# Adi CEO Reveals AI-First Fintech Strategy in Latin America

**Podcast:** a16z Podcast
**Published:** 2026-06-17

## Transcript

Don't let your ambition fall prey of commercial wisdom.
If you're a consensus play, there's just no alpha.
We serve over 3 million consumers, over 50,000 merchant partners.
We have incredible cost-to-serve economics, over 200 agents in production.
We built a company on a monorepo as opposed to microservices.
Monorepos today are all the rage.
Anthropics built on a monorepo.
Google, obviously, famously is a monorepo company.
When we started the company, microservices was where it was at.
So why did you make the decision?
What gave you the forethought to make it?
I hired the best CTO in the business.
How did you get your organization culturally moving as quickly as you have?
It's basically a combination of two things.
We have a very clear view of where we are and where we want to be.
We have been extremely explicit with folks about standards and expectations.
It's not going to be easy.
It hasn't been easy.
Building a technology company in Latin America often means solving problems that don't exist elsewhere.
Fragmented financial infrastructure.
limited access to credit, and underdeveloped software systems create challenges, but also opportunities to rethink how commerce and financial services work.
Over the last several years, Adi has grown from a buy-now-pay-later product into a broader platform spanning payments, commerce, logistics, and banking.
Angela Strange and Gabriel Vasquez speak with Adi founder and CEO Santiago Suarez about financial inclusion.
technology infrastructure, AI, and building one of Latin America's most ambitious fintech companies.
We're here today with Santiago Suarez, the founder and CEO of Adi.
Adi is a buy now, pay later, marketplace and payments platform in Colombia, and now just recently a bank.
And Adi has served over 25% of the population.
Welcome to the show.
Thrilled to be here.
Amazing.
All right, so for those in our audience who may be less familiar with Adi, can you give us a sense of what does the company do now and what scale are you at?
Absolutely.
So we provide consumers and merchants with financial solutions such as credit at the point of sale, instantaneous payments, logistics.
We just got our banking license.
So you should think of it as the kind of elemental fabric of commerce and financial services in one of the fastest growing Latin American economies.
Size-wise, we serve over 3 million consumers, over 50,000 merchant partners.
That's a little bit about the scale of Adi and what we do today.
Okay, so we're going to talk a lot about Adi.
I think it's interesting to talk about your career pre-Adi.
So you're from Columbia.
You came to the U.S.
to go to college.
You joined startup as a first employee, and you're pretty public about being fired from said startup.
And then left, very surgically created a list of the 10 best operators in your mind, and you wanted to go work for one of those.
And you chose Jamie Dimit.
So maybe walk us through the lessons from those experiences that relate to what you're doing now.
Yeah.
So it really starts with coming here for college.
I think I've been increasingly open about this most recent years, but for a while I was like, no one needs to know this.
But the first time I ever left Columbia was to come to college.
And I came to the U.S.
for college on a full ride because my parents couldn't afford college in the U.S.
certainly and most places in Columbia.
So that to me was probably the first time that I could.
really see doing something that no one thought was possible.
I mean, to the degree that I wrote my college essays on a borrowed computer, because we didn't have a computer, because in that particular case, those years were pretty tough, we didn't have one.
So I had to borrow someone's computer to write the college essays.
And being able to go from doing that to just getting a scholarship to this incredible place, and it remains incredible more so at the time, Yale, for an undergraduate education, it just really changes your definition of what's possible.
I think that also led me to work at Navia, the first startup.
Navia was doing probabilistic computing back in 2010.
And we used to say, this is, by the way, 16 years ago, we used to say our programs are so special because every time you run them, you get a different output.
This is back in 2010.
And we were...
pre-transformers, pre-image net, pre-all these things.
And we're out there talking to investors, being like, oh, our programs run and give you a different answer every time because they're probabilistic.
This is how AI is going to happen.
That was about the one thing we had going for ourselves there.
But I was so excited about the potential of that technology.
We were just a couple decades early.
And then to your point, we had four co-founders.
And in the 12 months I was there, we had four or five CEOs.
And I was the CEO for 35 days at the end of that stint.
And then I decided that what we needed was to fire a co-founder because he wasn't doing a good job in my 23-year-old head.
So I talked to the right people supposedly to gather support.
And then I fired him and within 24 hours, the four co-founders were like, yeah, that's cute, but you're out of here.
