From Uber IPO to $1B AI Fundraise: Josh's Secrets to Success

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Josh, thanks so much forcoming on the podcast today.
It's great to have you.
Um, super excited to be here.
I've loved hearing about all your stories.
And when I think of Uber and TravisKalanick, one thing that always
comes to mind is his expressionaround operating at the red line.
And, and continually pushing your limits.
And so I'm sure you havesome good stories where
you push your limits.

(00:22):
For sure.
I feel like I'm alwaysoperating a red light.
What's, what's the worst?
I wouldn't say that there'sanything that's worst about it.
Sometimes it's the frustrationof everyone on your team and you
seeing that they're always at thered line as well, and trying to.
incentivize them and keep them goingand, uh, and seeing that vision that

(00:47):
you're seeing and ensuring that,uh, you accomplish everything you
need to with that limited amountof resources that you're provided.
Like, that's the worst piece of it morebecause yes, you'd love to go and give
everyone tons of people, you know, timeback, but at the same time, then you're
just adding cost and adding inefficiency.

(01:09):
Like there's a certain pointwhen you balance your people
versus what needs to get done.
And if you add too much people,it just adds inefficiency.
So that's it.
That's the worst part.
The best part of it is just,I love operating like that.
Like every day.
Something new and whether it'sfirefighting or just something else that.

(01:32):
All of a sudden needs to get done.
Like that's kind of fun because itadds, it reduces the monotony and makes
it so that every day is something new.
And it's like, Oh, great.
This is a, this is anexciting journey today.
What's going to be tomorrow?
You know,
that's great.
Now, can you walk us throughwhat the IPO process is like?
Oh boy, I can talk about twodifferent IPOs and it, it differs.

(01:56):
And that's because whether ornot you are actually prepared.
One.
When I was at Twitter, they werevery much prepared into correct.
I was at PricewaterhouseCoopers,Twitter was the company, but I did
five years of robotics and then movedit into the IPO after that, all the
quarterization, all those types of things.

(02:17):
And they were very, very prepared.
They spent tons of time, tons ofmanpower and just planned ahead.
We had done three different.
Uh, even more than that, but just dryruns to make sure everything worked well.
Now, fast forwarding to beingat Uber and then kind of running
that IPO, very different.

(02:40):
We've got through tonsof leadership changes.
And then I was asked by this temporaryCFO that we had, like, let's start
getting prepared and let's do anassessment and let's get things moving
so that we can understand how far thatend goal is from where we're at now.
And so we brought in external big fourfirm that did the assessment, worked

(03:04):
with many others, had probably over 400different items that were on the list
of, hey, if you wanted to go public,here's all the things that you should do.
I would say most of them weren'tanything new that we weren't aware of.
And so from there, we started,we created a project management

(03:24):
team or Styrco that reported intokind of the A team or your A CEO.
And we just went through, we prioritizedand honestly, we went from 400 and then
said, we don't need to do all 400, likeway too many, prioritized the long poles.

(03:46):
And then I went forward and had a prettyrigorous process where we would check
it once a week, and then we would havea meeting with them once a week as
well, and we'd go through, make surethat priorities were still priorities
to them, uh, and then check it and makesure that they were making progress.

(04:07):
And we did that for six to nine months.
I want to say just constantly goingthrough setting the different phases,
what comes in phase one, phasetwo, and IPO has different phases.
And so there's things that youneed to do that are important.
Post IPO that is now public.

(04:28):
I got to operate like a public.
I actually have to meet all my deadlines.
I have to close my bookson time, all those things.
So you.
They, you know, they might not matter asmuch to you when you're pre IPO, but post
IPO, all those things come into play.
Right.
And so those you would throw in this kindof your later phase of accomplishing them.

(04:48):
Phase one are things like, well, actuallyneed to have a set of financials.
I need to have them audited.
I need to have quarterlyfinancial statements.
Those need to be repeat reviewedbecause as part of your IPO or S1
process, You're going to be askedfor comfort around those numbers
as well as your audit firm will be.
And so there's a lot of preparation, uh,kind of in that initial phase of getting

(05:13):
everything together, thinking through,Hey, what are my segments as a company?
Oh, how about EPS?
What do I need?
You know, if I had tons of differentfinancing rounds, as well as issue
tons of convertible debt, tons ofother things, all of that, like makes
a lot of complexity when you get intoyour EPS calculations and so getting

(05:33):
in front of that super important, butI mean I could go on and on There's
a massive list basically But what Iwould always recommend to someone is
going through that process of saying.
Hey, what are our long hauls?
What are the things that we need to do?
Engaging one of the big four firms whoeach of them has a consulting arm that
does exactly that will come in Makethe assessment, tell you where you're

(05:58):
short, and then putting together a teamof people that can work on a regular
basis to accomplish all of those things.
I even at Thumbtack, one of thethings we did is we created a
steering committee as well, becausewe were marching towards IPO.
Obviously the market isn't greatright now, hopefully someday, but

(06:18):
we found that to be super successfulwhere you had a steering committee
that was made up of the executives.
And then from there, you had a leader ineach of your departments or divisions,
GNA, within GNA, you have a legal person,you need someone in the finance team,
you need an investment, not investmentbanker, they're finance people, but, uh,

(06:41):
someone in IR, investor relations, andeach of them have a, a plate at the table.
And.
With that, they, and they come to themeeting saying, here's what I've done.
Here's where I need help.
And then you have the executives that arevery supportive to say, okay, I see that
that is something that we need to do.
Here's how I'll help you.

(07:02):
Or here's, you know, they'll provide someof the guidance or at least arms and legs.
If you're falling shortbecause of whatever might
be happening, whether it be.
Hey, we got pulled in because of arevenue matter that we're dealing
with and we can't get to ours.
Okay, well, let's get you a contractorfor a couple weeks or something in order
to give you the arms and legs to getthat accomplished, whatever it might be.

(07:24):
And I actually replicated that fromUber, where we had, uh, actually three
different, uh, We had one that was focusedon control specifically, one that was
focused on IT controls, um, and that'sbecause Uber was fully built, internally
developed, I should say, hundreds, well,thousands and thousands of engineers went

(07:44):
into the development of the applicationand all the systems that fed all the
information from a financial perspective.
And then we had the overall, Hey, forIPO, like you need your HR policies in
place that support more of a corporatecompany that is public versus a private
company, which sometimes you don'teven have document policies in place.

(08:06):
So we split it into thosethree different committees.
Each of those committees had a headand then all of that committee,
um, reported up into myself.
And then ultimately we reportedinto that executive team and, and.
We actually, or because of that Iwould say, we got a lot of support from
people when we needed it because theyrealized, you know, different things.

(08:29):
So we, we use the green, yellow, red,and we built these beautiful dashboards,
you know, and all those things sothat everyone can see progress as well
as the, the red stands out for sure.
And yeah, you're blockers.
And, and so we went through thatprocess and, and it worked well and
we got the resources as we needed itto ensure that we could be successful.

(08:49):
And the fun fact, sorry, I couldkeep going forever on this.
So you could also just cut me off.
But one of the fun factsabout the Uber IPO is.
I don't know if it's a fun fact, but itwas during the time when the government
shut down when Trump was the president.
And if everyone remembers, like, they'rekind of heading into this stalemate where

(09:09):
everyone's not willing to do anything,and then the government shut down.
And we had just filed our initial S1, and then the government shuts down.
And so we're like, okay, well,the sec is part of the government.
And so what are we going to do here?
And so we actually built like a milliondifferent timelines to say, okay,
if the government opens this, thenthis is when we would have to file

(09:30):
our next, you know, our responses.
This is when we couldexpect our responses.
Luckily, what happened was thegovernment opened up and well, to
step back, your financials go still.
Right.
And so when you file them.
If you pass certain periods, like, hey,a quarter happened, then your financials
are seen as still and you have to updatethem with new numbers in order for the SEC

(09:54):
to continue to have discussions with youand provide comments back and forth still.
And so that was the thing we were mostworried about is, hey, it's December.
Are we going to have to get ourquarterly financials for Q1 done?
Or are we going to be okay becausethey'll open up and they'll be
like, Oh, we'll give you a specialcar grace period because we know

(10:15):
that it was our fault anyway.
And luckily they said, Hey, look, we'llgive you a couple of correspondence.
Cause they opened up rightbefore our financials went stale.
And so with that, we were able to goback and forth a couple of times and
get some comments back and forth.
Then it kept the process moving,but we definitely had like a million
alternate alternate plans say whathappens if this happens or this

(10:37):
happens, you know, it was, there'sa lot of planning that goes into it.
And then I remember you tellingme a story about staying overnight
in a hotel across the street.
Yes.
There were definitely many instances.
I mean, IPOs are not easy, especiallywhen, you know, you're at this behemoth
of a company like Uber, where youhave so much data and information,

(10:57):
so many different entities, and whenyou're trying to, like, move all
of that to get it into a good spotto go public, it was a large lift.
So there were definitely, you know,Even while I was at Uber, we had
certain targets that we wanted to hit.
I usually got an Airbnb and I would stayin the Airbnb from January until at least
March usually just because I'd be workingtill 1, 2 in the morning and would just

(11:23):
rather than going home and spending, youknow, two hours commuting back home, quite
literally, it was easier to get an Airbnb.
We had nice, uh, what are those called?
Bean bags in the office.
Yep.
And I would, some people would come in.
I remember I was just passed out on thisbean bag and just totally asleep, but

(11:43):
someone came in and was like poking me.
Josh, are you alive, do you know?
And then there was another time whenit actually wasn't IPO related, but
it was when we were integrating.
We were on QuickBooks and we had 54entities on QuickBooks and literally
every single day QuickBooks wouldcrash because we had so much data

(12:04):
that we were trying to push inthere and we're running off of bill.
com.
would sync to QuickBooks and becausewe were paying so many bills, it would
just break QuickBooks all of a sudden.
So then, yeah, the CTO of Uber, Tuan,would call up his buddy over at, uh,
Intuit and be like, Hey, Can you guyshelp us out and fix, fix QuickBooks

(12:25):
so that we can start working again?
And so we got to the pointwhere it's like, okay, we
got to get a new ERP system.
And so we went with Oracle.
And as we're implementing Oracle, weactually had to go half day by half day
or day by day and pull the data out inorder to get the data out of QuickBooks
because there was so much data in there.

(12:46):
Um, versus, you know, I would alwaysrecommend, Hey, just pull a month at
a once or whatever, and we couldn't dothat, otherwise we would break it again.
And so we would spend superlate nights because we were
given a six month time period.
Usually you would take 12
months to work.
Yeah.
So we said, Oh, you know, we'reUber let's do it in six months.

