 (05:08):
So you have this problem, like what's the first step?
So you realize the maps aren't working. You're a mobility company,
that's that's not gonna work. How do you It seems
impossible to think, oh, well just map everything.
Speaker 2 (05:19):
How do you go?
Speaker 1 (05:20):
How do you start doing that?
Speaker 2 (05:21):
Yeah? Exactly right. It's a crazy problem, right yeah. I
mean the numbers that were out there when Google started
mapping the US, they apparently spent like anywhere between half
a billion to one billion dollars and it clearly didn't
have that amount of money to map it. And it
seemed crazy, right like people when we started doing this,
I mean, I got like so much feedback that people
(05:42):
thought we were completely insane because God cost us and
Southeast Age are right, I mean, we are home to
like close to seven hundred million people, which is like
a lot larger than the US. So you can imagine
like how much it would normally.
Speaker 1 (05:55):
Cost, yes, two x exactly and crazy dense cities, traffic,
tiny alleys. So how do you even start to undertake
this project? What do you do?
Speaker 2 (06:06):
Yeah? No, so this was this was really the fun part.
That's such a fun challenge yourself. But basically what we
started is we started taking it from first principles like
why was it so expensive to map? And then we
tried to solve those problems and the problems why it's
so expensive normally to produce a map are a few things,
the ones that I just said. The mapping vehicles are
(06:27):
extremely expensive. Then the people sitting in the mapping vehicle
are extremely expensive because you sent them around driving in
the country, sleeping in hotels. So the cost like to
send these vans with people around cost a fortune, and
like operating that is just super super costly. And that's
(06:48):
what we had a unique advantage because obviously we had
our drivers already crisscrossing the city in like insane amounts.
I mean our drivers. Any city is like crossed by
our drivers like one hundred times or more in a day,
so there's like no roads that our drivers don't see.
So we thought about two things that we needed to
drive the radically down. One is we build our own
(07:11):
collection hardware. So instead of building these like massive rigs,
we build small GoPro like cameras with an echip in
there that we can give to our drivers. And instead
of costing one hundred two hundred thousand dollars for full
mapping van, these cameras are under order of like a
bunch of one hundred dollars, so orders of magnitudes cheaper.
And then the second thing is we said, the driver's
(07:32):
drive around the city anyway, can they just have the camera?
And then we just need to pay them a little
for them. It's a great extra income. But they get
their main income from doing food deliveries or doing mobility trips,
and they get an extra income from that. And that's
because we could deploy like we have in Southeast Asia
probably one hundred times at least more cameras than anybody
(07:54):
else deployed because they're so cheap. And then we don't
need to send drivers traveling everywhere.
Speaker 1 (07:59):
So let's break it down a little bit like building
your own hardware, like the driver part is kind of obvious, right, like, oh,
we've got these guys already out there, could we get
them to do the mapping. It's not at all obvious
to me that you would need to build your own hardware.
So tell me about that, Like, first of all, why
do you need to build your own hardware? Why not
just dumb question, buy a camera?
Speaker 2 (08:19):
No, great, great question. Actually, I mean that's what we tried. Okay,
So I think when we went and looked into this,
because we didn't want to build hardware, and so basically
we tried two things. So they were these go pros
a few hundred bucks cameras, and then they were the
more professional mapping Greade cameras twenty thousand bucks. So the
problem with go Pro was a bunch of things they made,
usually as action cameras, right, like you can go skiing
(08:40):
and so on with them, but they don't have great
GPS in them, and basically because you don't need it,
right Like, if you go like somewhere mountain biking, it's
typically open sky. It's not like a dense urban environment
with big skyscrapers that reflect GPS. So they're decent for
what people usually use. Goal pros, but they're not the
media level precision we need for high density urban center mapping.
(09:03):
So that was one problem. And the second problem with
Goal pros was they just have no tooling that they
kept operate twenty four to seven drivers upload automatically data,
so they're not made They usually made for like a
one two hour recording and not this heavy duty recording
for the back of our motorbikes for twelve hours in
blistering heat in Southeast Asia, tough rail and so on.
(09:26):
So that's basically why both of these cameras that existed
didn't fit the needs that are very harsh environment.
Speaker 1 (09:32):
So would the mapping camera have worked, but you didn't
want to spend twenty grand per camera. That was presumably
the simple problem with the mapping camera.
Speaker 2 (09:38):
Yeah, twenty grand was the problem. And the other problems
are they're also quite heavy and bulky, So those cameras
are close to ten kilograms, which is not super easy
to operate. So they were too bulky and too costly.
Speaker 1 (09:52):
So what do you just get on a plane dest
engine and tell them what you need, Like, how do
you build your own camera?
Speaker 2 (09:59):
So this was really certing that we actually had a
small team and Scenzan already building building off our battery
swap blockers for scooters. And back then I talked to
our Cito at that time and he is like kind
enough to say, like I feel it, like we really
need to find something next for this team to do.
So you can have them all. They can build your cameras,
(10:20):
so amazing, We're very lucky.
Speaker 1 (10:23):
So tell me about the camera they came up with.
Speaker 2 (10:25):
Yeah, so the first camera that we built, we called
it a Karda cam, so like after like carda in
like some language, was like map. So basically those cameras
where the first version was mounted on the helmet of drivers,
predominantly it was two hundred and fifty grams mounted on
the helmet of drivers, and it would basically be a
camera with an AI chip and really highly accurate GPS,
(10:48):
like that's all we needed. And so they would just
mount them on their on their helmets and go about
their day and just at the end of the day,
take it home, take it on their Wi Fi, upload
the data and yeah, we would get data from tons
of cameras.
Speaker 1 (11:01):
And then you made a cardacam too, right, tell me
about Carteracam too.
Speaker 2 (11:05):
Yeah, we made a bunch of iterations of our hardware.
So Karda CAN two is our latest generation, which is
basically a three sixty camera, so it has basically four
camera lenses and all directions. Has also like more advanced centers,
like allied our sensors in there, so we basically made
the setup a lot more professional that we can capture
(11:25):
maps an even higher level of accuracy to get things
like lane level navigation for our drivers or advanced safety features.
So the latest camera is basically a great iteration that
has taken just a step further to capture more details
in the map.
Speaker 1 (11:40):
And where does it go? Where does the camera go
if somebody's driving a scooter.
Speaker 2 (11:43):
Yeah, so now we've built a mount on the back
of a motorbike. So we've built actually the cameras that
you can either mount them at the back of your
motorbike that it doesn't disturb you and in a pole
so that's high so that you can see because the
camera obviously needs to have a good view even if
there's cars around, so we've built a special mount on
the back of a motorbike. We have mounts also for cars,
so some drivers also mount them on cars. And then
(12:05):
we have a backpack that you can carry if we
want a map indoor. The other thing that's really important
in Southeast Asia is malls are gigantic. Do you have
like these malls which it takes you fifteen minutes to walk.
Speaker 1 (12:16):
Around, and so people will get a delivery to whatever
the noodle shop in the middle of the mall, and
you've got to put that on the map.
Speaker 2 (12:25):
Well, the other way around, people get a delivery from
the noodle shop of course, of course, and our driver
needs to find their way there, and if they get
into the wrong entrance of the mall, it might cost
them ten to fifteen minutes extra to walk back and forth.
So we need to precisely not only tell them go
into the mall, but we also precisely need to tell
them where to park their motorbike. So what we emphasize
(12:46):
when we build our maps that we build them like
and to end, and we say, here's where you park
your motorbike, here's where you walk, here's where you pick
up the noodles, and then you can do the same
in reverse.
Speaker 1 (12:57):
How many of these cameras do you have out there
like today? How many people are driving around mapping with
these cameras today ish by end of this year.
Speaker 2 (13:04):
We have about twenty thousand cameras in the field. And
to give you a sense, like professional mapping companies in
Southeast Asia, to my best of my knowledge, have about
tens of cars, but not tens of thousand tens of cars.