And I think the lessons there, one was...
I mean, there's a non-obvious one, which is obvious, which is, well, the power of AI.
The second was the power of getting organized internally to pursue a goal.
And I think one of the things that we've been able to do at Adi well, in spite of some challenges through the years, is being very good at organizing ourselves.
I was just having breakfast with one of our board members and we brought our new CFO.
And the new CFO has an 11-year career at Melly.
a board member asked him, hey, what is the thing that surprised you the most?
And he's like, how well organized the company is.
So I think that was one of the core insights was like, there's got to be a better way.
And that speaks to them, okay, how do I want to do this?
And there were two things I knew about starting a company.
One, I wanted to do it.
Two, I never wanted to do it on a visa.
Because one of the challenges in Navia was like, for a while, the government denied my H-1B, then I had to re-get it.
So I was like, I need a green card if I'm ever going to do that in the U.S.
That automatically means bigger company, lower upside, but lower downside, you can stay there.
And then the third thing, to your point, I was like, well, I want to go see how great CEOs operate.
Because clearly, I have no idea what I was doing in my 35 days as a CEO of a five-person company, but someone else did that.
So I wrote a list and we had Costco in it, I had Amazon in it, I had Wynn Casinos in it.
It wasn't even financial services really.
It was just who are these incredible people and how do they operate?
And I basically leveraging McKinsey, Yale, everyone I knew, I just try to get in a position to get a job somewhere close to these people.
And the first one I came up with this kind of legendary strategy group back at the time, which was 10 people at J.P.
Morgan because Jamie hated consultants.
So he had banned all consultants, but he had a strategy group of 10 people.
And I was like, okay, well, that feels like a good place to go learn.
And I was there for six years and it was incredible.
It was like a whole university in how to run companies and how to think about financial services.
Amazing.
All right, so I want to fast forward to starting Adi.
And I think you've done a few things that are usual, but most founders when they look at Latin America, or most investors when they look at Latin America, they expect founders to start in Brazil, which is the largest market, then Mexico.
Obviously, you're from Colombia, but decided you were going to do Colombia.
And I think the view at the time was Colombia is just too hard.
Like payments are mostly cash, 65, 70 percent.
Less than 20 percent of people even have a credit card.
And then a small number of banks own everything.
And so what did you see that gave you the confidence to enter this market?
And what do you think the outside observers get wrong?
So the answer is on why Colombia?
Hindsight is wonderful, and there are so many reasons, but part of it was I was in Colombia.
And I had landed in Colombia because I had followed my girlfriend and wife, and I refused to be employed in Colombia.
And I actually think being in Colombia was instrumental for being an entrepreneur because I knew too much about U.S.
financial services.
Every time someone would give me an idea in the U.S., I'd be like, the Bank Holding Company Act of 1960 actually makes this unworkable.
Or, you know, Durbin will make sure this doesn't work.
Or here's how you don't understand SWIFT.
And I had just been so into the matrix here that all I could see was the 10 ways in which your startup is going to fail.
Had I been that into the matrix in Columbia, I would have been like, yeah, your company is never going to work out.
I know what I know now.
But I didn't.
So I think a little bit of ignorance is always bliss.
A little bit of being the right place, right time and refusing to be an employee.
But there were two things that I saw that were very, very important.
The first one is extreme amounts of technological adoption.
And, you know, you're on our board.
So, you know, see, we have these metrics and we say, hey, there's more smartphone per capita than Brazil and Mexico.
which is incredibly powerful, right?
75% of Colombians have a smartphone.
But more than that, I remember in a three-year period from 2014 to 2016, smartphones went from being the thing that my banking friends had locally and always three years behind the models we had in the U.S.
to being something everyone had.
I was like, okay, and when you think about banking in particular, right, distribution has always been the issue with banking.
That's why branch networks are so valuable.
And that's why...
And Neobank in the U.S.
has been so challenged to have these very embedded distribution networks, particularly in consumer, I should say.
But I was like, wow, we now have a distribution device in everyone's hands at zero marginal cost of distribution.
So I was like, inside one, I was like, this is incredible.
And then the second thing to your point was, this is a country that has this very strong pay over time tradition, but no real work around it.
The only thing was Visa and MasterCard had actually changed their switches and rails.
to enable installments, a handful of countries, right?
Brazil, Korea, Turkey, us, where that is the default.