(13:07):
And so we, we did just that and weworked, we brought a ton of contractors
But they, all they did was literallyjust pull data and then provide the
match between QuickBooks and Oracleto make sure that, um, we had, you
know, imported all of the correct data.
But there was one particular time Iremember it's like three in the morning.

(13:27):
I'm just like, it makes no sensefor me to go home right now.
And then I was like, well, I want tobe frugal for the company as well.
So I don't want to go get a hotelfor three hours until, you know,
I had an eight o'clock meeting.
I was like, it makes no sense.
So I went, and we had a nap room, andso I went, I laid in this nap room.
And, you know, I'm just starting tofall asleep and everything's going good.

(13:49):
And then all of a sudden, boom,all the lights come on again.
I'm like, what the heck?
And it was then that I realized thatthe lights on the inside of the nap
room were also connected to like thetriggers or the sensors of motion.
So every hour.
The guard would walk through and allthe lights would pop on and just be
like, you know, as I'm dead asleep.

(14:11):
And then all of a sudden I'm waking upagain and then trying to go back to sleep.
That was a long, long, butI'll never forget that.
Don't have anything like that recently.
Yeah.
And Oracle in six months, I mean,that's just the impossible feat.
And so how did you guys plan that out
and execute?
Well, a lot of it was bythinking through, okay, okay.

(14:33):
How many ENT entitiesdo we need to integrate?
Because we actually had internationalentities, and the international
entities were actually kept on a wholedifferent system, which was called
like Intact or something like that.
And so we had to export outof that, all those entities.
And so at first we're like, okay,we're gonna do every single month,

(14:54):
then we'll take that month, we'llcompare the month results to what
we put into Oracle for that month.
And then we'll ensure it all matches.
And so we, that's how we started.
It wasn't an issue for allthe international entities.
I was able to have a list,just pulling them up at a time.
Like I said, likeQuickBooks, it was an issue.
We couldn't even pull a day at sometimes.

(15:15):
So we had to go account by accountand just try to pull out the data.
So that was the approach.
We had a gate, we brought in a PM,Walter from Amsterdam who came in
and he basically said, okay, this is.
What we're going to do, and hewould track every single thing.
Okay, for, you know, Uber, BV, did weget this month, this month, this month?

(15:35):
And did it match?
And we would go through thisconfirmation process to validate
and ensure that as we work through.
But initially, we scoped outand said, how many years of data
do we actually need in Oracle?
We wanted to go all the wayback to when Uber was founded.
Or do we just need like threeyears of data because you only
need that much for a good trend.

(15:57):
We decided that given the large amountsof data, we would just go with three
and then we could use the old systems,uh, if we ever needed to, which was
going back into QuickBooks and same.
So we actually paid like, you know,500 bucks a month or whatever to
keep QuickBooks live in case weever needed to go back to grab what
the heck happened three years ago.
But yeah, as we went through that, wewere able to move quickly and then you

(16:20):
can assign, like, hey, you're goingto do BB, you're going to do the, you
know, France entity, and you're goingto take the US entity, and, um, so we
had individuals assign to each entity.
And then we would pull all the data,and then we had the consultants who
would go, perform the reconciliation,import it into Oracle, and then
we would work month by month.

(16:43):
Once we had a month, then we could gothrough a consolidation process, too.
And it was then that I would say theycould see all the errors that I made
in my manual Excel consolidations.
It wasn't that bad, but, you know.
Mistakes happen, though.
I know I've migrated clientsoff of QuickBooks before and
it's like oh, here we go.
Especially QuickBooks isnot a consolidation tool.

(17:05):
So usually you get to consolidate itwill probably never be the most accurate.
And that's why we pulled out of QuickBooksinto Excel and consolidated 20 plus
currencies and 54 entities all in Excel.
Yeah, it was quite the process.
Believe me, we've had a lot of errorsas we, you know, went through that.
I still remember the day when.

(17:28):
I had done it.
I went to the CFO, um, at the timethis was like super early and I
owned like Technological Accountingand Consolidations and all that
right when I got to the program.
So I consolidated it, I'mlooking, I'm like, this is great.
And the CFO wants to meet with me.
I'm like, okay, let's do this.
I walk in the room.
The first thing he says is, Hey, thisnumber right here looks like super weird.

(17:49):
It doesn't look right.
And I'm like, No, I don't wantto say I made up an excuse, but
I basically was like, no, this,this is why it looks that way.
Why, you know, it's presentedsuch a media, lots of media
right back to my desk.
I'm like, okay, I got to look at this.
I look and there was a footing error.
And it just, so then I went backto him, you know, take that walk of

(18:11):
shame, am I going to get fired today?
And so I'm going back to him.
I'm like, look, you're right.
I've looked into it.
If this is what happened,it's like, okay, thanks.
Right.
You know, he's got that lookof I knew you were wrong, but
I just didn't want to tell you.
Yeah.
You learn.
It was definitely anexperience I'll never forget.
And I always double check on all of mywork now, even triple check sometimes,

(18:32):
just because of that experience.
That's, I don't know.
I wanna say that's one of lifelessons that always impart to someone.
It's good to make mistakes'cause you learn from them.
Don't ever make the same mistake twice.
I, I like to call thoselessons on the hard way.
Yes.
Because you only do it once . Yeah.
And you know never to do it again.
Yep.
Um, and that's, uh, how I try torun my teams always as well is

(18:55):
like, it's fine to make mistakes.
'cause the, you know, even think ofyour kid, you, they make mistakes.
You don't always sit there and I mean, Idon't know, some parents or the helicopter
parents that come in and every mistake,
yeah,
but others are like, yeah,what do you learn from that?
And what are we not going to do again?
And there's a lot of thoseat fast paced companies.

(19:17):
Cause you're just trying to.
Strap on band aids and hope that,you know, everything doesn't explode
as you work to get better processesin place and you make mistakes.
Yeah.
And I can only imagine like goingthrough that oracle implementation,
getting live, what your, your golive criteria must have been whittled
down into critical must haves.

(19:39):
And so what did that go life look likeversus where you want it to be eventually?
Yes.
I mean, on go live, the things thatI would say were lacking that we just
didn't have time to do were some ofthe reporting functionality as well.
Like Oracle is, Oh no, it'slike a behemoth of a tool.
It is really not that great.
Right.

(19:59):
It's not so bad.
Not the Oracle here, but, um,yes, this was on premise too.
It wasn't the cloud version back.
It didn't have a cloud version.
It was on premise.
And so.
We just didn't have time tocreate all the reporting.
We made it so they could do a simple,like, just whole P& L that gives you

(20:22):
a column, but then we're like, okay,we really need the reporting tools.
Then we said, okay,we're going to get HFM.
And we went through the process of scopingthat out, and then adding on HFM, and
then that's eventually when I would saywe got to a much better spot where we're
like, okay, this is much more complete.
Pull data we've manipulated in ourreporting and have much more ability to
actually see what's going on with thebusiness and pull reports that would

(20:46):
enable us to do that versus when wewent live or closed, like, oh, great.
We can have a consolidated P&L, but can'treally dig too much into everything until
I can actually find something we candrill in and give me intuition and a much
more legible fashion, you know, I don'tknow how else to better describe it.

(21:08):
But I mean, muscle wasjust that reporting.
I don't think there was really anything.
I think we would have loved to havegotten more information in there
besides just three years, but I think,especially like FP&A, you know, they want.
Because when you're going from QuickBooks,which is like very basic and you're
moving to Oracle, you have tons morestrings within your account, right?

(21:30):
So you have the account itselfthat you have like department costs
and all those types of things.
And a lot of that we didn't want, butthat FP&A wanted, obviously the FP&A
will take everything they can possiblyget so they could dice everything up.
And so that was one of the other thingsis we had to make some compromises there
because they wanted the full kitchen sinkand we weren't willing or didn't have

(21:51):
the time, six months to do all of that.
And so that's why we ended up with threeyears and then we did end up adding a
bunch of the departments and things likethat and making sure that we can get all
the information possible to enable themto have a lot more historical forecasting
ability and by being able to see a lotof the trends and what's going on with

(22:11):
the various departments, et cetera.
That's always a give and take between,I would say finance and accounting is.
Yeah,
exactly.
They're like, what's the source of truth?
I'm like, uh, right now at scale, right?
It's like, what's the source of truth?
I'm like, NetSuite is my source of truth.
Like not Adaptive.
That is you do a lot of stuff in there.

(22:33):
All I care about.
What's in NetSuite now?
Make sure that is correctfor GAAP reporting.
And then after that, we can talkabout whatever else you would
like to have this part, you know?
Yeah.
And then, you know, during your time,Uber diversified its revenue a lot and
expanded into Uber Eats, and Freight,and all other kinds of business.
Jump.
Jump?
Yeah, bikes and scooters.

(22:55):
We even had vehicles, and we purchaseda ton of vehicles because someone
had a brilliant idea, and I honestlythink it was a good idea, but, uh,
it was by, uh, you know, We had,I just want to say at one point in
time, almost a billion dollars ofvehicles, how I have books, probably
close to that, that we had purchased.
And then the thought process was tolease them to drivers that potentially

(23:20):
Couldn't, you know, afford to havea car or we had quality standards.
And so someone was like, Hey, youcan't have a 1994 Honda Accord.
and use that for Uber.
We want it to be good experience.
And so we would lease the car to themthinking, well, if they drove off the
platform, In our cars, then we couldjust not confiscate, but essentially

(23:43):
you could just take a little bit ofthat earnings and then pay off the
lease that they would else the carneedless to say that didn't work.
It's like subprime lending, you know,that didn't work out for the U S really.
And it's basically the same thing.

(24:03):
And so eventually thatwas sold off as well.
But, um, yes, sorry.
You were talking about differentrevenue streams and it got me
thinking of all the crazy things.
But like with jump bikes, it was like,okay, we have bikes and scooters.

(24:24):
How are we going to track it and makesure like they don't just disappear.
So we had GPS tracking on.
But then what would happen is whenthe GPS or the battery died, the
GPS signal would just disappear.
sure you could get like,okay, where's the last ping?
And you could go look for that area.
We would definitely find themup in like scooters and bikes up

(24:44):
in trees and things like that.
Like people would justchuck them off there.
So there was an evolution too, though.
I mean, that's what you do as a business.
You learn from your mistakes.
And so we're like, okay.
Initially it was like.
We would have to go out, collect allthe bikes, bring them back in, charge
them, and then put them back out.
And so we had these people that wouldgo and literally drive vans around,

(25:07):
pick all of them up and do that.
And I was like, wait, this isinefficient and not the, you
know, yeah, and expensive.
So let's make batteriesthat are swappable.
And then we could bring chargedbatteries and just swap them out.
That's a brilliant idea.
And so that's what they started doing.
And it made it so that, uh, One,you could continue to generate

(25:27):
revenue from having the scooters outand not bringing them all back in.
But two, then we had a lot moreinventory that we had to count
as a company and keep track of.
And so that made counting evenmore fun as trying to check what's
the value of the bike, what's thevalue of the battery, and so on.
Then we had inventory accounts whereeven half of the accounting and the

(25:47):
internal control's team went out andcounted because, um, at one point time
we got to where we didn't know howmuch we had That always reminds me and
so you guys had to,what types of stuff did
you have to count?
I mean, luckily, well, we did count cars.