Speaker 1 (13:17):
So does that mean they have basically seeded it to
you like that you have won the mapping Southeast Asia fight.
Speaker 2 (13:25):
I think it's still so so early. Fair. I mean,
there's so many things we want to do.
Speaker 1 (13:30):
Interesting. I love that. So let's talk a little more
about what you're doing now and then let's talk about
what you want to do. So you have an incredible
amount of data coming in now, right, I mean, if
you have twenty thousand cameras driving around every day like
that is a wild amount of input. What are you
doing with all that data? I mean there's basic mapping,
(13:50):
and I'm sure you've got that, but what's the next level?
Presumably have AI? You said you have light? Are like,
tell me all the things you know because you have
all this data.
Speaker 2 (13:57):
Yeah, you're right. I think the basic vision that you
have is get an accurate representation of the real world
in real time. That's that's the that's the goal of
mapping in real time?
Speaker 1 (14:09):
Is crazy? In real time? It's crazy? Yeah, that's wild.
But fine, that's just like right now, what do you know?
Speaker 2 (14:17):
Yeah, so a bunch of things is I mean, first
of all, everything that you need for navigation roads, how
our road's drivable? Is the road safe to drive? Does
it have a lot of potholes and things like that,
and that signage obviously is it a one way road?
Is it not a one way road?
Speaker 1 (14:32):
Can you worry me about a particular pothole? Could you
be like, be careful, there's a pothole on the left
coming up, try that right lane.
Speaker 2 (14:39):
And we're doing actually very cool things we're doing right now.
We've already launched a safety navigation that tells you, hey,
is the road safe to drive? Based on potholes? And
does it have street lighting? Is it safe? Because if
the road at night is well lit, it's a lot
safer to drive, but it's not. So that's kind of
like a practical use case that we that we already
can do with that.
Speaker 1 (14:58):
Interesting as I'm sure you know, like ways in this
country tells you where there's like a speed camera or
like a speed trap. Do you do that?
Speaker 2 (15:10):
Yeah. Absolutely. So we have an AI voice reporting that
drivers can share anything they're like, oh, the right lane
of this road is closed, and then it gets processed
and basically used for all the other other drivers. So
we've launched that and it's been really successful as well.
Speaker 1 (15:26):
So when you say it's AI, does that mean the
driver just says it and it gets integrated into the
maps or what does that mean?
Speaker 2 (15:33):
Yeah, that's that's exactly right. There's the future. We actually
work closely together with our friends at open Eye. So
the driver natural language reports an issue and then it
gets processed and if you're not clear, we ask your
follow up question. If you say, hey, the right lane
is closed and he said like, oh, do you mean
this road? And then he says drive but yeah, that's
exactly the road I'm talking about, and then we process
it and basically one all the other drivers accordingly.
Speaker 1 (15:56):
So you mentioned your friends at open ai. It seems
like good friends to have a good place to have friends.
I know you have a deal with them, Like what
is the nature of your deal with open ai and
more generally, what's the work you're doing with them?
Speaker 2 (16:08):
So MAP I think is one of the key things
we do together with them and use their models to
improve our maps. The voice stuff that I shared earlier
is like one of our one of our highlight features.
And in general, we're just embedding like AI and everything
we're doing. We're having we're having like over a thousand
different AI models that we're that we're working on, so
(16:29):
it's basically deeply, deeply embedded in many many of the
things that that we're doing.
Speaker 1 (16:33):
Tell me more, like, what are some more examples of
how you're using AI.
Speaker 2 (16:38):
So one of the latest things that that I really
like that's really cool is the other thing that has
really huge impact on our marketplace is weather. In Southeast Asia,
the rain is insane. It is like pouring down, the
roads are flooding, so and for our services for mobility
for deliveries, that has a tremendous impact on that. So
(17:00):
knowing when it rains early is super super critical so
we can adjust the marketplace. We've launched AI based rate
detection that a bunch of things, So we deploy sensors,
other sensors in cars. So we have basically a device
that's called a card to dongle that we deploy and
the cars that we own and it plugs into a
port in the car. That's called OBEDI two port, and
(17:21):
that reads when the windshield viper's going.
Speaker 1 (17:24):
Oh genius, such a simple way, such a simple way
to know if it's rating. Does the car have the
windshield wipers on? It's so second order. I love it though.
Speaker 2 (17:35):
Yeah, and how fast it is?
Speaker 1 (17:36):
Oh? How fast? It's how fast the car is going,
because the slower it's going, the heavier the rain. Is
that the inference how fast.
Speaker 2 (17:42):
The windshield wiper is going. Because you kind of like this,
are this right?
Speaker 1 (17:47):
So okay? So so and what do you do? What
do you do with that data? That's the input, what's
the output?
Speaker 2 (17:53):
The output is basically we know the moment it rains,
we know demand will go up and supply of drivers
will go down. So what we do We try to
activate more drivers, so we would send out to all
the drivers, Hey, it's starting to rain. There's a fantastic
earning opportunity. If you're not worth now's a great chance
to get on the road and make extra money. And
(18:13):
then we try to actively get the supply of drivers
up so that we can keep our reliability at the
levels we need.
Speaker 1 (18:20):
We know people get all worked up about sarge pricing.
Do you use searge pricing in that setting? It sounds
like a classic use of search pricing. Everybody wants a driver,
nobody wants to drive. What you need is a higher price.
Do you do that?
Speaker 2 (18:30):
That's exactly right? Or like you need to motivate the
drivers to come on the road.
Speaker 1 (18:34):
Yeah, no, there is a market. I'm pro searge pricing.
That's a great example. What's another example of the way
you're using AI?
Speaker 2 (18:41):
So we used AI to translate in many scenarios. For example,
when I go like to Jakarta, I obviously don't speak
any Bahasa, but I can message our drivers in the
real time, translate any messages, send them a chat to
the driver and he can reply back to me in basade.
I see it in English on my side, he sees
it in basade my site. But the more important one
(19:03):
for me was like food menu translations. So we invested
quite a bit because whenever I go to Thailand and
then like I mean, I can't read any tie script obviously,
and I in the past I look at like some
things on the menu and I have no ideas that
like for me, always are worries that ultra spicy can
I eat it or not, and I didn't even know
what the item is, Like can look at the picture
and kind of guess, but sometimes merchants don't have pictures.
(19:26):
So we used AI to to translate all these menus
in all kinds of languages, so that that has been
a super impactful one as well. We'll be back in
just a minute.
Speaker 1 (19:50):
So I'm curious about grab maps enterprise, right Like, you're
selling something related to your maps to big companies right
to to Microsoft, to Amazon and to government. It's like,
what tell me about that business?
Speaker 2 (20:04):
That was quite an interesting one, right Like, we would
have never imagined early on when we built our maps
that this business will be created. But we've got people
like once we publicized our maps and people have seen
that they work generally quite well, we've got people approaching
us say hey, can we use them as well? And
that was like the genesis of the enterprise business. And
(20:24):
then we started working very closely with AWS where any
developer on their platform can use our maps now, so
we have over one hundred different developers already using it.
We partner with Microsoft, as he said, a bunch of
other large tech companies that in Southeast Asia started using
our maps because they've seen i mean, all these things
(20:45):
that I shared earlier that just really serves their needs
in Southeast Asia a lot better than anything else out there.
Speaker 1 (20:50):
What's an example if something someone has built, you know,
on top of it.
Speaker 2 (20:55):
A really cool startup that I've seen that used our maps.
What they do They go from merchant to merchant and
collect like the old oil that they use for cooking
and that basically recycle it. So they send somebody around
going to all these merchants collecting that. And in the past,
and a lot of these merchants are like small neighborhood
mom and pop shops in all these like little side
(21:16):
rolls and alleys, and that had a hard time for
the for the person going around collecting all of this
stuff finding that. And that's one of these examples where
they use grab maps to make the navigation of the
person going around and collecting all of that a lot
a lot easier. And there's many of these kind of
stories where people use it for things that that are
(21:36):
very similar in those kind of finding the last mile
has been really really hard before. So I'm curious.