So it was like those two things.
I was just like, you know what?
There is a great opportunity here.
And the last and most important one, the UX.
At the end day, UX really matters.
And when I say UX, I mean UX not in the black turtleneck, beautifully designed thing, but UX in customer pain.
And I always give two examples.
One is I went to buy a t-shirt on installments.
and it took me 22 minutes because they had to fingerprint me.
They had to take two photos of me.
They had to call two of my friends, and then they would only do this during working hours.
Wait, pause on that for a second because that I think is very unintuitive.
Like, why are they calling your friends?
That's basically a way of, like, what's your personal reference, right?
So bankers literally just making sure they exist.
So you're trying to buy a retail purchase, and the KYC is literally phone a friend.
Two friends.
Yeah.
Not one, two.
And then fingerprinting you and taking photos of you and all these things.
And you're just sitting there like, there's got to be a...
And the other thing that I found asinine was that banks would charge you to use their online banking and not to use them to like wire mine, to access it.
So my original bank account in Colombia, I was able to enter my online bank portal twice in a month.
But after the third time, they would charge me like 20 cents.
And I was like, this makes...
First of all, why?
Secondly, now it means I have to go to the branch where I am more expensive to serve than if I'm self-serving myself or call you.
Like, any other channel of service is more expensive.
Why are you charging me to use the channel that for you is expensive?
But that just gives you a little bit of that dynamics of the market.
Now, I think that's what we saw originally.
And I thought, wow, this is a great test, but I will confess what I didn't see was how big the opportunity was.
Like, even, like, nowadays, I'm like, It's huge.
But even at the time, it was like, okay, we'll be able to build a good base.
But then we got to go abroad, which we did eventually.
But nowadays, I'm like, no, this is a massive market.
And Santi, you know, there's an example of a global company that reached success by just focusing on one company, in one country, that's Caspi.
And, you know, the story goes that you messaged the CEO four times on LinkedIn before he replied.
And then you flew to Kazakhstan to meet him, to spend time with him.
We'll be curious if you can share a little bit more of those lessons on having spoke to him and find somebody that found success only focusing on one country.
Yeah, so I give a lot of credit to one of our investors who introduced me to Caspi well before.
They are now a little bit known in fintech circles.
At the time, they were not.
And when we exited Brazil, I think everyone was very supportive, but I think even those folks who were supportive were worried.
Like, you know, what's going to happen?
These guys were so excited.
They were like, this is great.
We're very excited.
By the way, we've been telling you to do this.
The right comp is this country, a company in Kazakhstan.
I'm like, okay, great.
But they were public in London, so we had a lot of info.
So we started reading about them.
And then what Daniel, my co-founder, and I try to do every year is we try to take a trip to see things.
So we then went to the Middle East.
Obviously, the first one we did was to Brazil.
And this time around, we said, well, let's go to Central Asia.
We were Central Asia, but everyone knows about Kazakhstan, but also Uzbekistan is incredible.
In Uzbekistan, there were like three fintech unicorns, Oliver Hughes, who was the long-term Thinkoff CEO, now runs, what's the name, TBC Uzbekistan.
So we're like, let's go.
But obviously, you know, the big price is Caspi.
I started like trying to get in front of this guy in many ways.
I pinged one of their investors.
I got sent to their IR team.
But the way it worked was after like three or four cold LinkedIn emails, he was like, okay, fine.
I'm like, come.
And then we go meet him and he was so jealous.
But she sat there, he's like, what questions do you have?
I was like, okay, great.
Well, I shall ask you some questions.
And then Danny and I just spent 90 minutes with him.
And he was extremely, I was like, he just, you know, obviously public company CEO, but it's just like, just the way he thinks.
And there were two important lessons for us.
Number one, NPS.
And when I say NPS, I don't mean it.
He's actually been public about this where he's like, I think NPS for a company with less than a thousand employees is a terrible metric.
You should just listen to customer service calls.
So, you know, weekly business reviews, we always start with what happened to the customer last week.
And we have two transcripts pulled at random in addition to a number of a battery of KPIs just on customer experience.
And he was like, the first thing you need to think about is NPS.
And in fact, they measure all their product managers on NPS alone these days.
They don't have a P&L.
And then the second thing that he said, and it was very useful, particularly as we think about our own product roadmap, he was like, everyone thinks you got to do everything at once.