(26:08):
there was a time, we'll call it earlyUber days so that no one can sue Uber for
this one, but we didn't even know whatour fixed asset ledger, like we had to
rebuild the entire fixed asset ledger.
We'd purchased so many cars.
We had, I mean, we even had abuilding that we had purchased
and We just did not know.

(26:30):
And so we had to go through this process.
Uh, we, we hired Protiviti to come inand we formulated a plan of how are we
going to rebuild our fixed asset ledger?
And so we went through a processof looking at old invoices to come
up with like a cost value, andthen we did an average cost value.

(26:51):
And then we went and sent peopleout, we counted the cars, we had cars
that were over in Singapore becausethat was a big, uh, Asian market.
And then we had cars all over the U.
S.
And so we had Protiviti who was, youknow, fairly global as well, they went
around, they did all these counts.
And with the bikes and scooters, wehad to go through the same process.

(27:14):
And then, we went to the buildings,we counted various furniture items.
And then with the laptops, we were able tojust go and say, Hey, but then we also had
server rooms, so we had to send people outinto these secure data centers, and counts
over like how many servers, how many ofthe peripherals are there, et cetera.
And, From all of that, we thenwould say, okay, what's the

(27:38):
average age based on our ledgers?
And then we would say,what is the average cost?
Well, we basically reproducedan entire fixed asset ledger.
And we were able to get the auditorscomfortable with everything that we did.
They would ask questionsto say, how about this?
We could go and give them proof andevidence of, but that was like a.

(27:59):
At least a three to four monthprocess where we had, it was like all
hands on deck trying to figure out.
Cause you can't, can't go through anaudit if you can't say, Oh, here's the
amount of fixed assets that I have.
And so that was a bigjourney in and of itself.
And I wrote this whole memo toget the auditors comfortable.

(28:20):
I guess it worked.
I don't want to say those are some of thepain points that you run into as well.
I mean,
as a company grows as fast as Uber did.
It was next to impossible toget all the process in place.
I mean, you're launching a new city,all of a sudden you're buying a bunch of
other things, you're opening a new officeand all of a sudden you have all this

(28:42):
furniture and sure as the accounting team,you're trying to stay in front of it.
But when you're running at thered line and not having all the
assets that you need or the peoplethat you need in order to do that.
Yeah, you lose track ofsome of those things.
We did outsource a lot of it over toIndia to help at that point in time
to where they could actually processthe invoices and things like that.

(29:04):
But that's where controls come into placebecause what we found is that They could
potentially enter the same invoice twice.
But what would they do to get around itis they would go up and say, Oh, if the
invoice number is one, two, three, four,it was a one, two, three space four.
And then it would showup as a new invoice.
And so then all of a sudden,you know, you're there.

(29:28):
You could double pay someonebecause someone had entered in that.
And so then we got, went through a wholenew process of saying, okay, How can
we mitigate these types of things, stopit, and some of it's education, but
some of it is like adding in certaincontrols, checks and balances, and, I
mean, and, you know, talking about PACE,that's what happened there, I think.

(29:50):
A lot of it was because of such a fastpaced culture environment and so demanding
that you just had to get things done.
And everyone ran and sprinted.
And then at the end of the month, you'relike, okay, where do we get everything
that we need to, to be able to figureout what the heck we just purchased,
you know, and all that type of stuff.

(30:10):
And so that everyone's running around.
And that was like early, early Uber days.
And then eventually it was like, okay,we need to put a place I called the PPS,
the people process systems in order tostart getting towards where we could
actually go through an IPO processand be an effective public company.
But, uh,
Well, and then, you know, therewas, there was that divestiture as

(30:34):
you guys got into China and thenrealized that model wouldn't work.
Could you shed some coloron that journey as well?
Oh, yeah.
I mean, China has one of thelargest markets for Ubers in a way.
You have a massive, dense populationof people that use taxis and other
forms of transportation to get around.

(30:56):
it was on Travis's list ofsomething that, like, we gotta
go after, we gotta go into China.
Which makes sense.
It's, it's probably one of thelargest markets in the entire world.
And so we went into China andspent tons and tons of money.
What we would do in order to win overa new territory usually is you would

(31:17):
just pay tons of driver incentives,rider incentives in order to incentivize
people to start using your product.
I would say the issues that yourun into is the potential for
fraud when you're growing so fast.
When you're trying to pay everyoneanything to try to get them to use your
product or drive for you, people aregoing to take advantage if you don't
have the right checks and balances.

(31:39):
We did learn that, but we didput in place a lot of checks and
balances subsequently and were ableto mitigate a lot of those things.
But yeah, we ended up spending severalbillion dollars winning over China.
And ultimately, uh, DiDi came andsaid, Hey, look, like we're basically
going to sit here and fight each otherand spend money to fight each other.

(32:01):
And why don't we just call a truce andwe'll acquire your Chinese side of things.
And so we went through that process,determined, you know, negotiated,
determined what the value was, andthen ultimately divested China.
But interesting enough,I was in Amsterdam.
I was the EMEA controller atthe time when we were divesting.
China.

(32:22):
And so I would have calls like earlymorning during the day I would work and
then late at night and just working, youknow, every hour of the day, basically
to work through that to divestiture.
I mean, it was what it was valued.
It's 9 billion initially, andthen I think it became 6 billion.

(32:43):
And obviously it fluctuateswith the value of DiDi.
Um, so we have a stake in it.
Um, but yeah, it was quite the processbecause it was a very different market.
And I think that's what we determinedor saw is that each market was very
different and how you would go about.
You know, when you go over the marketor even, you know, just gaining drivers,

(33:07):
uh, we used to have a launch team thatwould go out and would hit up all the
airports and just sit and talk to all theblack car drivers or try to talk to the
taxi drivers or anyone like that and belike, Hey, do you want to drive for Uber?
It's this new company andthat's how they want it.
So I say that only because,you know, it's a business.

(33:28):
Sometimes you just have to do a lotof different crazy things in order to
win over and influence your market.
And so the same goes withaccounting, you know, sometimes
in order to just get things done.
It means jumping in all heads on deckand doing some crazy things in order
to get the books closed, but it works.

(33:49):
And so leading that divestiture,what, what does that process
look like from an accounting
standpoint?
That one was actually interesting aswell because we hadn't fully separated.
Meaning there was a lot of support upat the parent company still for all the
operations that were going on in China.

(34:09):
In order to divest, you need to have, youknow, a clear set of financials that is
like, they call it a carve out, right?
So we had to go through a carve out andwe had to work through the entire set of
financials to determine ultimately whatthe carve out set of financials that DiDi
was receiving as part of divestiture.
So, We brought it, I mean, I had awhole team on the technical accounting

(34:34):
side, but we also had to bring in, um,other consultants because of just the
lift and the, you know, with everythingin Uber style, it's like, Hey, we've
got to get this done in like a month.
And so I was like, okay, uh,let's see what we can do.
So we went through that process,spent days looking at like different
balances, different receivablesand going through kind of every

(34:56):
single line item, speaking of.
Okay.
What do we need to do here?
And how do we need to do it?
What's our time frame?
And so we went through a projectmanagement process of working down
each one of those, assigning them outto different people and then having
them work through, and ultimatelywe're able to get to where we had a
supportable carve out set of financialsfor the historical period that

(35:20):
DiDi could then leverage in there.
Filings and both all the testsfor acquisition purposes.
But,
and then how does the tangible
divestiture look,
you know, because there's, there's thecarve out financials, which is really
numbers exercise, but then you had anumber of cars and technology and IP
and all of that good stuff like people,

(35:40):
honestly, well, with people, they sat downand said, Hey, look, we're divesting and
now you're going to be a part of a DiDi.
We spent plenty of timeworking with DiDi's.
Uh, finance team and accounting team thatkind of helped them walked it through,
like what the different account were,different accounts were in the process.

(36:00):
But then to your point, like from asystem perspective, we didn't want
DiDi to have access to all of the data.
So we had to go through a process ofkind of in a way, carving out that data,
sticking it onto a separate set of serversand all that type of stuff so that they
could still access some of your data,things like that, which adds value to

(36:21):
have some of those things, but it enabledus to put it onto a separate site.
And then once we were done, we all almostjust closed it down or closed off our
access so that then it was kind of almostall theirs, um, subsequent to that.
So we had to do that witha lot of different things.
And then even like leases or things likethat, that you have for all your different

(36:42):
campuses over there in China, we had togo through the process of reassigning
and doing those types of things.
So that's the process that we wentthrough is similar to what you're asking
in a way, you have to think throughit as if you're just giving this away.
And so what do you need todo in order to give it away?
And sometimes it wasreassigning different contracts.

(37:02):
We had to go through and make amassive list of all the contracts
and then go through a reassignment.
And that, yeah, there was a constantsupport for at least several months
after that, until they got comfortablewith everything that was going on.
And we were able to kind of moveon and just enjoy either the
valuation changes in your P&L thatyou constantly are reflecting even

(37:26):
post, uh, domestic jurisdiction.
Right.
One of the things that we also haveto think about is like, we have
an investment now and we have tobe able to value that investment.
Right.
And so how are we going to getthe information timely enough to
be able to determine if there'sa change in the valuation?
And so that was something that weactually worked out as part of the

(37:47):
deal economics as well, to ensurethat, you know, you don't all of a
sudden divest, give them everything.
And then you're like,okay, we need information.
And they're like, screw you,you know, you don't want that.
And so it's something that'svery important to work out as
part of the deal economics.
Like Okay, you're going to give usfinancial statements every quarter

(38:08):
within 30 days of the quarter thatwill allow us to then figure out, take
that, figure out what's going on andthen reflect the changes as necessary.
Yeah.
And then what kind of transition supportagreement did you have in place with them?
Four different divestitures, essentially.
One was John, one was Yandex.
When it was that and then withthat, basically we just had a time

(38:32):
and materials, that's all it was.
Literally, it was like, what we'll dois we will charge you time materials,
um, for any support post divestiture.
And we would create a bill, everysingle one, and then we would just
ship over the bill and say, here it is.
And sure, there was some back andforth sometimes saying, I have this,

(38:52):
you know, 20 hours of accounting.
That doesn't seem right.
It's, you know, back and forth,but overall it was pretty easy.
See, unless we agreed on a billing format,actually, to say who, how, what they did.
And then we would shipthat over every month.
Then we would get paid.
Yeah, so four divestitures, so youmust have some good lessons in terms of

(39:15):
what to, you talked about what are thedeal parameters, what are some others?
Well, there'd be a divestiture, I meanwe also had several large acquisitions,
and I think the The planning andtiming around it was always critical.
And I think what we learned is wealso put together almost like a
playbook, like a check checklist ofsaying, Hey, okay, it's divestiture.