Speaker 1 (21:43):
What you're working on now, Like, Yeah, what what are
some of the things you're trying to figure out that
you haven't figured out yet.
Speaker 2 (21:49):
I think the one thing that we're really passionate about
is solving more of the indoor problem. So that's one
thing that we're really mapping more so. We have a
decent amount of malls already map, but there's still so
much more to do. So hopefully we can find that.
Butever you go into these malls, we can exactly help
you to find whatever you're looking for, specific store or
general a general kind of like shop or so. So
(22:13):
that's that's one thing that I really want us to crack.
Speaker 1 (22:16):
So it's the general shop idea that like, if I'm
just not working for grab, I'm just in the general
public and I'm like, I want to buy a pair
of shorts, I just type that into grab maps, and
grab maps tells me where to go.
Speaker 2 (22:28):
Is that what you're thinking of there, Yes, but we
always think about it with a hyperlocal twist.
Speaker 1 (22:32):
Yeah.
Speaker 2 (22:33):
So, and what that means. As an example, let's say
in Indonesia, a large part of the population is Muslim,
which means they generally eat halal, and this is like
all the mapping platforms they don't support that. You can,
of course for every search for you can say restaurant halal,
this halal, this halal, but you cannot make it part
of your user profile. But it's it's like, it's not
(22:55):
it's not something that you switch every day. Well, like
either your preference is halal or is not. And and
a lot of the mapping platforms support putting in your profile.
I only want halal restaurants because I don't eat anything else.
And those are the kind of things that we see
in Southeast Asia that we need to really solve because again,
nobody else would solve those kind of problems. So capturing
(23:18):
this data accurately and knowing all these details, those are
the kind of things that we really put a lot
of emphasis on.
Speaker 1 (23:26):
Are there technical problems you're working on, Like are there
things where you haven't figured out the right tech or
where you need to build something, or where AI models
aren't quite where they need to be, but you're trying
to push them.
Speaker 2 (23:39):
I think what we are trying to do is what
I said earlier, like you want to have a real
time accurate model of the world.
Speaker 1 (23:44):
Yeah, so let's talk about that. Like real time is
a crazy phrase in that context, right, Like when you
say real time a model of the world. What are
you thinking of?
Speaker 2 (23:55):
Yeah, I mean a simple use case would be nowhere
there's parking right now? Yeah, not like where there's parking
in general, but I'd say a roadside parking. If you
had another crap card driving past and say, hey, fifteen
seconds ago there was a freak parking spot on the right,
that's that's extremely useful.
Speaker 1 (24:15):
That is extremely as well that I know people in
Brooklyn and San Francisco who would love to have that functionality.
Speaker 2 (24:22):
Yeah. I think for us, really the goal is always
what we know is like shaving off seconds of every delivery, right, Like,
that's what really really makes a delta. I think that
I really loved Steve Job's old mental model. But he
convinced people to make the macboot like ten seconds faster,
and he said like, well, if you take this, I
(24:42):
don't remember the exactly now mark for him, but he said, like,
if you make the macboot ten seconds fast and there's
fifty million people using it every year, you save like,
I don't know, it's like ten lifetimes of people's time
waiting for the Magic Boot something like that, and fast
right Like across I mean across I calculated to share
this always with a team. Across a billion deliveries, two
point five seconds saved across every delivery is roughly one
(25:05):
lifetime that you can save. So any second we can
shave off by getting cash to park faster, by getting
motorbikes to park faster, park at the right space, at
the right time. That's kind of the problems that we're
really passionate on solving.
Speaker 1 (25:19):
How long does it take you to do a billion deliveries?
How is it a billion per per way?
Speaker 2 (25:24):
We don't publish exact data, but it's the auder of
magnitude of it, like a year or less.
Speaker 1 (25:29):
Oh wow, okay, that's a big number. So how do
you get real time parking data? I mean, so, is
the constraint now just getting more cameras to more drivers,
like the what's the rate limiting step?
Speaker 2 (25:42):
Yeah? Great, great question. I think more cameras is one constraint.
The other constraint is doing all the smart processing on
the act because obviously you cannot upload all these data
because that would be extremely costly, and mobile networks in
Southeast Asia aren't quite that powerful that you could upload
millions of video streams to the cloud. So we're working
a lot on what is called like ATGYI that we
(26:04):
can run all these models that are powerful but not
in the cloud but on the app or at least
on a mix of boths.
Speaker 1 (26:11):
So in this case, is the edge the actual camera?
I mean, what what is like is the dream the
camera itself is doing the work and just uploading something
very simple to the network.
Speaker 2 (26:21):
That's in many places already happening. We already have quite
powerful AI chips in the cameras. But I mean, of course,
like no mobile phone, no edge camera right now is
as powerful as let's say CHET GPT with GBT four O,
that's sure.
Speaker 1 (26:34):
I mean, you have this sort of spectrum where you
have a whatever one hundred million dollar data center on
one end and one hundred dollars camera on the other
and some things in between. Right, So what can you
do on the camera now?
Speaker 2 (26:46):
We do things like privacy, so we blur people's faces,
we blow license plates because we never want to upload
this information. We do weather detection what I said earlier,
with rain, so we detect on the cameras if it's raining,
things like that. We detect traffic signs and see if
they've changed. So we actually already run large part of
processing on the edge and then only if something has changed,
(27:10):
then we upload some data into a validation on the server.
But that already allows us to reduce what we upload
by more than ninety percent to.
Speaker 1 (27:18):
Just tell the server if something is different.
Speaker 2 (27:20):
Basically, yeah, exactly.
Speaker 1 (27:22):
So I know, you know, we've talked mostly about maps,
but obviously GREB is doing a lot of different things.
Are there other parts of the business that we should
talk about?
Speaker 2 (27:32):
The business is so fascinating, so if I have time,
we should talk about all of our business.
Speaker 1 (27:36):
Yes, just tell me what's one other sort of frontier,
one other thing you're working on.
Speaker 2 (27:41):
I think the other thing which we've really invested deeply,
which is also quite closely connected to our maps, is
when we look at across the delivery journey. The other
part where a lot of time is spent is in
the merchant cooking and preparing the food. And that's an
area where you spend a lot of time optimizing together
(28:01):
with the merchants to make sure that the food is
prepared in the right time, that the driver arrives in
the right time. So you've done lots of lots of
cool things. So, for example, we've built a data science
model that accurately predicts at every time of the day
how long the merchant needs to prepare an order. So
we can detect the busyness of the merchant and say,
(28:22):
if we know all the merchant has gotten a lot
of orders, normally they take to prepare, They prepare to
bud me in like three minutes, but when they're very busy,
they take seven minutes. And the merchants don't want that
our drivers crowd their store and weigh in troves, So
we typically allocate the driver only when we know, okay,
the food is ready in seven minutes, and we don't
(28:42):
allocate the driver immediately, but we know all there's a
driver two minutes away, we just allocate them like five
minutes later so that he's in the shop just in time.
And that also cuts the delivery journey, makes a lot
more pleasant for the merchant not having like a lot
of people crowd their store, and allows us to offer
more affordable price to the consumer because they don't need
to pay for all these minutes of the driver waiting.
Speaker 1 (29:03):
So just all these optimizations, all these different margins where
you can optimize. So if we think about whatever five years,
ten years out, when you think about this sort of
medium to long term like what's your dream, Like, how's
it work? What's going on?
Speaker 2 (29:20):
I think like for me, the world is changing so fast.
Five to ten years prediction is really really hard with
all the AI advancements.
Speaker 1 (29:26):
How far you want to go?
Speaker 2 (29:27):
Two years? Four years?
Speaker 1 (29:28):
You tell me, like what you just give me a
dream for the future.
Speaker 2 (29:31):
I mean the obvious one that I'm very passionate about
is robotics that make so much sense in our marketplace.