But he literally got on a board and he's like, that's not quite the way it works.
Here's how we thought about structuring our roadmap, potentially think about it.
And that informs a lot of product roadmap decisions because it's very counterintuitive, the work we do.
You know, we had an event last week and we announced in the same event, a bank account, next day shipping, and an integration with a third of the country's entire POS device fleet.
I was like, in the same event, to the same audience.
That's just not very intuitive.
And then the third piece which he gave us, which, you know, we always stick to heart, he was like, look, this is a very unusual play.
Equity investors will not understand this.
Just ignore them for five years to 10 years.
And then one day you'll come out and they'll be like, why didn't you call me?
And you'll say, I tried many times.
And it was just this clarity of purpose, clarity of focus, high conviction, serving a market in a different way.
And then on the marketplace in particular, Caspi has done an incredible job at owning e-commerce without logistics.
And it's not because we're tourists.
It's because you wouldn't build a social network today like you built Facebook in 2004.
You wouldn't build a marketplace like you built.
And we even said, and we were a tiny company and smaller, but when we built the ID in the first iteration, we had to do all of our ID checks in-house.
You remember this?
Nowadays, we modularize some of it.
We keep the real valuable part, the algorithms, the software.
But like for a commodity run-of-the-mill biometric check, there's great providers that'll do it.
And zero equity value generated there.
That's just a commodity.
So thinking modularly is very important.
And that was another lesson there.
Very productive trip.
Yeah, very productive.
And that leads you to the next trip.
You're here in New York.
You're receiving an award from the Fast Company, most innovative fintech companies in the world.
How do you achieve that from Colombia?
Yeah, this is a big deal.
And we were talking before the podcast that we travel very rarely these days.
But this was worth the exception.
So, yeah, we've been named...
Investors and the world finally woke up past five years to your earlier point.
Exactly.
So we've been named the third most innovative company in fintech by Fast Company globally.
Look, I think...
There's obviously a little bit of our being a little more disciplined in telling our story, but the real story is in the last 24 months, a lot of our long-term investments have really paid off and have allowed us to ship extremely exciting product, right?
So, you know, we built a monorepo-based unified data architecture company that once LLMs became workable, made it, I'm not going to say very easy, but made it feasible to like drive them at scale.
Today we have a company that has the fastest market place, has one of the most viable and at scale closed loop payment networks in the country.
We are in over a thousand cities in a country with 1100 cities with our own payment rails.
We do our own clearing, our own settlement, all in our own tech.
And then I think obviously because we live in the year 2026, we've done extremely attractive and powerful things on AI.
So we built our own in-house agents.
that currently handle 100% of all customer service queries.
They handle 100% and they resolve close to 80%.
So full resolution, no human in the loop.
So there's not a lot of dispatching.
Oh, the human, no, no.
Full resolution.
The same thing on a merchant onboarding.
We have a merchant onboarding agent that allows us to abort over 2,000 to 3,000 merchants a month.
Same stuff, 100% handle rate, over 20% improvement in conversion.
We're launching our own Transformer later this year.
So there are a number of these things that I think when you see it, we've always, you know, we've always benchmarked ourselves against the best.
And there are a number of things we've done that I think are now putting us in that direction.
I mean, there's a reason we run the company in English, which is we always run as talent, even though we're now in Colombia.
It's very counterintuitive that you run a company in Colombian English, but it is not once you think about where the world class mark is.
So I think, We're very grateful and it's pretty humbling to be in such company, but it's also just, I think, the recognition of four or five years of incredible investment spanning up.
Maybe just to stick on AI, I think one of the things I learned when we first started working together is unlike in the U.S.
where you can build financial services and then every piece of the stack, there's some other company doing it as a service.
Like, none of that existed in Columbia, right?
To your point, like, KYC might be undefensible, but it's not like there was another service that you could just bolt in there.
And so I think Addy has got a real history.
You had to build your own loan management system of just having to build every layer of that stack.
And so maybe talk about how did you approach, like, I think you were very fast on we need to rethink Addy.
to be very AI forward.
How did you decide where to start?
So I would say we actually started seven years ago when we made two foundational decisions that to this day help us.
The first one was we built a company on a monorepo as opposed to microservices.
Monorepos today are all the rage.
Anthropics built on a monorepo.
Google obviously famously is a monorepo company.
But when we started the company, microservices was that.