(39:36):
What are we going to need?
We're going to need carved out financials,we're going to need all these types
of things in order to be able to, runthis process from start to finish.
And so we ended up building achecklist and putting that in
place and we would follow that.
Um, I would highly recommend that,especially if it, whether it be
an acquisition or a divestiture,there'd be different checklists,

(39:58):
but you will still have one, right?
For an acquisition, you'd be like, okay,well, I need to look at their assets.
I need to look at this.
I need to value this.
I need to go and get a valuationperformed depending on the size.
With the divestiture, technicallyyou potentially also need to go get
a valuation performed, but I thinkthat was the main learning is just.

(40:18):
It's always going to take more timethan you think it will to do whatever.
So give yourself a little handicapso that way you can have a little
additional time to get through it.
Create a playbook so that youmake sure that you've checked
all the boxes every single time.
And then make sure you'recommunicating on the differences.
And what I mean by differences is.

(40:38):
You could have your US GAAP books,but theirs could be an IFRS, and so
you're going to have to go througha transition process and work
through that with them as well.
And that was one of the hardlearnings, not hard learnings,
but things I've learned islike, you know, like, Hey, sure.
For DiDi is over at China.
It's there for a CubanIFRS or local standards.

(41:00):
Same with Yandex.
But when we had like jumpbikes are in their divestitures
there, they were US entities.
And so we didn't necessarilyhave that accounting difference.
But at the same time, when you'reacquiring, You are also looking at
what are their accounting policies?
Is it actually the sameaccounting policy that I follow?
And I would venture to saymost of the time it was not.

(41:20):
Even your thresholds you're using forprepaid or fixed assets or whatever,
like you're capitalizing at 2,500.
They're capitalizing at.
500 bucks.
And so sometimes there's justchanges, write offs of other things
that you're making to make theadjustments into your, uh, to, to be
reflective of your actual, policies.

(41:41):
Sure.
And on the integration side, what doyou, what would you say was the biggest
challenge of integrating your companies
I'd say there's two.
I would say on the integrationside, One, integrating the people.
You want to keep the good peoplefor a period of time, but what
happens a lot of times when theyannounce that there's going to be a
divestiture and acquisition, peoplejust are like looking for a new job.

(42:05):
And so acting quickly on that, I think itis important to make sure that the people
that, Uh, because, you know, in, in predeal discussions, you're saying, okay,
let's start looking at the good people,who are your good people, and they can't.
Who are your good people here?
And so, as you're looking, you want toact quickly to let those good people know
that we really want them to be there.

(42:26):
Because, you know, The peoplewith all the knowledge and
information are super critical.
When after you've acquired them, youdon't want to just be left alone.
And you're like, waita second, what is this?
And then you're looking around andit's crickets because everyone's gone.
Right.
And so making sure that you canbring over the correct people that
can add value subsequently to, yeah,getting ahead of the accounting

(42:48):
policies and things like that.
To re performing validation.
Uh, you know, everything that you'reacquiring, performing a due diligence,
we always had, uh, those things done.
Uh, but on the integration side,you also have to think about, like,
chart of accounts and all thosethings and how does it map over.

(43:09):
And so we had a whole entireteam that actually worked and was
designated as M&A integration team.
And they would go out and theyalso had a play book of like, okay,
let's get their chart of accounts,let's start the mapping process so
that we can ensure that we can map.
Now let's look at their policiesand determine if their revenue

(43:30):
policy, if we're on a net basis forUber, are they on a gross basis?
And do we actually believe thatthey should be on a gross basis?
Or if we did an assessment now,would we say it's actually net and
we need to go through adjustmentsand make all of those changes?
And so, There's quite a bit of processthat goes into an acquisition that

(43:50):
people don't necessarily think about.
So it just takes time to integrate,update, make the necessary adjustments.
Then again, you have to get the valuationand go through all those types of things.
And there's, there's a nice long checklistthat goes into acquiring or divesting.
I mean, it's not as easy.
It was just, you know, ridingoff and calling it goodbye.

(44:11):
Uh, there's, pre and postactivities that are taking place
and are necessary in order to besuccessful in all those things.
But it sounds like you guys had a prettystrong playbook because I mean I've seen
companies that are very acquisitive andand they just bring them on but they
don't integrate them and they have themoperating on standalone, you know, ERPs.

(44:32):
They're just upload theirmonthly trial balances..
Yes.
Sounds
like you guys fully integrated them.
In most
instances we would, but to yourpoint, a lot of companies do at
first, Oh, they're on QuickBooks.
We'll just keep it on QuickBooksrather than bringing up to Oracle.
And then over time we'll getup to start booking in Oracle.
So when we acquired Careem,it was out of the Middle East.

(44:52):
They were slightly larger and so itwasn't as easy as just saying, okay,
yeah, start using everything that we do.
They had probably 15, 20 plus entities.
And so we had to go through a processof saying, okay, chart accounts,
each of these entities is different.
And how do we think about the IFRSaccounting side of things and getting
that captured versus the US GAAP side,which we needed for all of our, you

(45:15):
know, 10 Qs, 10 Ks and going forward.
And so we, you know, we had to workthrough the process of saying, what are
the gives and takes of us forcing them toinstantly take on and start using Oracle,
or do we make it a gradual process?
So with that, we actuallydid work with them.

(45:36):
They implemented Oracle cloud, startedusing our set of trial balances, which
enabled us to then, you know, leverageeverything that they were doing.
And, and essentially take the trialbalance and created it to our Oracle
instance, then use it to consolidateand do all those types of things.
So it sped up the process because ifyou don't integrate them and let's say

(45:58):
they keep QuickBooks, then yeah, you'rehaving to go through that, that extra
steps versus that seamless integration.
And each of those extra steps, one,you need to get controls around it.
Two, are usually manual,it can cause errors.
And so you're adding in more riskassociated with it if you don't fully

(46:19):
integrate them, but there are definitely,I would say 50 percent of the time we
would fully integrate and say, okay,we're going to start booking QuickBooks.
Whereas if they're very large, youwould have more of a step process of
what creates, you Now we'll get themto start using that child bounce.
Next, we'll have them startintegrating into our Oracle instance

(46:39):
and taking those steps soon.
So, it just depends, but yeah.
What, what lessons learned,I guess, by country?
Because I know China isreally tough for ERPs.
Brazil, India, what areyour thoughts there?
Yes, Brazil, India, all of those extremelycomplicated, The amount of taxes and
special accounts that you would needassociated with just for the different

(47:02):
taxes or things that are specific tothose countries was very different.
We had a full team in India justbecause of the differences and, um,
the statutory books were so much morecomplicated and so we needed that
special team just to address all ofthe different nuances that were there.

(47:23):
We had to do the same with Brazil, whichI think everyone knows has one of the
more complicated tax structures, and theycan pay every single tax for every single
thing there is in Brazil, but, uh, yeah,so that, you know, and that's where some
of the decision, do we outsource thekind of statutory reporting or special

(47:44):
nuance things, and then companies, wedid this, um, And many instances where we
just send over, here's the US GAAP books.
We would work through withthem all the differences.
And they would go through and makeall of those changes with IFRS.
And then they kept aseparate set of books.
Um, so Oracle was the, like, U.
S.
GAAP set.

(48:04):
And then we'd have a secondaryset for statutory purposes.
Yeah.
Yeah.
Yeah.
The memories.
So, one of the things that came upwhile we were kind of moving through
the different Uber eras was, hey, youknow what we should do is we should make
it so that we have our own autonomousvehicles so that way, like you cut the

(48:28):
driver out eventually, and then you'resaving money because that's where most
of your money is going to from each ride.
And so we went through this whole process.
We first went to Carnegie Mellon and weactually acquired a piece of the company
there, uh, CRI, and with that, we wereable to get a lot of technology that

(48:52):
helped us with kind of LiDAR imaging.
After that, Travis thought, Hey, youknow, I think he met, uh, so I remember
correctly, and maybe this is just mecoming and remembering it from watching
super pumped versus my actual knowledge.
But, uh, it was at some sort ofconference and bumped into him.

(49:12):
Anthony Levodowsky and Anthony Levodowskywas someone from Google and he was
part of their autonomous vehicles unit.
It was helping develop allthis technology and Travis was
like, Hey, come work for Uber.
And eventually we did bring Anthonyover to Uber and, and he kind of helped
lead and run a lot of the developmentaround our autonomous vehicles and

(49:36):
everything that went into that.
And then.
Yes.
You know, I was part of.
So we actually acquired him and afew other individuals and things like
that because I had to work on allof the, you know, the acquisition
agreements and things like that.
I, so I remember reading this acquisitionagreement and thinking like, and we're

(49:56):
totally going to get sued for this.
You know, we're basically hiring theperson from Google that is in charge
of all the autonomous vehicles.
We're asking him to come over andbasically replicate that for Uber so
that we would be able to then haveour own fleet of autonomous vehicles.

(50:17):
And this just doesn't seem like,well, it won't end pretty, but
hey, we did the acquisition andthen yeah, a few years later, uh,
Google won a lawsuit against us.
We ended up paying for it, so I can'tsay that I didn't say that it wasn't
going to happen, but uh, it was awork, it was, you know, I would say
sometimes the businesses you take risksin order to try to make advancements

(50:41):
in areas that you want to focus on andsometimes they're large risks and end up
costing your business in the long run.
Wow, that's a, that's a big lawsuit andI can only imagine over the years, you
know, there's been a number of scandalsthat you, you led the disclosures and
financial reporting for a lot of thoseyears, you must have just been, you

(51:02):
know, catching at left, right and center.
Yeah, there was definitelya lot of interesting things
that I was privileged to.
Uh, over the course of that time, um,some are probably still privileged, but
it definitely was interesting sittingin a room sometimes and just be like,
what are we going to do with this?
And you're just like, I don't know.

(51:26):
Let me think throughthat and get back to you.
There was even a time, you know, goingback in with the cars, some of it,
it's like, Well, we, we paid, you know,millions of dollars for these, but now
they're telling us that they need millionsmore in order to deliver our cars.
And do we want to pay the millionsmore or just bite the bullet?