So if you can add robotics to our marketplace, that
will be a huge, huge change and something we're actively
working on to make happen. So that's I think probably
the thing I'm most passionate about to change.
Speaker 1 (29:51):
Tell me what you're actively working on with robotics.
Speaker 2 (29:54):
We haven't shared much which we do on the delivery side,
but for example, I mean things like autonomous cars. We've
signed an agreement with a bunch of like autonomous car
providers to come with us to Singapore and so on,
so we'll do a bunch of things in that space.
But basically you can imagine that it will be in many,
many parts of our delivery chain.
Speaker 1 (30:14):
You need to build an autonomous scooter based on what
you've told me, right, I feel like the analogy to
the map story is the autonomous scooter that can go
down all the alleyways.
Speaker 2 (30:22):
Great, if you ever want to have a job in
our product team, to join us, build autonomous schooters with us,
because you're exactly right, but you need to build products
that work in our region. So that's exactly the right
mindset that we're trying, like what we say always and
we want to build hyper local products that work in
Southeast Asia.
Speaker 1 (30:42):
We'll be back in a minute with the lighting round. Okay,
let's finish with the lightning round. So I read on
your bio you write that you mostly travel between Singapore,
(31:03):
San Francisco, Berlin and Cluge. Tell me about Clue the
other three I'm familiar with Clues. I don't know anything
about How does that wind up on that list?
Speaker 2 (31:14):
As a city in Romania, in the heart of Transylvania actually,
and we have a we have a small engineering center
there in our maps team, So I've spent a lot
of time there because for my own startup, that's that's
how I wound up Inclusure. Originally for my startup, Scobbler,
which was a maps and navigation startup. I built a
(31:34):
engineering team and cluge, so I've been going to Clusion
working with engineers there for the last fifteen years or so.
What's clue Like, it's fun. It's a student town, so high,
high energy, young population, very smart, the computer science department
very good. They used to clone the IBM mainframes for
the Soviet Union, so a bunch of hardcore engineers.
Speaker 1 (31:58):
You also write in that bio you wrote in Berlin,
I experienced some of the best parties ever. Tell me
about a Berlin party.
Speaker 2 (32:08):
That was prior line, That was before I moved to
before I moved to Southeast Asia. But Berlin was a fun, fun,
fun journey for I lived there for five six years.
Speaker 1 (32:17):
Maybe he didn't tell me anything about any party. So
and then you also say you lived and worked in
Singapore and Berlin and San Francisco, and I'm curious, like,
how is sort of whatever professional culture, work life different
in those places. What are the sort of striking differences?
Speaker 2 (32:35):
I mean, Germany is extremely direct, but like that's where
I'm originally from. I'm originally German, and Germans are known
to be super super direct, which in Southeast Asia doesn't
work super well. If they're not accustomed to it. So
I think for me, the thing that I needed to
adjust most is to become a lot more indirect, to
(32:56):
become a lot more like have those conversations and like
smaller groups, not in front of everybody. So I definitely
had to just my style quite a lot when moving around.
But I found that always incredibly fun. I always loved
learning about new cultures, So after some adjustments, I really
really like it here.
Speaker 1 (33:16):
That is super interesting. Was there, as you were saying
that I was wondering, like, was there a particular moment
when you realized, Oh, I'm not behaving correctly in this
cultural context?
Speaker 2 (33:31):
Yeah, I mean this predate grap So this was when
I was first doing business in China. So I sold
my startup to a Silicon Valley company, but I also
got a two hundred people team in China report to me,
which was the first time that I ever managed a
team in China, and I was extremely surprised why they
wouldn't tell me about all the mistakes, all the things
(33:52):
that I did wrong. And I said, like, that's very unusual.
Normally I get a lot of like pushback from engineers.
And then I started like asking people is why don't
they do this? And then people said like, well, it's
kind of root in a public forum to say the
boss's wrong. And then I realized, like, oh, that's that's
(34:13):
why I just need to like change how I'm asking questions,
how I'm managing teams. And that's when I first managed
teams in China. Took me quite a while to figure out, honestly.
Speaker 1 (34:22):
And how does San Francisco fit in relation to both
working in Berlin and working in you know, China and Singapore.
Where's the US on that continuum?
Speaker 2 (34:34):
Do you ask for me? I think the thing that
I really loved in San Francisco was just the craziness
of ambition. When I spent like April back in in
SF four month to like vibe code and hack on
a bunch of hobby projects, and like you go in
every random coffee shop and there's somebody who said they
work on a billion dollar idea and they they're just starting,
(34:55):
they have nothing yet they're just like, Okay, I'm going
to make it big. So I think that that ambition
and that willingness to openly declare it even if people
know it's super unlikely. There's not thousands of people who
start billion dollar company, but in the Bay Area that's
thousands of people who say they will and this conviction
and this like optimism. I think that was for me
(35:16):
one of the most striking things in the US, and
that I still love the energy whenever I go there.
Speaker 3 (35:22):
Basically, Philip Condall is the chief product officer at GRAB.
Speaker 2 (35:33):
Please email us.
Speaker 1 (35:34):
At problem at Pushkin dot fm. We are always looking
for new guests for the show. Today's show was produced
by Trinamanino and Gabriel Hunter Chang. It was edited by
Alexander Garreton and engineered by Sarah Bruguer. I'm Jacob Goldstein
and we'll be back next week with another episode of
What's Your Problem.
Pushkin, what do you understand about maps and the world
because you have been, you know, studying it and working
on it for so long.
Speaker 2 (00:26):
I think the reason I personally love maps is just
because there's a super super complex problem and very very
hard thing to solve. That's why why I'm so passionate
about it.
Speaker 1 (00:37):
This is Philip Kundall. He's the chief product officer at
a company called Grab. Grab is huge in Southeast Asia
and its main businesses are delivery and mobility, kind of
like a combination between Instacart and Uber, and as a result,
maps are at the core of its business.
Speaker 2 (00:53):
But the fascinating thing for me is like, once you
get a digital representation of the real world, you can
basically the vision that I always had for long, long years.
If you haven't solved that yet is imagine when he
went from Yahoo to Google. Right. Yahoo was like the
static index of web that you could say, for something
that just leads you to a website. Right if you
would search for sports and it gets you to a
(01:14):
sports side and then you need to navigate your way.
And then Google you can say you search for a
very specific thing and it brings you exactly to that page. Right, Yeah,
and that's what we haven't solved with with maps yet.
Speaker 1 (01:26):
Right if you say, if we're still in the Yahoo
era of maps, is what you're telling me.
Speaker 2 (01:31):
Maybe slightly beyond the Yahoo era. But let's say like
when you say, where do I find the in Southeast Asia?
Durian is like a super popular fruit? Right say where
I said, where do I find the freshest Durian for
this price? Right now, right now, the real world, right now,
in the real world, right like, which store stocks a
Durian right now? Show me where it is and then
(01:52):
guide me to that store. There's nobody who has solved
that problem. And and that's the fascinating part for me
that I want what Google has done for the web.
I want that for the real world.
Speaker 1 (02:03):
I mean, that's a wild problem. When you formulate it
that way, you would have to know everything all the time.
There's an information problem there that is quite hard.
Speaker 2 (02:13):
I mean, the hot problems are the fun ones and
a yeah.
Speaker 1 (02:23):
I'm Jacob Goldstein, and this is what's your problem? The
show where I talk to people who are trying to
make technological progress. Philip Bilton sold a mapping startup before
he wound up at grab in twenty nineteen. Today, he's
based in Singapore, which is one of several countries where
Grab operates. Others include Indonesia, Malaysia, the Philippines, Thailand, and Vietnam.
(02:44):
And in addition to doing delivery and mobility, they also
do payments. It's what they call a super app. And
I wanted to talk to Philip about maps in particular
because I mean, I guess I just kind of thought
maps were solved or assumed that without really thinking about it,
and then when I started learning about GRAB, I realized
I was very wrong. So there's that dream of real
(03:08):
time mapp apps that Philip mentioned a minute ago. But
there's also something that's really interesting and kind of more
immediately relevant, and that is this The online maps we
use in the US and other developed countries just don't work.