So why did you make that decision?
What gave you the forethought to make it?
I hired the best CTO in the business.
I'd be like, you know, now I can tell you exactly why.
But he was like, you're all idiots.
We're going to build a monorepo.
And the monorepo gives you, obviously, you know, for folks who don't know, what the monorepo is, is basically all your code is in a single repository.
So for agents to read it.
It's a lot easier than if you have multiple code bases in many parts of the org.
They also have some extreme economic advantages because there's very little duplication.
You reuse components almost automatically.
We have 60 product engineers supporting a gigantic product surface.
So Monorepo was kind of a canonical decision, one that allowed us to drive AI.
And the second one is...
You know, we built a company using an event sourcing architecture, which means that every single event that happens at this company gets logged.
And events are anything from we texted you to you added something to our shopping cart, to your bureau score, to your financial, everything.
You know, we generate over 10, what is it?
Over 10 million events per day.
So those two things were there.
critically, these event architectures we use Kafka for hours are brutal to query.
So that's why most people end up moving away from this.
And again, four years ago, no, five years ago now, we partnered with Databricks.
So what that means is you, in real time, are able to ingest these events and showcase them in vector form for LLMs.
SQL form for classical machine learning models, obviously tabular forms for still the humans we've got because we do have some humans.
So the whole engine and foundation was there.
And you're able to build on a very solid foundation.
And then, of course, once we went in, and I'm lucky enough to work with a great team that is very anti-hype, the question was, okay, where do we start?
And then we decided to start counterintuitively with legal.
And legal has two reasons.
One, and now there's a wonderful Brazilian company that has figured this out, which is you get sued for everything in these countries, right?
So there are these things called in Colombia tutelas, which are emergency constitutional actions, and you get 48 hours to respond.
And if you don't respond, the legal representative of the company, usually the CEO, goes to jail.
Like, it is liable to go to jail.
Like, it is that high.
And then like that, you have all these consumer protection lawsuits, all these things.
So with insane response times.
And by choice, we never sell them because that's a precedent.
So, you know, there's a cottage industry in Colombia of people who will be like, oh, we'll sue the bank to get them to remove your credit report.
And it costs the bank two cents to remove the credit report and it may cost you a few thousand dollars to fight the lawsuit, maybe ten thousand dollars.
We've taken a couple all the way up to like.
the little below the Supreme Court, because we will just not settle.
Then people know.
Yep, matter of principle.
So we started with legal.
Because with legal, you actually need to take this incredible foundation that I was talking about, but you now need to make sure your LLMs work extremely well.
So you need to build the right RAC pipelines, you need to build the right kind of feedbacks and effectively RLHF in-house loops, because 48-hour clock starts.
You need all the data from every customer service interaction this customer had, all the data from your financial core, all the data from the transaction history log.
They need to package this in a way that's sensible, legally allowed, with the right precedence.
And they need to get ready for a fairly junior lawyer to say, yep, this looks good on print.
So, and I remember when we were discussing this, I was like, but let's start with customer service like everyone else.
Carlos, my CEO, and then Mauro, they were like, no, this is the harder problem, but this is the one that scales.
Because all the pipelines we need to build for the, A, by the way, we are saving a few lawyers, if not more on this, because we're able to get incredible efficiencies.
But B, think about what it takes to resolve a lawsuit in 48 hours that otherwise the CEO is liable to go to jail, where you have to pull all the right data, you have to pull all the right information, all the right pricing.
with great accuracy.
So it was a six-month investment to build these pipelines.
We built these pipelines.
And then, you know, within 90 days, we had V1 of our customer service agent, which, as I mentioned, you know, off the gate.
It was crazy.
Off the gate, 100% resolution.
100% not resolution, 100% handling.
And off the gate, it was 60% resolution.
Because if you could resolve a lawsuit, you could resolve most customer service interactions.
Cynthia, I wanted to touch a little bit on a point that you mentioned about talent and building a great company because, you know, we invested in a couple of companies in Latin America.
And one differentiation that we've seen is people that attract global talent tend to have a larger success case.
And I wanted, you know, from your experience having worked in the U.S.
and also building Latin America, you know, curious how you think about attracting the best talent and how you build an organization that can attract this talent from this year.
Yeah, it's been a selfish focus of mine because I think one of the reasons I didn't last as long as I did in some of my jobs is I just thought my teammates were not great.