(51:47):
You don't want to get left.
I don't know.
Eat the cars.
Um, so you've got to go through that costbenefit analysis, but there's definitely
a lot of different decisions, um, andsometimes maybe I'm just too frugal.
Like.
But we, you know, the building inOakland that's Square now we opened
that at one point and we boughtit and we were like, Oh, we're

(52:07):
going to build an Oakland campus.
And then, you know, a year later it'slike, Man, that was a terrible idea.
We're selling the Oakland buildingand never even occupied it,
but put a lot of money into it.
I would say that was one thing thatI learned from Uber as well, is like,
focus on like your bread and butter,versus trying to do everything possible.

(52:27):
Uh, and we had that tendency, like, Imean, we had Uber rides, then we had Uber
Eats, and yeah, we had Uber Freights.
But then after that, um, we had like,Where it's like, Oh, let's do alcohol,
let's do prescription, let's do, youknow, anything basically delivery services
in New York, you could buy like shoesthat have them delivered for Nordstrom.

(52:50):
And like, there was all these thingsand each of those came, but then,
uh, we even had like Uber works.
This was a good one.
Whereas like, Hey, all ofthese like drivers have
pockets where driving is great.
And you have peoplethat really want rights.
But then in certain instances,there's just downtime.

(53:10):
So what can we do with them?
Let's rent them out as labor.
And so it was like, Oh, restaurantsneed dishwashers and restaurants
might have someone that didn't showup that needs to do X, Y, and Z.
And so let's We could charge 15bucks an hour, then have them go
wash dishes at certain places.
And so they're like, Oh, we'lljust stick them in houses and we'll

(53:31):
have them sit there in the houses.
And then we'll get a call from therestaurant saying, okay, I need a
dishwasher and then we'll send them offto go wash dishes or whatever, you know?
And so we did that for a while.
That didn't really work out.
And I guess I use that as an example oflike, We're jumping at everything that
could, Oh, we could get money from this.
We could get money from that.
What a great idea this is.

(53:53):
And you kind of lost your focus on, youknow, what is the core value of principle
that you're trying to establish at Uber?
And we should stick to that and ensurethat we're, you know, always focused on
that versus keeping it or taking youreye off the ball and losing your focus.
And I, you know, I try to take thatto Scale as well in any of my teams.

(54:15):
Just like, okay.
What is our focus and whatare we trying to accomplish?
And let's make sure that we don't getdistracted by all of those things.
You'll have tons of fires that comeup, but you need to not let them
distract you from like your end goal.
Cause if it does distract you,I mean, in the long run, you're
not going to accomplish your, thegoal that you're setting out to.
And so, and that's what I have toremind my team constantly as well.

(54:37):
so what I do now is I actually haveweekly meetings, day one, Mondays.
It's like, okay, it's a, this is whatwe're trying to accomplish this week.
And then every single day I still haveto do check ins with each one of my
directs because it's such a fast pacedenvironment and everything changes that
I might get marching orders like, youknow, five o'clock on Monday and I'm

(54:59):
like, oh, we've got to change our wholeentire approach to what we need to do.
And that doesn't mean that I stilldon't need to get my audit done,
that I still don't need to close thebooks, but I do have to switch focus
quickly in order to be able to helpthe company continue to operate and,
and, you know, meet their needs.
It's not to say it's super criticalwhen you're in such a, an environment

(55:22):
that you are communicating verywell and also questioning, do you
really need this, like tomorrow?
cause there have been way toomany times, whether it be at Uber.
or whether it be at Scale where it isso fast paced and everyone's running
like I need this get this for me I needthis that like you're running around

(55:46):
getting all this done and then youdeliver it and then they just sit on it,
you know and you're like I just workedthrough the night to get this for you
and then you sat on it for a week likeand so at least I've learned That's why
I really need to communicate and trulyask Is this really necessary tomorrow?

(56:09):
Or can I get it to you in a week?
Or when do you actually need this?
And so I've tried to instill thaton my team too, because At a fast
paced company, what happens is youhave all these operators that are
constantly, you don't even know, butthey're coming to your team and asking
questions and asking for information.
And so your role as their leaderis to help them ensure that they

(56:31):
remain focused and don't losefocus on all those other things.
Otherwise they'll work all day, all nightand not even get you what you needed
because they're focused on all the otherthings that everyone's asking of them.
And so, um, I found that having those
meetings, those touch points, even if it'slike a quick 15 in the morning, whatever
it is, can really help drive valuebecause then your team can flag, like

(56:56):
hey, Billy from ops is asking for this.
Do we really need toget this to him today?
Cause that's on my list.
It's like, no, let me go andtell Billy he's not getting it.
Let me go and tell him that we'll getit to him at a later time to be able to
keep your team focused and moving forwardon what you're trying to accomplish.
And that fast paced companies, that'slike one of the most critical things

(57:18):
is, you know, maintaining that course.
And, Um, making sure everyone feelssupported and not lost in, in the fold of
everything that's going on in the chaos.
Yeah, and, and you guys recentlygot one billion dollars this month.
One billion dollars.
Where, where do you start with that?

(57:38):
Do you have a bigspreadsheet to budget that?
No,
um What's funny is you would think,Oh, we just got a billion dollars.
We're going to go do something.
And it was more like, Oh, we justgot a billion dollars and we're
going to just put that investments.
And we're not even goingto touch any of it.
And we're going to continue tooperate under the assumption that,

(57:59):
like we need to become profitable.
We need to only burn X amount of cashflowor we need to become cashflow positive.
And so I would say the fundraisingis more like a rainy day fund,
but I don't think we need it.
And it was more of we're going to continueto operate as if none of that even
happened and just try to ensure that wecan progress towards that profitability

(58:23):
target that we had in the first place.
And that we told all the investorslike, Hey, you want to invest in us
because we actually are trying to movetowards profitability and this is our
growth plan and this is what we're doingversus just going and wasting money.
I I've seen the opposite it Uber.
I would say it was like, Oh, we'vegot, you know, 3 billion and then.

(58:43):
Hey, we just went and spent it allon fighting in China or whatever,
you know, versus actually saying,Hey, we just got 3 billion.
Let's say, you know, figure outwhat we're going to do with that.
And obviously I wasn't a partof the leadership at that time.
So maybe they're like, we need toraise 3 billion so that we could go
and fight China for those 3 billion.

(59:04):
Um, and that could havebeen the plan all along.
You know, I guess in my finance,accounting mindset, it's more like.
Let's just not go waste all this money.
And let's actually think that, youknow, pretend you didn't have it.
Something I was always taught.
It's like, even if you get all this bigcash influx personally, it's like, just
set it aside, pretend it didn't happen.

(59:24):
And you don't have it.
Otherwise you will always know itwasn't the saver, you know, but
you know, I've quickly learnedthat it's important to just.
Put it aside and pretend it never came.
Otherwise, you have a tendency to think,Oh, I'm just going to go hire 10 more
people, even though you don't need them.
And so you actually begin to drive moreinefficiencies with your processes and

(59:46):
your team, if all of a sudden you justopen up a spigot of like, do whatever
you want with all of these funds.
Versus ensure that you are very methodicalin your operational process and not
wasting any of it and still drivingthe efficiency that you're trying to
drive before you even have the funds.
I love that.
And there's, in my opinion, I'veseen so many Bay Area companies

(01:00:08):
that just waste too much cash.
And so whenever I hear a pre- IPOcompany trying to get cashflow
positive, I'm like, yes, good for you.
Although a lot of themsay that too, right?
It's like, oh, we're, we'regoing to be cashflow positive.
That's what we're going for.
Well, you have to change certain thingsin order to actually hit those targets.

(01:00:30):
And so it's, it's great to speak thespeak, but, uh, it's more important
to walk the walk and make sure thatyou're actually doing things within your
organization and finance it whether peoplelike it or not, is like the example child.
You know, it's like.
If we're going to have an offsite, okay,well, we're not going to spend a dime

(01:00:51):
on the offsite, you know, we'll eatmac and cheese for lunches so that we
could have this offsite and show thatwe didn't spend that much on our offsite
because we're finance, you know, versus.
Like legal's like, Oh,we're going to Vegas.
And as a matter of fact, legal, Iwon't say which company, but not
the one I'm at, but one of my priorcompanies, legal's like, we're

(01:01:14):
going to have an offsite in Vegas.
And I was like, man, I wish wehad that much money to just go
and have an offsite in Vegas.
Instead, we're having it in theoffice and using the conference
rooms in the office and then doinga scavenger hunt in San Francisco.
Finance.

(01:01:34):
Building a great culture.
Yeah,
it is.
But, I mean, that's the thingthough, is you have to walk the walk.
And if you want the operators thatyou're trying to control with your
financial levers to comply, youneed to show them that you can
actually operate that way and still.
You know, maintain morale in yourteam, um, even as frugal as we are.

(01:01:58):
So in your mind, oh gosh, aside from,you know, saving money with mac and
cheese with on-sites and scavengerhunts, what are some cashflow positive
initiatives you guys are tackling first?
What's like the biggest bangfor your buck that you can do?
That's a great question.
The most difficult trick is growthwhile maintaining at least, you

(01:02:20):
know, at a level cost structure.
And so how do you get that?
I guess that's the magical questionthat every CEO and CFO asks themselves
too, is how do you constantly growand not, not increase your costs.
What we are doing is, I mean, some ofus just around the financial rigor,
a lot of companies don't even havea planning process in place, right?
When you're a startup, You'll have asmall forecast and things like that, but

(01:02:44):
what we've done is made it so that weactually have, uh, a weekly rollup to
where we are actually looking at, um,what happened revenue wise, whether the
deals that were closing, what happenedcost wise, because we've actually created
SQL queries where, um, all the databasesare capturing all the information
associated with the transactions,um, which was also developed.

(01:03:08):
And then we'll SQL query it.
We'll put it like dashboards.
And essentially we've createddashboards for like everything.
So I can see if like, Hey, we'rewrapping up on a project that we call
Flamingo, and as we wrap up on that, wecan see how much revenue we're driving
versus how much costs we're driving.
And if the costs are way outweighingwhat we're actually obtaining revenue,

(01:03:30):
then we can kind of ratchet downsome of that, or look at ways that
we can be some more, more effectiveand efficient in how we're spending.
Um, some of the moneytowards that project.
And so that's one thing that we've done,honestly, is just more create dashboards.
And so on a weekly basis, we go through aprocess of looking at, you know, those two

(01:03:51):
things, and then also a snapshot of OPEX.
And so we'll look and say, okay,how's OPEX looking for this week?
And are we going to be over under plan?
And if we're going to be over orunder, over typically, We'll look at,
okay, what are some levers that wecan pull in order to be able to reduce
our spend and still hit our targets.

(01:04:14):
So, on Monday, we go through a processand we will, you know, we kick it off.
We'll get an all of our revenue and cost.
And then on Tuesday we'll build out like,okay, how's the forecast looking now?
Um, and so then we'll go through thatprocess of saying, okay, what are
levers that we need to start pulling?