Speaker 3 (03:22):
Very well in a lot of other parts.
Speaker 1 (03:24):
Of the world. And when Grab launched several years ago
in Malaysia, that turned out to be a big problem.
Speaker 2 (03:30):
I mean, from day one, you needed maps, so you
can't run the company without maps, and grabs started using
a third party service. But then what we've realized a
lot of these services are built for like a developed
market like the US, and very built around the mental
model of like cars, and Southeast Asia is really different.
We're operating primarily on motorbikes, even two thirds of our
(03:53):
transport trips on the back of a motorbike. And then
if you know Southeast Asian cities, then you have these
narrow alleys and sideways and so on, and traditional maps
just don't cover them because interesting, basically how maps are
made traditionally is with these big mapping vans that I'm
sure you've seen driving through cities. One hundred and fifty
two hundred thousand dollars basically look like a way more right,
(04:16):
That's kind of like how they look like, and they
don't cover the roads that we need to deliver our
services and really reach our customer because they expect to
be picked up in their home. They expect that the
food get delivered to their front door, and not just
to the nearest street that might be two hundred meters
away in terms of like a big car drivable street.
Speaker 1 (04:35):
So lots of life, lots of people live, lots of businesses,
earned streets that a van couldn't even drive down if
it wanted to.
Speaker 2 (04:42):
Yeah, no chance. I mean, when I do the immersions
like on the back of a motorbike. There's absolutely no
chance that with a man you can get there. It
was a scooter centric world, and Southeast Asia updated so quickly,
like I mean, there's like entire new neighborhoods springing up,
and it was traditional maps are built in a way
that these big vans collect once every one or two years,
(05:05):
and we need to refresh the maps in like days.
Speaker 1 (05:08):
So you have this problem, like what's the first step?
So you realize the maps aren't working. You're a mobility company,
that's that's not gonna work. How do you It seems
impossible to think, oh, well just map everything.
Speaker 2 (05:19):
How do you go?
Speaker 1 (05:20):
How do you start doing that?
Speaker 2 (05:21):
Yeah? Exactly right. It's a crazy problem, right yeah. I
mean the numbers that were out there when Google started
mapping the US, they apparently spent like anywhere between half
a billion to one billion dollars and it clearly didn't
have that amount of money to map it. And it
seemed crazy, right like people when we started doing this,
I mean, I got like so much feedback that people
(05:42):
thought we were completely insane because God cost us and
Southeast Age are right, I mean, we are home to
like close to seven hundred million people, which is like
a lot larger than the US. So you can imagine
like how much it would normally.
Speaker 1 (05:55):
Cost, yes, two x exactly and crazy dense cities, traffic,
tiny alleys. So how do you even start to undertake
this project? What do you do?
Speaker 2 (06:06):
Yeah? No, so this was this was really the fun part.
That's such a fun challenge yourself. But basically what we
started is we started taking it from first principles like
why was it so expensive to map? And then we
tried to solve those problems and the problems why it's
so expensive normally to produce a map are a few things,
the ones that I just said. The mapping vehicles are
(06:27):
extremely expensive. Then the people sitting in the mapping vehicle
are extremely expensive because you sent them around driving in
the country, sleeping in hotels. So the cost like to
send these vans with people around cost a fortune, and
like operating that is just super super costly. And that's
(06:48):
what we had a unique advantage because obviously we had
our drivers already crisscrossing the city in like insane amounts.
I mean our drivers. Any city is like crossed by
our drivers like one hundred times or more in a day,
so there's like no roads that our drivers don't see.
So we thought about two things that we needed to
drive the radically down. One is we build our own
(07:11):
collection hardware. So instead of building these like massive rigs,
we build small GoPro like cameras with an echip in
there that we can give to our drivers. And instead
of costing one hundred two hundred thousand dollars for full
mapping van, these cameras are under order of like a
bunch of one hundred dollars, so orders of magnitudes cheaper.
And then the second thing is we said, the driver's
(07:32):
drive around the city anyway, can they just have the camera?
And then we just need to pay them a little
for them. It's a great extra income. But they get
their main income from doing food deliveries or doing mobility trips,
and they get an extra income from that. And that's
because we could deploy like we have in Southeast Asia
probably one hundred times at least more cameras than anybody
(07:54):
else deployed because they're so cheap. And then we don't
need to send drivers traveling everywhere.
Speaker 1 (07:59):
So let's break it down a little bit like building
your own hardware, like the driver part is kind of obvious, right, like, oh,
we've got these guys already out there, could we get
them to do the mapping. It's not at all obvious
to me that you would need to build your own hardware.
So tell me about that, Like, first of all, why
do you need to build your own hardware? Why not
just dumb question, buy a camera?
Speaker 2 (08:19):
No, great, great question. Actually, I mean that's what we tried. Okay,
So I think when we went and looked into this,
because we didn't want to build hardware, and so basically
we tried two things. So they were these go pros
a few hundred bucks cameras, and then they were the
more professional mapping Greade cameras twenty thousand bucks. So the
problem with go Pro was a bunch of things they made,
usually as action cameras, right, like you can go skiing
(08:40):
and so on with them, but they don't have great
GPS in them, and basically because you don't need it,
right Like, if you go like somewhere mountain biking, it's
typically open sky. It's not like a dense urban environment
with big skyscrapers that reflect GPS. So they're decent for
what people usually use. Goal pros, but they're not the
media level precision we need for high density urban center mapping.
(09:03):
So that was one problem. And the second problem with
Goal pros was they just have no tooling that they
kept operate twenty four to seven drivers upload automatically data,
so they're not made They usually made for like a
one two hour recording and not this heavy duty recording
for the back of our motorbikes for twelve hours in
blistering heat in Southeast Asia, tough rail and so on.
(09:26):
So that's basically why both of these cameras that existed
didn't fit the needs that are very harsh environment.
Speaker 1 (09:32):
So would the mapping camera have worked, but you didn't
want to spend twenty grand per camera. That was presumably
the simple problem with the mapping camera.
Speaker 2 (09:38):
Yeah, twenty grand was the problem. And the other problems
are they're also quite heavy and bulky, So those cameras
are close to ten kilograms, which is not super easy
to operate. So they were too bulky and too costly.
Speaker 1 (09:52):
So what do you just get on a plane dest
engine and tell them what you need, Like, how do
you build your own camera?
Speaker 2 (09:59):
So this was really certing that we actually had a
small team and Scenzan already building building off our battery
swap blockers for scooters. And back then I talked to
our Cito at that time and he is like kind
enough to say, like I feel it, like we really
need to find something next for this team to do.
So you can have them all. They can build your cameras,
(10:20):
so amazing, We're very lucky.
Speaker 1 (10:23):
So tell me about the camera they came up with.
Speaker 2 (10:25):
Yeah, so the first camera that we built, we called
it a Karda cam, so like after like carda in
like some language, was like map. So basically those cameras
where the first version was mounted on the helmet of drivers,
predominantly it was two hundred and fifty grams mounted on
the helmet of drivers, and it would basically be a
camera with an AI chip and really highly accurate GPS,
(10:48):
like that's all we needed. And so they would just
mount them on their on their helmets and go about
their day and just at the end of the day,
take it home, take it on their Wi Fi, upload
the data and yeah, we would get data from tons
of cameras.
Speaker 1 (11:01):
And then you made a cardacam too, right, tell me
about Carteracam too.
Speaker 2 (11:05):
Yeah, we made a bunch of iterations of our hardware.
So Karda CAN two is our latest generation, which is
basically a three sixty camera, so it has basically four
camera lenses and all directions. Has also like more advanced centers,
like allied our sensors in there, so we basically made
the setup a lot more professional that we can capture
(11:25):
maps an even higher level of accuracy to get things
like lane level navigation for our drivers or advanced safety features.
So the latest camera is basically a great iteration that
has taken just a step further to capture more details
in the map.