So I think when I decided to start my company, I was like...
I have control of picking my teammates.
I can only point out...
Yeah, exactly.
I get to choose who I work with.
And as I, you know, we hosted them last week.
We meet like six times here in person.
And in real time, I said, guys, I just have a great time working with all of you, which is true.
It took me a while to get to that place.
But how we thought about talent was multiple ways.
One, I think I mentioned earlier, we run the company in English.
And we run the company in English.
The obvious reason is it allows us to attract non-Spanish speaking talent.
The non-obvious reason, which we figured out later, was Spanish speaking talent wants to work.
for great companies.
And being English, actually, there is a certain raising of the bar.
A premium.
It's a premium.
And people want that.
And I remember when we went back to Colombia, my co-founder and I were discussing, he's like, oh, should we now migrate to Spanish?
Because there's a huge tax about running the company in English.
An easy one.
All our copies in English and they need to get translated because our customers want to see it in Spanish.
And we were discussing this.
And then I was like, no, we had not quite an exec offsite, but more like a manager.
Once a year, we do this thing and we brought the full 200 of them.
And most of them are local.
But the fact that they get to, like, it changes their mind.
It's like, we are here to do serious work because we're speaking in English.
You know what I mean?
So I was like, that was a piece.
So we run the company in English, which actually counterintuitively attracts local talent.
The other thing we've done is we have a very high bar and we push the bar very high and we're very editorial.
And we're very editorial on what we want to see from our colleagues and teammates.
And my view has always been there may not be that many of them, but we only need, you know, a few hundred of them.
And in whatever, three, four hundred million pool of people, we should be able to get this.
And that speaks to the other one, which is we hire regionally.
I would also say, We're also a remote-first company.
As we think about our AI adoption, we actually think being a remote company is a huge competitive advantage because it ensures all the context is explicit and it allows agents to work on that context.
Like, if you want to take an extreme example, the three of us were building a company right now, we wouldn't be sending memos to each other.
We would probably just be hashing things on a board and then we want an agent to drop on, like...
It's just hard for the agent to have context.
We have a lot of like formal APIs at the company that allow agents to craft them and run with them.
And then the other piece is we give people great development opportunities.
So we have a model where we bring world-class leaders and we become very good at hiring them, even though each search takes like nine months, true story.
And then we train young up-and-coming talent.
And that combo is extremely powerful.
And then related to that, and we were talking about this before the podcast, it is impossible not to be AI pilled if you're in the Valley.
Like just stand in a coffee shop.
It's the only conversations you're going to hear.
You walk down the street, you bump into five people from labs.
That is not the environment in Bogota.
No.
How, so you obviously had a CTO that was leaning into it.
You were leaning into it.
But how did you get your organization culturally moving as quickly as you have?
I think the same way we've done a few of these transformations, right?
Like four years ago, we're like, we're going to make money come hell or high water the summer of 2022.
I was just like, one day, I remember.
And I remember why.
It's like, well, this world has changed.
And I think it's probably the similar thing.
It just becomes the top priority for the company.
Obviously, for a company of our scale and our product surface, as I am reminded often by my team, there are many priorities.
I was like, that's fair.
But this is the one, I think, it's basically a combination of two things.
One, giving you the tools.
And then two, it is not optional.
And we have a very clear view of where we are and where we want to be.
We have been extremely explicit with folks about standards and expectations.
It's not going to be easy.
It hasn't been easy.
It has been easier than I thought.
But we're asking.
our workforce the same way everyone is asking their workforce to undertake like three high jumps in sequence in a span of months.
But, you know, I send, like most people do these days, I send a note to the company every Monday and it suddenly became month-matic.
And the other thing I did was before I got conviction to push this, I built my own stack.
So I got my own cloud provisioned.
I started playing with, and I went the whole way.
So I was doing my own DevOps.
I was provisioning my instances.
I was getting my own API keys.
I can just really playing with the core tooling.
It was so fun because like my CTO and I would text at 10 p.m.
I was like, hey, this thing on Amazon is breaking.
He's like, what are you doing with this?
Like, I was just literally asking for DevOps style support that had nothing to do with the actual AI, but I was like, I want to build the whole end-to-end stack.
I don't want to just play, because there's a lot of now agent wrappers, right?
So they'll be like, hook all these MCPs, and you're off to races.