(01:04:34):
And so that way by Thursday, whenwe, our CEO and all the A team were
saying, okay, well, you need to dothis and you need to do this in order
for us to kind of, Move the leversaround to meet all of our targets.
And then if things, you know, sometimesyou have deliveries that get off
track and it's like, okay, well, wewere supposed to deliver, which would
have resulted in X amount of revenue.

(01:04:56):
We're not going to get that anymore.
So what can we do?
Or you give that news to your CEOand he goes and yells at everyone and
tells them to work harder and get thedelivery on time, one of two things.
And so in our case, though, wetypically try to say, okay, if we're
not going to deliver on this, thenwhat are some other things or ways
that we can get revenue in, um, with.

(01:05:18):
You know, the team that wehave, or are there new contracts
deals that can make up the gap?
And so we, uh, so we have weeklybusiness reviews, um, as we get weekly
financial reviews, and then we have themonthly forecast, um, business reviews.
And we just go through that kindof cyclical process where we're

(01:05:38):
constantly taking snapshots and seeingwhat's going on in the business.
Um, which enables us to kind of havedifferent levers to pull back and
forth as we either need to expand orcollapse in different areas, we use the
finance team and to do a lot of that.
And then you have the accountantsthat are kind of your.
You know, you're being counters thatsay, Oh, this is where it ended.

(01:06:01):
And all of a sudden you're like, Oh crap,that's not where it's supposed to end.
Um, what can we do now?
And then you've worked through thatprocess, but I would say that's probably
the thing where the cadence that we'vegotten into, I, I, I'm not going to say
there's anything magical because I'msure any finance person's like, yeah,
everyone doesn't finance, but it's gettingyour business to operate and understand.

(01:06:24):
That that is the cadence and thenunderstand that as things happen and
you need to make changes that they needto quickly adapt and make those changes
in order for us to come and find them.
meet our targets, stayon track, et cetera.
And that also helps you identify kindof your excess spend and cut it back
to, to work towards that profitabilitythat, that you're looking for.

(01:06:46):
So if you don't have that, I, youknow, I've had the flip side and you
could shut me, shut me down anytime youwant, but I've also had the flip side.
Uber, whereas I feel like itwas like spend at all costs.
And so a lot of times you didn't havethe levers, especially initially.
And then that month, then you're tryingto pull in everything, figure out what

(01:07:08):
the heck just happened and trying tomake sure that you're still on target.
And at times you didn't necessarily carewhat happened, especially when you got all
this new fresh cash, you're like, okay.
These are the initiatives that we have.
We're going to go full inand do whatever we can.
Then at the end of the month, youwould take a snapshot and say,
okay, well, really missed on that.
We're going to have tomake some adjustments.

(01:07:29):
And so getting into that kind ofoperational cadence helps to drive, in my
mind, value to the company because you'reable to quickly make more decisions that
will have that financial impact and allowyou to move towards that profitability.
Versus, you know, on the flipside, we didn't really have a
lot of those great processes.
And so I'm kind of getting stucksaying, Oh, we missed this one.

(01:07:52):
Oh, we missed that target.
Oh, we missed.
And then that adds up quickly.
And then you're going backout to the market and saying,
Hey, we need some more money.
Um, shit.
I mean, at Uber, we did it multipletimes because it was such a rapid
growing company, but there comes atime with every company where it's
like, okay, We need more money, but noone's going to give it to us anymore

(01:08:14):
because we can't hit our targets.
We haven't shown that we can doanything that we keep telling investors
not to give it to us tomorrow.
So I know that that didn'treally answer your question.
No, no, I love it.
And when you said, you know, I feellike what you're building at scale
is what I would consider, like abest in class, finance function.
We call it future readyfinance function, right?

(01:08:35):
And it's really where the financefunction is enabling the business to
drive important decision makings withreal time analytics and dashboards.
And this is like, you know, creme dela creme of, you know, consulting that
everybody wants to strive towards.
So it's impressive thatyou guys are already there.
And so I'm asking the question, howdo you build a team Cause again,

(01:08:56):
these are all very data intensive.
You need a certain skillset.
These aren't your typicalaccountants on your team.
I think some of that is in your interviewprocess, we do a series of different
things in order to identify if this, if
the people that we are looking atare going to meet those demands

(01:09:16):
and be those types of people.
Some of it is, you know, through modelbuilding, like on our finance side,
how to build models, things like that.
On the accounting side, we actuallygive them data sets and have them work
through and make sure they know how toactually kind of work through some of
the massive issues of data that we couldhave, um, on the cost or other side.

(01:09:39):
But then also there's, we interviewedthem for credos, um, and as part of the
credos process, you go six levels deepand that's to ensure that they actually
have done what you're going to be askingthem to do and getting concrete examples
of that and asking question after questionto really dig into that, to ensure that.

(01:10:01):
They're not just giving you fluff,but they're actually giving you things
that, um, couldn't truly going tobring value to your team when you
actually add them to the team andrequire them to do, um, certain things.
And so we, initially we go through thatprocess, but then I would say we look for
people that have been in both very scrappytype startups that have had like kind of

(01:10:26):
improvised might've not been, no offenseto Oracle, we'll keep using them though.
But, like, Oracle probablyhas, like, you know.
An accounting team of 500 people or more.
Right.
And someone from there is nevergoing to survive at scale.
If I brought them intoscale, they just wouldn't.
It's too fast paced.
There's too much lack of structure.

(01:10:47):
And so we target ambiguousquestioning as well.
Like tell me a time when youhad to deal with an ambiguous
situation, how did you approach it?
What did you do?
What did first?
And so some of it is the lack ofquestions you have to determine.
If I just drop them in the bucket ofscale and say, run, you know, are they

(01:11:08):
going to have the ability to do thatbecause they have the framework that's
been built for them previously to do that?
Our questioning line and in ourinterview process is extremely
rigorous and gets to that andtypically has been pretty successful.
After that, it's about training them.
And obviously each company is different.

(01:11:29):
Okay.
Thank you.
So sure, you could just drop them in, buta lot of times that's going to make it
more difficult and maybe they will fail.
So it's having regular check ins, butthen in my mind, I've tried to get my team
to actually say, okay, walk them throughone, then have them do it back to you.
And then, you know, set them loose andthen come back the next day and say,

(01:11:51):
okay, let's walk through what you've done.
Let's make sure that it's okay.
And then it's having, creating thismore of an operating cadence where
people are familiar with each other.
I try to, what I bring to my teams iskind of like this open, I'm a very open
person, like way too open sometimes.
But I find that that actuallymakes people more comfortable to

(01:12:13):
come to me and talk to me about.
You know, dumb mistakes that theymight've made or things like that.
If I tell them about a mistake I made,like, like, look, I haven't ever done
that since because I learned from itand helped them realize like, it's okay.
Then they could come to me withmistakes and then we can learn about
it together and work through it.
And that helps them think throughsituations and constantly be

(01:12:36):
learning, which helps them furtheradapt it into kind of our scale
of culture and environment that isquick, fast moving as fire drills.
And you're just like a utility knife,constantly going all over the place
and having to jump in and do things andSo between interviewing and then just
the constant on the jump training andlearnings and how I, you know, earlier

(01:12:58):
described the meeting structures andsetting out priorities, all those
different things, help people to hitmy mind, be able to adapt, really
competent and do well in a culturelike, uh, scale or other fast-paced
based companies like Uber as well.
That's great.
And you've had a lot of experience kindof scaling up big teams quickly as well.
Okay.

(01:13:19):
Yes, you might have a question there,but I will just, I'm going to read
in your mind right now and I'm goingto tell you about an experience
where we had to, we were at Uberand we were hurting, we really,
really needed to build a team fast.
What we found is in Silicon Valley.
I don't want to say it's stable.

(01:13:39):
I don't want to say there's noloyalty or no stability, but in a, in
a way it is like, and we notice andnothing like I'm an old dude, right.
And nothing against the youngergeneration, but it's almost like
they have no loyalty and sometimes.
And so, you know, six months in, they'relike, you know, I don't like this.
I'm taking off.

(01:14:00):
And then you're stuck back in that hole.
And so we were in this process where we,our teams were running at the red line.
And as a result of running there,a lot of people, you know, would
just be like, this isn't for me.
I didn't, I'm not willingto do that anymore.
And so we'd lose people quickly.
So we got to a place where ourteams and then it compounds, right?

(01:14:24):
Like all of a sudden that personleaves and I'm doing their work
and then I leave and then someoneelse is doing both of our work.
And so it keeps compounding.
And so we had to do something quick.
So we're like, man,what are we going to do?
And that's when we said,why don't we go to Dallas?
We will hire a ton of people in Dallas.
There's not a ton of tech companiesthere, so we'll be able to keep the

(01:14:47):
resources incentivized because they'relike, Hey, this crazy, awesome tech
company, you got to work for it now.
Instead of some airline orsome oil rig company, you know?
So we went in, we said, okay, we'regoing to hire 20 people in Dallas.
And we're going to do it basicallyover a weekend or over a week.

(01:15:09):
And so we got all the resumes, wescreened out everyone that we wanted to.
And then we went down for three daysin total and we interviewed probably.
60 plus individuals over that time.
We had six people that flew out andthey just went like room to room and
swapped around and essentially, youknow, and we all interviewed them.

(01:15:31):
End of the day, we would go to a,I don't want to say a dart board.
It wasn't a dart board.
Yeah.
Can you imagine this one and that one?
Um, No, it was, we had a dry eraseboard and we basically put them all
on there and we'd sit there and I,I don't want to say we'd rank them,
but we would say like, who's yournumber one, who's your number two.
And we would sit there,I really like this.

(01:15:52):
And then we would work through and discusswhat we saw out of the interview process.
And then, yeah, by day two, wewent through the same process.
And then day three, we didn't have anyone.
We actually just tried to wrap up,finalize and make sure we're all aligned.
These are the people we'regoing to move forward with.
And then, yeah, we walked out of theremaking 20 offers and within a month,

(01:16:14):
month and a half time, we had likea whole entire team out there for
our accounting team up and running.
Then from there, what we did.
Cause we actually had like, youknow, four or 20 people, right?
And so we had each team member assignedto a different like training task.
Okay.
You're going to trainthem how to use Oracle.