Speaker 1 (11:40):
And where does it go? Where does the camera go
if somebody's driving a scooter.
Speaker 2 (11:43):
Yeah, so now we've built a mount on the back
of a motorbike. So we've built actually the cameras that
you can either mount them at the back of your
motorbike that it doesn't disturb you and in a pole
so that's high so that you can see because the
camera obviously needs to have a good view even if
there's cars around, so we've built a special mount on
the back of a motorbike. We have mounts also for cars,
so some drivers also mount them on cars. And then
(12:05):
we have a backpack that you can carry if we
want a map indoor. The other thing that's really important
in Southeast Asia is malls are gigantic. Do you have
like these malls which it takes you fifteen minutes to walk.
Speaker 1 (12:16):
Around, and so people will get a delivery to whatever
the noodle shop in the middle of the mall, and
you've got to put that on the map.
Speaker 2 (12:25):
Well, the other way around, people get a delivery from
the noodle shop of course, of course, and our driver
needs to find their way there, and if they get
into the wrong entrance of the mall, it might cost
them ten to fifteen minutes extra to walk back and forth.
So we need to precisely not only tell them go
into the mall, but we also precisely need to tell
them where to park their motorbike. So what we emphasize
(12:46):
when we build our maps that we build them like
and to end, and we say, here's where you park
your motorbike, here's where you walk, here's where you pick
up the noodles, and then you can do the same
in reverse.
Speaker 1 (12:57):
How many of these cameras do you have out there
like today? How many people are driving around mapping with
these cameras today ish by end of this year.
Speaker 2 (13:04):
We have about twenty thousand cameras in the field. And
to give you a sense, like professional mapping companies in
Southeast Asia, to my best of my knowledge, have about
tens of cars, but not tens of thousand tens of cars.
Speaker 1 (13:17):
So does that mean they have basically seeded it to
you like that you have won the mapping Southeast Asia fight.
Speaker 2 (13:25):
I think it's still so so early. Fair. I mean,
there's so many things we want to do.
Speaker 1 (13:30):
Interesting. I love that. So let's talk a little more
about what you're doing now and then let's talk about
what you want to do. So you have an incredible
amount of data coming in now, right, I mean, if
you have twenty thousand cameras driving around every day like
that is a wild amount of input. What are you
doing with all that data? I mean there's basic mapping,
(13:50):
and I'm sure you've got that, but what's the next level?
Presumably have AI? You said you have light? Are like,
tell me all the things you know because you have
all this data.
Speaker 2 (13:57):
Yeah, you're right. I think the basic vision that you
have is get an accurate representation of the real world
in real time. That's that's the that's the goal of
mapping in real time?
Speaker 1 (14:09):
Is crazy? In real time? It's crazy? Yeah, that's wild.
But fine, that's just like right now, what do you know?
Speaker 2 (14:17):
Yeah, so a bunch of things is I mean, first
of all, everything that you need for navigation roads, how
our road's drivable? Is the road safe to drive? Does
it have a lot of potholes and things like that,
and that signage obviously is it a one way road?
Is it not a one way road?
Speaker 1 (14:32):
Can you worry me about a particular pothole? Could you
be like, be careful, there's a pothole on the left
coming up, try that right lane.
Speaker 2 (14:39):
And we're doing actually very cool things we're doing right now.
We've already launched a safety navigation that tells you, hey,
is the road safe to drive? Based on potholes? And
does it have street lighting? Is it safe? Because if
the road at night is well lit, it's a lot
safer to drive, but it's not. So that's kind of
like a practical use case that we that we already
can do with that.
Speaker 1 (14:58):
Interesting as I'm sure you know, like ways in this
country tells you where there's like a speed camera or
like a speed trap. Do you do that?
Speaker 2 (15:10):
Yeah. Absolutely. So we have an AI voice reporting that
drivers can share anything they're like, oh, the right lane
of this road is closed, and then it gets processed
and basically used for all the other other drivers. So
we've launched that and it's been really successful as well.
Speaker 1 (15:26):
So when you say it's AI, does that mean the
driver just says it and it gets integrated into the
maps or what does that mean?
Speaker 2 (15:33):
Yeah, that's that's exactly right. There's the future. We actually
work closely together with our friends at open Eye. So
the driver natural language reports an issue and then it
gets processed and if you're not clear, we ask your
follow up question. If you say, hey, the right lane
is closed and he said like, oh, do you mean
this road? And then he says drive but yeah, that's
exactly the road I'm talking about, and then we process
it and basically one all the other drivers accordingly.
Speaker 1 (15:56):
So you mentioned your friends at open ai. It seems
like good friends to have a good place to have friends.
I know you have a deal with them, Like what
is the nature of your deal with open ai and
more generally, what's the work you're doing with them?
Speaker 2 (16:08):
So MAP I think is one of the key things
we do together with them and use their models to
improve our maps. The voice stuff that I shared earlier
is like one of our one of our highlight features.
And in general, we're just embedding like AI and everything
we're doing. We're having we're having like over a thousand
different AI models that we're that we're working on, so
(16:29):
it's basically deeply, deeply embedded in many many of the
things that that we're doing.
Speaker 1 (16:33):
Tell me more, like, what are some more examples of
how you're using AI.
Speaker 2 (16:38):
So one of the latest things that that I really
like that's really cool is the other thing that has
really huge impact on our marketplace is weather. In Southeast Asia,
the rain is insane. It is like pouring down, the
roads are flooding, so and for our services for mobility
for deliveries, that has a tremendous impact on that. So
(17:00):
knowing when it rains early is super super critical so
we can adjust the marketplace. We've launched AI based rate
detection that a bunch of things, So we deploy sensors,
other sensors in cars. So we have basically a device
that's called a card to dongle that we deploy and
the cars that we own and it plugs into a
port in the car. That's called OBEDI two port, and
(17:21):
that reads when the windshield viper's going.
Speaker 1 (17:24):
Oh genius, such a simple way, such a simple way
to know if it's rating. Does the car have the
windshield wipers on? It's so second order. I love it though.
Speaker 2 (17:35):
Yeah, and how fast it is?
Speaker 1 (17:36):
Oh? How fast? It's how fast the car is going,
because the slower it's going, the heavier the rain. Is
that the inference how fast.
Speaker 2 (17:42):
The windshield wiper is going. Because you kind of like this,
are this right?
Speaker 1 (17:47):
So okay? So so and what do you do? What
do you do with that data? That's the input, what's
the output?
Speaker 2 (17:53):
The output is basically we know the moment it rains,
we know demand will go up and supply of drivers
will go down. So what we do We try to
activate more drivers, so we would send out to all
the drivers, Hey, it's starting to rain. There's a fantastic
earning opportunity. If you're not worth now's a great chance
to get on the road and make extra money. And
(18:13):
then we try to actively get the supply of drivers
up so that we can keep our reliability at the
levels we need.
Speaker 1 (18:20):
We know people get all worked up about sarge pricing.
Do you use searge pricing in that setting? It sounds
like a classic use of search pricing. Everybody wants a driver,
nobody wants to drive. What you need is a higher price.
Do you do that?
Speaker 2 (18:30):
That's exactly right? Or like you need to motivate the
drivers to come on the road.
Speaker 1 (18:34):
Yeah, no, there is a market. I'm pro searge pricing.
That's a great example. What's another example of the way
you're using AI?
Speaker 2 (18:41):
So we used AI to translate in many scenarios. For example,
when I go like to Jakarta, I obviously don't speak
any Bahasa, but I can message our drivers in the
real time, translate any messages, send them a chat to
the driver and he can reply back to me in basade.
I see it in English on my side, he sees
it in basade my site. But the more important one
(19:03):
for me was like food menu translations. So we invested
quite a bit because whenever I go to Thailand and
then like I mean, I can't read any tie script obviously,
and I in the past I look at like some
things on the menu and I have no ideas that
like for me, always are worries that ultra spicy can
I eat it or not, and I didn't even know
what the item is, Like can look at the picture
and kind of guess, but sometimes merchants don't have pictures.