I was like, no, I want to build my own.
So I started from an empty EC2 instance.
And from there, we built everything.
And once I figure out, okay, if this is helpful for me at this scale, and if I'm able to do this kind of in between meetings, basically, then the entire company should be able to do this.
And then working with a partner and saying, God, now we've got a very powerful kind of four pillars, right?
So we talk about A, our customer service, Audrey, we call our agents, and we've talked about them.
You know, we're investing in transformers.
So we're building our own transformers using our own GPUs to train our own kind of like, GPT of sorts, we call it ADDNA, with incredible early results.
Obviously, we've got kind of like our whole refactoring of the engineering code base.
So we were moving away mostly from reactive Java to more declarative language, sort of kind of code as hardness.
And we've seen extremely powerful early results.
We just shipped a version of our web marketplace.
So, you know, our marketplace is in our app.
because we're very app-centric.
But having it on web, great customer acquisition tool.
It also allows TikTok to index you, Google to index you.
It's just a lot of benefits.
But we've always been like, man, it takes six to nine months, five engineers, 60 product engineers.
Not quite sure it hits the bar.
Two engineers, two months.
So then that's the piece.
And then the last piece, which is the one that I'm spending most of the time, even though it has the least economic impact in the short term, is changing the operating model of the company.
to be kind of agent and AI first.
And that's where we are making great inroads, but where we're further behind against the other ones.
One thing that's very interesting about Addy is that, you know, it started pre-AI where obviously you described opportunity in financial services in Latin America.
Curious if you can share some thoughts if you were to go back to Santi when you were starting the company back then.
What would be some of those lessons that...
things to watch out, especially for entrepreneurs that listen to the podcast and starting companies, not only in Latin America, but globally.
But yeah, it would be great to hear from you.
The first one is, remember you're a technology company.
If there's one thing we did extremely well, but only became obvious five years down the line, was our investments in technology.
Like this stuff that we're talking about, about having, we have over 200 agents in production.
We have...
incredible cost to serve economics, all these things are only possible because we're a technology company.
Truly.
And, you know, we would have some people in the management team because for about a year or three years ago, we would spend every week talking about how long it would take to compile when an engineer pushed a code into production.
And it had gone from sub 10 minutes to 33 minutes.
And we treated this as like as a major problem.
And so we were like, why are we even wasting time on this?
But it's like...
Guess what?
Like, you know, 100 engineers times 20 minutes twice a day, that's a lot of minutes.
But it just gives you a sense that you cannot just one day show up and be like, oh, I'm a technology company now.
So I think it's something we did extremely well, in part because we made some early good founding decisions, but in part because we actually knew that.
And I think particularly in Latin America, it's very easy to forget that.
And people treat technology kind of as an afterthought or as an enabler.
We're like, no, no, this is a real technology company.
And that's actually where a lot of your equity value eventually compounds.
The second lesson I would say is go a bit slower at the beginning.
That'd be great.
If I think about some of our early near-death experiences, they could have all been avoided by going a little bit slower.
We had some awful fraud experiences at the beginning, and it was because we were just going too fast.
But again, remember, right?
This was the go-go years.
So in the go-go years, you have to go very fast.
And then the last piece is, be very big in your ambition, but be very contrarian how you get there.
Even this morning, I was in a meeting and I was like, oh my God, this just feels very conventionally, a very conventional path to getting to where you want to get.
So it sounds obvious, but particularly LATAM.
The conventional wisdom in that case was go to Brazil and Mexico or thin layer across many markets.
And I'm not saying our playbook is the right one for everyone.
You know, there might be cases where you should go thin layer many markets.
But don't let your ambition fall prey of conventional wisdom.
Because then there's no alpha, right?
If you're a consensus play, there's just no alpha.
Those would be the three.
And then maybe bonus, have a great co-founder.
I think it'd be interesting to talk a little bit more about how you're running the company now.
And you do a few unique things, but one of the ones that you've moved to recently is North Star metric versus long list of quarterly OKRs.
Right.
So the North Star metric was the foundational organizing principle of the company.
a forevailing time.
It remains so, but we're playing with it.
But it started when, we always hated OKRs because I found OKRs to be very heavy.
Like, just so heavy.
It was like, okay, five OKRs, five KRs per OKR.
So now we're 25 things and we've got to look at just in the company.
And then if you're very impurist, then you cascade them.