(01:16:35):
And so then we had someone thatcreated an entire video on how to
use Oracle and the accounting streamsand all those types of things.
Then we had someone, okay, you needto train them on our accrual process.
And so then we did the same thing.
And so we created all thesedifferent, almost like desktop
procedures and then videos.
And with each of those, we had twodays when they were onboarded, where

(01:16:59):
they just watched all the videos.
And then after they got done with allthe videos, we would fly team members
out for a week to two weeks at a time.
And they would just work withthe team and make sure that they
got trained on a different topic.
And then with work, uh, you know, and,and review their work and make sure
that they were all, um, doing whatthey needed to, and they would just

(01:17:23):
sit there and rotate team members.
And so people from San Francisco, acouple of time, just fly out and then
just work to get them on boarded.
We found that they got on boarded prettyquickly, um, trained up and started
having a good impact pretty quickly.
I wish we would have been in need of thatmany roles at Scouts where I do that.
We have continued to look at like, Oh.

(01:17:45):
You know, we continue to see that theSan Francisco Bay Area talent that
can be very difficult to break intoand, and maintain someone for a long
period of time that isn't going toleave in a year and a half, two years.
And we found that Dallas is very stable.
Several of the people I know that Ihired, and this was seven years ago.

(01:18:09):
Probably at this point are stillit, but Uber in the Dallas office.
And so we've also looked at that as ascale, like, do we want to build the
Dallas office like we did at Uber that'dbe successful and just determined right
now, if we only have two or three people,they would happen, feel alienated and
by themselves versus all of a suddenhaving 20 people in an office, like you

(01:18:30):
have an office basically, you know, youcould feel that complimentary and don't.
And now what are your viewpoints on,you know, Dallas is easier to manage a
couple hours away, you know, versus anoffshore, um, shared services center.
Have you had experience managing those?
Yeah.
So when I was at Uber as well, and evenat Thumbtack and we tried it at Scale,

(01:18:51):
although still working on the hiccups,I would say, but at Uber, because
we're probably transacting at the time,10,000 new invoices, maybe a month or
something like just tons of new voices.
And.
You can either just continue to hiremillions of people, or you could look
at a solution from a system perspective.

(01:19:12):
Um, it was a mix of both for us.
And so we went, we broughtin the shared service center.
And so we work center to partnerout and we went through a
process of hiring probably 250.
Like they are AP techpeople for that side.
And then what we found is for sharedservices, they work well for like

(01:19:32):
transaction volume based stuff.
But if you want them to make anaccrual that requires judgment,
they're usually not going to bethe best or the appropriate kind of
workforce to be doing those things.
And so we in total probably had500 people that were working on the
Across the APAR and general accountingfunctions, do the transaction work.

(01:19:56):
Even with that, we would still go througha process of interviewing, making sure
that we were comfortable with them.
And then we had groups of peoplethat would still fly out to India.
It was a different group.
Probably once a month, we would havedifferent people that would be flying
out just to constantly have thattouch base and ensuring that one, they

(01:20:19):
feel like they're part of the team.
Even though they are the shared servicecenter to, they continue to receive the
appropriate training, attention, um,and input that is necessary in order
to be successful, um, with your team.
And so it, I mean, standing itup, we just needed a partner.
And so that's why we moved tocenter because they were big

(01:20:40):
enough to be able to support.
At Thumbtack, I came in and wealready had a Philippines, like
one person in the Philippines.
Uh, and we had anoperational center there.
And so I said, okay, let's.
Build out and hire morepeople in the Philippines.
And so as our accounting needs kind ofcontinue to, okay, we need more people.

(01:21:01):
Any people it's like, okay,let's look at the Philippines.
And so we went straight there.
We actually tapped the resourcethat we had that was already there
and said, do you have good people?
And then we actually opened up and.
Just like a standard recruiting process,we had hundreds of applications coming
in saying we'd love to work here.

(01:21:22):
And as a result, we were able tohire another five people out there.
And like, I just the other day theywere in town and so I met up with them.
It was great.
Still, still keep in touchwith them on the scale side.
We've tried the same.
We said, okay, well, wewant low cost resources.
And it hasn't worked the same.

(01:21:42):
And Some of that I would say is becausewe still are like much older processes.
And so typically a shared service center,I, I compare it to like RPA, right?
With RPA, you kind of have, I thoughtthe square of RPA, you have the rules.
If it falls outside the rules,it's like hitting your head against
the wall because it cannot do it.

(01:22:02):
Yeah.
And in similar with a shared servicecenter, if you have a desktop
procedure, if you have things like that.
They'll be able to follow it and youwill not have issues, but if it's
outside of that desktop procedure, orif it's something else that they just,
uh, I would have to guess, and I don'tknow how to guess, so I'm just stuck.
You get to a point whereit's no longer efficient.

(01:22:25):
And so that's where we'refinding at scale is one.
I mean, obviously we don't have desktopprocedures, we're still running by
the seat of our pants on the red lineand trying to get everything done.
And since there's not that structurethat call it a shared service center
resource would be able to kind of thrivein, they haven't thrived and it's been a

(01:22:46):
struggle finding a person that is trained.
Like we train people in the U S toreally kind of take outside the box and
just Go at it, you know, do everything.
And so we're, we're stillstruggling there now.
And honestly trying to think, do Ipull out of there and just not use
it until we can get more structure?

(01:23:07):
Or do I continue to just have them domy cash reconciliation instead of my
cash to let the wire money somewhere?
And like how much value of timeare they saving me in that it's
kind of been doing that assessment.
Still up in the air tryingto figure out what to do.
Yeah.
Yeah.
And then you talked a littlebit about RPA automation, which
I know you're a big fan of.

(01:23:29):
Talk to me a little bit about that interms of your plans for how to scale out.
Scaling with people andleveraging automation.
Yeah, well, it goes back to myPPS system, you know, people.
So it's all about getting the rightpeople, which we've discussed process.
It is about building out scalableprocesses and then systems.

(01:23:49):
And so that's where I'd say your questioncomes into play is on the S side.
And so I'll be honest, like probablyon a daily basis, I have at least five
Hey, I'm from blank company.
Uh, I was about to mention someoneI was like, but I better not, um,

(01:24:10):
cause they're annoying, but it'slike, Hey, I'm from this company.
I want to sell you my product.
Uh, we're using AI and we cansave you so much time and money.
You know, if I went and listened toevery single one of those people's
speeches to me and tell me how they'regoing to save me money and what they do.
That's a complete waste of my time.
And so the person that I typically take.

(01:24:31):
In my career saying, okay, weneed to stack rank are our issues.
And it's like, okay, what are wetrying to solve for first versus
saying, what is your solution?
And then let me see where Icould put it in to my process.
And so right now, what we'redoing is we're trying to We
did do that on the P2P site.
Um, we had like five different systemsin each territory or region, whether we

(01:24:57):
have Mexico, Philippines and Germany andall these other FTE hubs, um, that we
needed to capture the P2P side of things.
We had, like, spreadsheets and GoogleDocs that we were using for Mexico.
And then we had to call it,you know, another tool for
Germany, because the Bill.

(01:25:18):
com or whatever couldn'thandle I mean, it's not Bill.
com, because that's a different system,but it couldn't handle, you know, the
international currencies and whatnot.
And so we went and we identified asolution and we implemented the solution
that they helped to implement that system.
And actually it saved us a ton oftime and it's helped us to scale more.

(01:25:39):
We had at the time, we hadthree AP accountants that did
all T&E and all of our PV.
And we've been able to maintain thatand haven't had to add any more,
even though we've quadrupled insize over the two years that I've
been there, including our spend.
And the number of invoicesprocessing, et cetera.
And I give that example becausethat to me is an example of how

(01:26:03):
you can use a solution to scaleand support scale versus constantly
adding more people at your problem.
And that's how we go about our process.
And on a quarterly basis, we have ameeting, we get together and say, okay,
what are the largest pain points thathave come out of this last quarter?
Almost like a retro is what we call it.

(01:26:23):
We'll do a retro and say, okay,these are our pain points.
How are we going to solve for them?
And that's where we startgoing and looking at different
solutions in particular.
Right now, we're very highlyinterested in AI solutions.
We've been looking at tools.
There's a tool that willtake and leverage AI.

(01:26:45):
It'll ingest all of youraccounting transactions from
this month and last month.
And then it will prepare a flux for youand tell you what the differences are.
Like, I think of where is oneof the areas that I spend a lot
of my team time is trying tofigure out what the heck changed.
Oh, AR went up 500, 000.
What the heck drove that?

(01:27:05):
I mean, AR is a little easier, butyou know, things like your accruals,
like what did you do last month versusthis month and why did it change?
If you have hundreds of differentcustomers that you're trying to accrue
for, it's very difficult to go and,and write or develop that fluctuation
and then add in the business reason.
And so we're looking at different toolsthat would do that at the same time.

(01:27:27):
That's just an example.
That's kind of how our approachto leveraging or putting in
place systems or things such asRPA, which RPA had its purpose.
I would say it's irrelevant in mymind now with AI and things they
can do so much more than that.
Another thing that we're looking atcurrently is like contract summary

(01:27:49):
tools, uh, in a way like So wesigned so many different contracts
with so many different people.
And in accounting, you haveto capture and make sure that
all that's reflected, right?
And the last thing you want to do issit there and go through ironclad one
contract at a time and say, Oh, howis this going to impact my accounting?
So there are some amazing toolsout there that will literally

(01:28:10):
scan through every single contractwithin a matter of minutes.
And you could even add the fields that youwant it to look for and summarize for you.
And then it will return essentiallylike an Excel schedule or worksheet
that tells you, okay, you know,here's your term, here's your tenant
improvement allowance, whatever else.
They have them for revenue, theyhave it for leases, they have it

(01:28:32):
for any other types of contracts.
And I mean, that, that the power behindtools like that and the automation
that it can break is enormous, right?
Like it's endless.
There are some nuances to that.
I would say one is aroundsecurity with everything.
When you're implementing a new tool,you need to do the, you know, test.

(01:28:52):
And then as you're executing the contract.
You want to make sure you're lookingat who owns what, um, because you
don't want them to obtain all of thisconfidential information of yours and
then have the right to build all thesederivatives and use it, uh, essentially
go out and pitch to other customerson business because they have all this

(01:29:13):
information that they shouldn't be using.
And so that's one thing that's criticalat working at an AI company, knowing
that you're looking at working with AI.
Companies, you need to be veryfocused on the security around
the information that you're goingto be ingesting tools, et cetera.
And so what specific kind ofclauses or tactical recommendations

(01:29:35):
can you make for other people?
You know, you, you need toknow who you're dealing with.
Some are just almost a pass throughand they are leveraging AI from, you
know, even at open AI or Anthropicand they're using their models.
Built into their own AIplatform, their platforms.
And so what you're looking at iswhere's the information flowing.