(19:26):
So we used AI to to translate all these menus
in all kinds of languages, so that that has been
a super impactful one as well. We'll be back in
just a minute.
Speaker 1 (19:50):
So I'm curious about grab maps enterprise, right Like, you're
selling something related to your maps to big companies right
to to Microsoft, to Amazon and to government. It's like,
what tell me about that business?
Speaker 2 (20:04):
That was quite an interesting one, right Like, we would
have never imagined early on when we built our maps
that this business will be created. But we've got people
like once we publicized our maps and people have seen
that they work generally quite well, we've got people approaching
us say hey, can we use them as well? And
that was like the genesis of the enterprise business. And
(20:24):
then we started working very closely with AWS where any
developer on their platform can use our maps now, so
we have over one hundred different developers already using it.
We partner with Microsoft, as he said, a bunch of
other large tech companies that in Southeast Asia started using
our maps because they've seen i mean, all these things
(20:45):
that I shared earlier that just really serves their needs
in Southeast Asia a lot better than anything else out there.
Speaker 1 (20:50):
What's an example if something someone has built, you know,
on top of it.
Speaker 2 (20:55):
A really cool startup that I've seen that used our maps.
What they do They go from merchant to merchant and
collect like the old oil that they use for cooking
and that basically recycle it. So they send somebody around
going to all these merchants collecting that. And in the past,
and a lot of these merchants are like small neighborhood
mom and pop shops in all these like little side
(21:16):
rolls and alleys, and that had a hard time for
the for the person going around collecting all of this
stuff finding that. And that's one of these examples where
they use grab maps to make the navigation of the
person going around and collecting all of that a lot
a lot easier. And there's many of these kind of
stories where people use it for things that that are
(21:36):
very similar in those kind of finding the last mile
has been really really hard before. So I'm curious.
Speaker 1 (21:43):
What you're working on now, Like, Yeah, what what are
some of the things you're trying to figure out that
you haven't figured out yet.
Speaker 2 (21:49):
I think the one thing that we're really passionate about
is solving more of the indoor problem. So that's one
thing that we're really mapping more so. We have a
decent amount of malls already map, but there's still so
much more to do. So hopefully we can find that.
Butever you go into these malls, we can exactly help
you to find whatever you're looking for, specific store or
general a general kind of like shop or so. So
(22:13):
that's that's one thing that I really want us to crack.
Speaker 1 (22:16):
So it's the general shop idea that like, if I'm
just not working for grab, I'm just in the general
public and I'm like, I want to buy a pair
of shorts, I just type that into grab maps, and
grab maps tells me where to go.
Speaker 2 (22:28):
Is that what you're thinking of there, Yes, but we
always think about it with a hyperlocal twist.
Speaker 1 (22:32):
Yeah.
Speaker 2 (22:33):
So, and what that means. As an example, let's say
in Indonesia, a large part of the population is Muslim,
which means they generally eat halal, and this is like
all the mapping platforms they don't support that. You can,
of course for every search for you can say restaurant halal,
this halal, this halal, but you cannot make it part
of your user profile. But it's it's like, it's not
(22:55):
it's not something that you switch every day. Well, like
either your preference is halal or is not. And and
a lot of the mapping platforms support putting in your profile.
I only want halal restaurants because I don't eat anything else.
And those are the kind of things that we see
in Southeast Asia that we need to really solve because again,
nobody else would solve those kind of problems. So capturing
(23:18):
this data accurately and knowing all these details, those are
the kind of things that we really put a lot
of emphasis on.
Speaker 1 (23:26):
Are there technical problems you're working on, Like are there
things where you haven't figured out the right tech or
where you need to build something, or where AI models
aren't quite where they need to be, but you're trying
to push them.
Speaker 2 (23:39):
I think what we are trying to do is what
I said earlier, like you want to have a real
time accurate model of the world.
Speaker 1 (23:44):
Yeah, so let's talk about that. Like real time is
a crazy phrase in that context, right, Like when you
say real time a model of the world. What are
you thinking of?
Speaker 2 (23:55):
Yeah, I mean a simple use case would be nowhere
there's parking right now? Yeah, not like where there's parking
in general, but I'd say a roadside parking. If you
had another crap card driving past and say, hey, fifteen
seconds ago there was a freak parking spot on the right,
that's that's extremely useful.
Speaker 1 (24:15):
That is extremely as well that I know people in
Brooklyn and San Francisco who would love to have that functionality.
Speaker 2 (24:22):
Yeah. I think for us, really the goal is always
what we know is like shaving off seconds of every delivery, right, Like,
that's what really really makes a delta. I think that
I really loved Steve Job's old mental model. But he
convinced people to make the macboot like ten seconds faster,
and he said like, well, if you take this, I
(24:42):
don't remember the exactly now mark for him, but he said, like,
if you make the macboot ten seconds fast and there's
fifty million people using it every year, you save like,
I don't know, it's like ten lifetimes of people's time
waiting for the Magic Boot something like that, and fast
right Like across I mean across I calculated to share
this always with a team. Across a billion deliveries, two
point five seconds saved across every delivery is roughly one
(25:05):
lifetime that you can save. So any second we can
shave off by getting cash to park faster, by getting
motorbikes to park faster, park at the right space, at
the right time. That's kind of the problems that we're
really passionate on solving.
Speaker 1 (25:19):
How long does it take you to do a billion deliveries?
How is it a billion per per way?
Speaker 2 (25:24):
We don't publish exact data, but it's the auder of
magnitude of it, like a year or less.
Speaker 1 (25:29):
Oh wow, okay, that's a big number. So how do
you get real time parking data? I mean, so, is
the constraint now just getting more cameras to more drivers,
like the what's the rate limiting step?
Speaker 2 (25:42):
Yeah? Great, great question. I think more cameras is one constraint.
The other constraint is doing all the smart processing on
the act because obviously you cannot upload all these data
because that would be extremely costly, and mobile networks in
Southeast Asia aren't quite that powerful that you could upload
millions of video streams to the cloud. So we're working
a lot on what is called like ATGYI that we
(26:04):
can run all these models that are powerful but not
in the cloud but on the app or at least
on a mix of boths.
Speaker 1 (26:11):
So in this case, is the edge the actual camera?
I mean, what what is like is the dream the
camera itself is doing the work and just uploading something
very simple to the network.
Speaker 2 (26:21):
That's in many places already happening. We already have quite
powerful AI chips in the cameras. But I mean, of course,
like no mobile phone, no edge camera right now is
as powerful as let's say CHET GPT with GBT four O,
that's sure.
Speaker 1 (26:34):
I mean, you have this sort of spectrum where you
have a whatever one hundred million dollar data center on
one end and one hundred dollars camera on the other
and some things in between. Right, So what can you
do on the camera now?
Speaker 2 (26:46):
We do things like privacy, so we blur people's faces,
we blow license plates because we never want to upload
this information. We do weather detection what I said earlier,
with rain, so we detect on the cameras if it's raining,
things like that. We detect traffic signs and see if
they've changed. So we actually already run large part of
processing on the edge and then only if something has changed,
(27:10):
then we upload some data into a validation on the server.
But that already allows us to reduce what we upload
by more than ninety percent to.
Speaker 1 (27:18):
Just tell the server if something is different.
Speaker 2 (27:20):
Basically, yeah, exactly.
Speaker 1 (27:22):
So I know, you know, we've talked mostly about maps,
but obviously GREB is doing a lot of different things.
Are there other parts of the business that we should
talk about?
Speaker 2 (27:32):
The business is so fascinating, so if I have time,
we should talk about all of our business.
Speaker 1 (27:36):
Yes, just tell me what's one other sort of frontier,
one other thing you're working on.
Speaker 2 (27:41):
I think the other thing which we've really invested deeply,
which is also quite closely connected to our maps, is
when we look at across the delivery journey. The other
part where a lot of time is spent is in
the merchant cooking and preparing the food. And that's an
area where you spend a lot of time optimizing together
(28:01):
with the merchants to make sure that the food is
prepared in the right time, that the driver arrives in
the right time. So you've done lots of lots of
cool things. So, for example, we've built a data science
model that accurately predicts at every time of the day
how long the merchant needs to prepare an order. So
we can detect the busyness of the merchant and say,
(28:22):
if we know all the merchant has gotten a lot
of orders, normally they take to prepare, They prepare to
bud me in like three minutes, but when they're very busy,
they take seven minutes. And the merchants don't want that
our drivers crowd their store and weigh in troves, So
we typically allocate the driver only when we know, okay,
the food is ready in seven minutes, and we don't
(28:42):
allocate the driver immediately, but we know all there's a
driver two minutes away, we just allocate them like five
minutes later so that he's in the shop just in time.
And that also cuts the delivery journey, makes a lot
more pleasant for the merchant not having like a lot
of people crowd their store, and allows us to offer
more affordable price to the consumer because they don't need
to pay for all these minutes of the driver waiting.
Speaker 1 (29:03):
So just all these optimizations, all these different margins where
you can optimize. So if we think about whatever five years,
ten years out, when you think about this sort of
medium to long term like what's your dream, Like, how's
it work? What's going on?
Speaker 2 (29:20):
I think like for me, the world is changing so fast.
Five to ten years prediction is really really hard with
all the AI advancements.
Speaker 1 (29:26):
How far you want to go?
Speaker 2 (29:27):
Two years? Four years?
Speaker 1 (29:28):
You tell me, like what you just give me a
dream for the future.
Speaker 2 (29:31):
I mean the obvious one that I'm very passionate about
is robotics that make so much sense in our marketplace.
So if you can add robotics to our marketplace, that
will be a huge, huge change and something we're actively
working on to make happen. So that's I think probably
the thing I'm most passionate about to change.
Speaker 1 (29:51):
Tell me what you're actively working on with robotics.
Speaker 2 (29:54):
We haven't shared much which we do on the delivery side,
but for example, I mean things like autonomous cars. We've
signed an agreement with a bunch of like autonomous car
providers to come with us to Singapore and so on,
so we'll do a bunch of things in that space.
But basically you can imagine that it will be in many,
many parts of our delivery chain.
Speaker 1 (30:14):
You need to build an autonomous scooter based on what
you've told me, right, I feel like the analogy to
the map story is the autonomous scooter that can go
down all the alleyways.
Speaker 2 (30:22):
Great, if you ever want to have a job in
our product team, to join us, build autonomous schooters with us,
because you're exactly right, but you need to build products
that work in our region. So that's exactly the right
mindset that we're trying, like what we say always and
we want to build hyper local products that work in
Southeast Asia.
Speaker 1 (30:42):
We'll be back in a minute with the lighting round. Okay,
let's finish with the lightning round. So I read on
your bio you write that you mostly travel between Singapore,
(31:03):
San Francisco, Berlin and Cluge. Tell me about Clue the
other three I'm familiar with Clues. I don't know anything
about How does that wind up on that list?
Speaker 2 (31:14):
As a city in Romania, in the heart of Transylvania actually,
and we have a we have a small engineering center
there in our maps team, So I've spent a lot
of time there because for my own startup, that's that's
how I wound up Inclusure. Originally for my startup, Scobbler,
which was a maps and navigation startup. I built a
(31:34):
engineering team and cluge, so I've been going to Clusion
working with engineers there for the last fifteen years or so.
What's clue Like, it's fun. It's a student town, so high,
high energy, young population, very smart, the computer science department
very good. They used to clone the IBM mainframes for
the Soviet Union, so a bunch of hardcore engineers.
Speaker 1 (31:58):
You also write in that bio you wrote in Berlin,
I experienced some of the best parties ever. Tell me
about a Berlin party.
Speaker 2 (32:08):
That was prior line, That was before I moved to
before I moved to Southeast Asia. But Berlin was a fun, fun,
fun journey for I lived there for five six years.
Speaker 1 (32:17):
Maybe he didn't tell me anything about any party. So
and then you also say you lived and worked in
Singapore and Berlin and San Francisco, and I'm curious, like,
how is sort of whatever professional culture, work life different
in those places. What are the sort of striking differences?
Speaker 2 (32:35):
I mean, Germany is extremely direct, but like that's where
I'm originally from. I'm originally German, and Germans are known
to be super super direct, which in Southeast Asia doesn't
work super well. If they're not accustomed to it. So
I think for me, the thing that I needed to
adjust most is to become a lot more indirect, to
(32:56):
become a lot more like have those conversations and like
smaller groups, not in front of everybody. So I definitely
had to just my style quite a lot when moving around.
But I found that always incredibly fun. I always loved
learning about new cultures, So after some adjustments, I really
really like it here.
Speaker 1 (33:16):
That is super interesting. Was there, as you were saying
that I was wondering, like, was there a particular moment
when you realized, Oh, I'm not behaving correctly in this
cultural context?
Speaker 2 (33:31):
Yeah, I mean this predate grap So this was when
I was first doing business in China. So I sold
my startup to a Silicon Valley company, but I also
got a two hundred people team in China report to me,
which was the first time that I ever managed a
team in China, and I was extremely surprised why they
wouldn't tell me about all the mistakes, all the things
(33:52):
that I did wrong. And I said, like, that's very unusual.
Normally I get a lot of like pushback from engineers.
And then I started like asking people is why don't
they do this? And then people said like, well, it's
kind of root in a public forum to say the
boss's wrong. And then I realized, like, oh, that's that's
(34:13):
why I just need to like change how I'm asking questions,
how I'm managing teams. And that's when I first managed
teams in China. Took me quite a while to figure out, honestly.
Speaker 1 (34:22):
And how does San Francisco fit in relation to both
working in Berlin and working in you know, China and Singapore.
Where's the US on that continuum?
Speaker 2 (34:34):
Do you ask for me? I think the thing that
I really loved in San Francisco was just the craziness
of ambition. When I spent like April back in in
SF four month to like vibe code and hack on
a bunch of hobby projects, and like you go in
every random coffee shop and there's somebody who said they
work on a billion dollar idea and they they're just starting,
(34:55):
they have nothing yet they're just like, Okay, I'm going
to make it big. So I think that that ambition
and that willingness to openly declare it even if people
know it's super unlikely. There's not thousands of people who
start billion dollar company, but in the Bay Area that's
thousands of people who say they will and this conviction
and this like optimism. I think that was for me
(35:16):
one of the most striking things in the US, and
that I still love the energy whenever I go there.
Speaker 3 (35:22):
Basically, Philip Condall is the chief product officer at GRAB.
Speaker 2 (35:33):
Please email us.
Speaker 1 (35:34):
At problem at Pushkin dot fm. We are always looking
for new guests for the show. Today's show was produced
by Trinamanino and Gabriel Hunter Chang. It was edited by
Alexander Garreton and engineered by Sarah Bruguer. I'm Jacob Goldstein
and we'll be back next week with another episode of
What's Your Problem.
Popular Podcasts
New Heights with Jason & Travis Kelce
Football’s funniest family duo — Jason Kelce of the Philadelphia Eagles and Travis Kelce of the Kansas City Chiefs — team up to provide next-level access to life in the league as it unfolds. The two brothers and Super Bowl champions drop weekly insights about the weekly slate of games and share their INSIDE perspectives on trending NFL news and sports headlines. They also endlessly rag on each other as brothers do, chat the latest in pop culture and welcome some very popular and well-known friends to chat with them. Check out new episodes every Wednesday. Follow New Heights on the Wondery App, YouTube or wherever you get your podcasts. You can listen to new episodes early and ad-free, and get exclusive content on Wondery+. Join Wondery+ in the Wondery App, Apple Podcasts or Spotify. And join our new membership for a unique fan experience by going to the New Heights YouTube channel now!
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!