(01:29:59):
And so you're getting different,you might know this better than I,
it's like a digital security, um,and it talks about the ownership
associated with your, the information.
And so you're looking at thosetypes of things and there should
be clauses within the contract.
It's a, that you can't make a derivativeusing our, you know, our data, or we own

(01:30:23):
all derivatives that are made, and thosearen't to be used for other customers.
It's ensuring that you're puttingin place clauses that will enable
you to continue to maintain.
And, and operate off of that data andthe new models that are being generated
as a result of it essentially consumingall of your data and putting it in

(01:30:44):
those clauses helps add that protection.
And now, uh, what I love about Scale AI,I was doing some You know, YouTube videos
of your CEO, and I love how he explainsthe concept of, you know, the data
infrastructure and the basically the threelayers of data and how to think about it.
I'd love for you to, you could probablyarticulate it much better than I can.
I don't know if I can.

(01:31:05):
Well, first you start with AI, right?
So within AI, there's threedifferent pieces of AI.
You have your compute side.
And if I said compute,what are you gonna say?
NVIDIA
NVIDIA.
Right.
So the compute side, all theprocessing power goes through NVIDIA.
Then you have the, uh, call itthe algorithm or model side.

(01:31:26):
Um, the algorithm side ofthings include the two largest
companies, OpenAI and Anthropic.
So, uh, when you think aboutthose two pieces, you know who
the, the common players are.
Then the third pieceis data infrastructure.
And that's where scale AI comes intoplay is we add in that third piece.

(01:31:50):
And between the three, you havelike a fully operating AI model.
I can't, I'll be honest.
I can't necessarily articulate allthe layers, the three layers of the
data infrastructure, but essentiallythe way that I think about it is you
have the computing and the modeling.
But you need the data to,uh, operate your models.

(01:32:14):
And if I gave you Let's take the mindof a two year old and you use the mind
of a two year old as the data that'skind of inputting into your model.
You know, how well is that going to work?
You know, if you ask a two yearold, Hey, where do you want to go?
Let's go to McDonald's andget a Happy Meal, right?
Versus if I'm an adult and I'm saying,Hey, I would like to go and eat.

(01:32:37):
Give me a place to eat and Idon't want to go to McDonald's.
So then take a, you know, fullyfully grown adult that's, you
know, 45 years of age and usetheir brain to develop a model.
There's a lot more information there.
And so that model is goingto be much more efficient.

(01:32:59):
It's going to be much more accurateand you'll be able to get a lot
more value out of that model.
And so The data infrastructure sideof things is essentially what we call
helping to fine tune the underlying dataassociated with models and algorithms
that are running off of your compute.

(01:33:20):
And so we, what we do is we havethousands of taskers that are
trained in different specialties.
One could be nursing, it couldbe engineering, and then you have
different language models as well.
And so you have, you know, whetherit's Arabic, Spanish, whatever.
And so we will, uh, essentially go out andinterview and then find all these people.

(01:33:42):
And then they are the ones that havecalled the data that then helps to
develop and fine tune these models, makethem smarter and smarter and smarter.
And so that's where scale AIkind of falls in the AI ecosystem
is a company could come to us.
Um, and ask for a model around, Imean, nursing as an example, or one

(01:34:08):
of our public partnerships is withCHEG, um, and I don't know if you know
who CHEG is, but they essentially,you know, help college students and
helps learning of students and thingslike that at different age groups.
And so what we do is we've actually beenasked to help build models that they can

(01:34:31):
then use to, you know, if your collegestudent wants to learn more about nursing,
can go into this model and sit there andinteract with it and ask us questions
that be being taught, essentially.
And so that's an example of,like, ways that a customer could
be leveraging the modeling andthe data that we're helping to.

(01:34:53):
Um, you know, bring to marketand, and provide to them, uh,
which is what they run off of inorder to make money in the way.
Yeah.
Just getting that kind of like,what is, what do you parents do?
How do you view it?
Um, how do you make money?
Yeah, it's, it's pretty crazy too.
We're definitely the largest,I won't say the best, but.

(01:35:15):
When you think of those three AIcomponents, like scale, AI is the
leader of data infrastructure.
They're like the NVIDIAof data infrastructure.
We want to be as big as NVIDIA and hopesomeday we will be, but essentially
you think of it as there's not a lotof, it would be very difficult to
build a platform where you essentiallyhave a prompt and then a response.

(01:35:40):
And you have this whole entire processthat you have taskers going through
and updating, providing responses.
Cause if you think about amodel, you want the question and
then you want an exact answer.
And so what we're doing is helpingto provide the inputs for all of
that and fine tune them to makeit so that it's smart and smarter.

(01:36:02):
I would say an example that I used,and I use their software at Winning.
I said, Hey, I have my Uber teamfrom Amsterdam coming into town,
they want to have brunch on Saturday.
And I was like, I'm not fromSan Francisco, I'm from the, you
know, the cornfields of Fremont.
Um, I don't know where to go inSan Francisco to have brunch.
So I went in and I asked thequestion, I said, name the top five

(01:36:25):
places for brunch on a Saturday.
And it gave me a response.
I went and I looked it up.
Four out of the five were closedpost COVID and never opened up again.
I was like, well, this is garbage.
Like I'm stuck with one.
So then, um, you know, I'm just Googlingafter that and saying, where can I find
now fast forward, done the same thing.

(01:36:48):
And we have much more fine tuned itto the point where I've asked the same
question that gives me like five verynice places that I can go and have brunch.
And that's because that expertise thatwe bring through the different taskers.
This helping to really drive thatvalue and, and break in called the

(01:37:08):
expert and the efficiency that willhelp any business be able to leverage
that model and that data to be ableto drive results that they're needing.
And that's, that's how we'retrying to provide, or we are.
We're definitely providing value tomany different companies and they're
involved with a lot of the largecompanies and helping to drive, bring
a lot of those efficiencies and drivea lot of the model building that

(01:37:30):
they're looking for and helping there.
So it's exciting.
It's a lot of fun.
So then what's next for you?
Are you looking, you've gotsuch an incredible experience.
Are you sitting on boards right nowor helping advise other companies?
Yeah.
I mean, you mentioned that,but yeah, about six months ago.
So I was asked to beon the other committee.
Uh, Careem was actually one of thecompanies that we acquired when,

(01:37:53):
while we were at Uber and a lot ofthe finance team reported into me.
And so subsequent to leaving Uber,we kind of split into two different
companies and one of the companiesthat is kind of more investor run was.
Looking the form without a committeeand start the process with, uh,
because they have been carved outas we had referred to earlier.

(01:38:16):
And so, yeah, they came and they asked meif I will sit on the audit committee, but
because of kind of a lot of what we haddiscussed, just that ability to kind of
think through financial statements and.
Ask questions about processes, makesure that they are appropriate processes
that have been involved heavilyin kind of control side of things.

(01:38:37):
And then one of the things thatwe're even looking at is like, Oh,
can we automate or add AI into someof the processes, it was like that.
And so it's definitelysomething that I I'm currently
interested in is just kind of.
Being involved more and like,I guess it is a individual then

(01:38:57):
kind of finance the accounting.
One of the things that.
It drives a lot of like, gives me afeeling of satisfaction as being able to
see like margin packs that you're making.
And sure, I see that at scale, but I lovethe opportunity to go and work with other,
other companies or other teams and justthinking through like ways that they could

(01:39:20):
either leverage AI or ways that they canenhance processes or things like that.
Bye.
Bye.
Matter of fact, I, I just did that lastFriday, I had a call and yeah, he was
asking me like all about our planningprocess and our weekly, you know, metric
things and trying to replicate those sortsof things and how do you go about it.
So, you know, when you think about it,you don't think, oh, I've got all this

(01:39:43):
great knowledge and information like.
But then when you're, you know, whenyou're talking and you're like, wow,
I've learned all these things that Ididn't even remember or know that I
know, but you just kind of know thatnaturally, you know, um, being able to
share that with others and just thinkthrough situations and how to approach
things like love that type of stuff.

(01:40:04):
That's, that's great.
I mean, honestly, what led me backto scale is knowing that, I mean,
when I got there, we were on a cashbased accounting, okay, that's a big
hill to climb to get to, you know,GAAP accounting and things like that.
And, um, those are the things that are alot of fun and it's trying to figure out
how to get from point A to point B andmake things more scalable and sustainable.

(01:40:25):
So for sure.
Oh, that's great.
Well, how can, how can ourlisteners get in touch with you?
Yeah.
LinkedIn
is a great place.
Yeah.
Um, no.
If you reach out through LinkedIn, Iam typically pretty good to respond
within a day or two of when I get, um,messages, but I would definitely put some

(01:40:45):
information as to why you're reaching out.
Like I said, I get probably like.
20 Sales calls a weekthrough LinkedIn as well.
Come, let's see this, anddefinitely have to block those ones,
. And from the sounds of it, you know, I, I, I try to reconcile Josh dad at four
with, you know, your work ethic and sowhat, what do you when you get time?

(01:41:09):
Oh boy.
I'll have a whole lot of downtime.
Yeah.
My wife hates it because, no, it's'cause I can't, like, I, I don't know
if it was, I was raised this way.
And I think I probably was, butI was raised like work hard.
And so as a result of that,like, I can't relax in my life.
That's what I say.

(01:41:29):
My wife hates it.
Cause I'll be sitting at home justlike, but now I'm like, holy crap.
Like, what can I do?
I can't just sit here on the couch.
Like, she's like, you need to relax.
I'm like.
Relaxing to me is doing something that'snot work then, but it's still work.
Like I would love to go.
And so I've got a cabin up in Arnoldand so in my free time right now, what

(01:41:50):
I'm doing is I actually built, you know,different shelter, um, a garage that
has like a game above it, and we knockeddown all these trees all the time, but
some in order to make space for it.
And so now I'm like, cutting up allthese trees and using, I've got a log
splitter that I spent time like, youknow, cutting up all the logs for the

(01:42:12):
firewood for the winters and, and thenI also, you know, I've got two boys and,
and so I've got season passes we have.
The epic pass.
And so we'll go up toKirkwood and go to park city.
And, um, they tried getting intosnowboarding likely, but just not feeling
that they'll stick to their scheme.
But, uh, we have a lot of fun doing that.

(01:42:33):
Um, definitely spending time with them.
That's probably what I tryto do in my pastime the most.
Yeah, that's great.
Well, thank you again for all your time.
It's been very enlighteningand I've learned a lot,
so I appreciate you.
Appreciate you having me.
I always enjoy our partnership withyou and your team and appreciate the

(01:42:55):
opportunity to be able to speak my mind.
It's been great.
Thank you.

All Episodes

Episode Transcript

Popular Podcasts

United States of Kennedy

Dateline NBC

Stuff You Should Know

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}From Uber IPO to $1B AI Fundraise: Josh's Secrets to Success

Episode Transcript

Popular Podcasts

.css-r6mb8g{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:1;overflow:hidden;}United States of Kennedy

Dateline NBC

Stuff You Should Know

All Episodes

From Uber IPO to $1B AI Fundraise: Josh's Secrets to Success

United States of Kennedy