So suddenly you have like 150 things.
I recently read this incredible post on Twitter by this Sequoia investor, Sean Maguire, about why X was going to win or not.
And, you know, I don't want to talk about that in particular, but it was an incredible treatise on how Elon Musk runs his companies.
And we end up realizing he spends all his time on the most important thing.
And very little time on everything else.
So I think the North Star became that because we ended up realizing, again, in the summer of 2022, equity capital markets are done.
at least for a while, we need to make money.
And it's just becomes a very useful organizing principle.
We're making this amount of dollars.
We're losing a lot more.
What is the first thing we can move to not lose money?
So we started with risk-adjusted margin.
Then we went to gross margin.
Then we went to gross margin minus sales and marketing.
And finally, we're like, okay, EBITDA.
And now these days, we even optimize our taxes because we're trying to optimize that income.
So it just focuses the discussion.
By the way, it worked really well, but now that you have three very different business lines, we're starting to play with it a little bit more and saying, what are the, but so we're calling the L1 metrics, one or two, maybe three KPIs that we want to track.
But even then it's three, not five, seven, because once you get to five and seven, no one knows them.
So that becomes hard.
Yeah.
If everybody at the company can't name them, then you've got too many.
Which was a great thing about profitability.
Yeah.
And by the way, for what it's worth, I actually think one of the biggest challenges we had after we hit profitability, it's kind of like the dog that catch the bus.
And they're like, profitable.
Now what?
But no, it turns out we knew it, but it was such a core thing of every weekly business review for three years.
Then once we actually got there, we even had to do a lot of like company education because we stopped talking about it on a weekly basis.
But we said the same way that no one says you should show up to work wearing appropriate clothing, that's like, that is profitability.
It's just like an expectation.
Like we've gotten, because people are like, well, if we don't talk about it, can we be unprofitable again?
And we're like, no, you cannot be.
It's just a necessary part of what we do.
And I think the other...
The other principle that you instituted very early on and have been very systematic about enforcing it is that everything is written down.
Like even obvious shit that feels like maybe it shouldn't be written down, you managed to get written down.
And so let me talk through, like that turns out to be very prescient in the world of agents.
All of your SOPs are by definition written down.
But what drove that before there was that kind of necessity?
So the original insight was certainly not agents are coming.
The real insight was a...
I had read enough about the early Amazon culture with the memos, but B, I had been a consultant and I had been in enough places where PowerPoint was used to just model thinking that I was like, okay, we don't want to do that.
And then C, which is, I think, where what you probably have seen is as we build a company, you know, as you build a company, you have some stated values that you're like, nah, we're actually not.
That's actually not a value.
Like, we don't value that.
But there are some...
even originally unstated values that become extremely powerful.
And one of them was, can you articulate the why?
And can you articulate the why, even if it's the most obvious why?
And that's when, for us, it became extremely important for everyone to write things down.
And even, you know, we were always remote first for engineering and product, even pre-COVID.
And post-COVID, when everyone left, then it became very easy.
But even when we were in person, we would just print memos.
and just have people read the thing and mark it up and ask questions.
Because I think it's just, all these companies are very idiosyncratic and I'm very idiosyncratic.
And one of the things that drives me nuts is when people don't know why they're doing things.
So then we basically said, well, you always got to be able to say why you're doing things.
By the way, myself included, right?
Like I will tell you once a week, someone on the management team will be like, well, can you just articulate why this is a good idea?
And even when we became a bank, some of the folks on MASH were like, I understand you think it's a great idea.
I don't understand it.
So would you mind just writing a note explaining why you think becoming a bank is a good idea?
So it just, so it works.
And by the way, in that process, we A, we committed to becoming a bank, but B, found a few things that we otherwise would not have found out.
So Santi, let's look ahead.
Everything goes well the way that it's going.
Five years from now, where is Addy?
What's the great vision?
We've been doing this now for seven and a half years.
So had I predicted we would be here in seven and a half years, I don't think I would have gotten most things wrong.
But I think I would have gotten a few things right, which I think will remain the same.
So the first one is we are a force for financial inclusion and economic development, not only in Colombia, but in many countries.
Like we figured out a way of packaging.
kind of retail and commerce technology with financial services to just drive economic activity.
And that, you see that that has massive impact.
We have cases upon cases of people who are able to send their kids to school.
You know, we just featured, we featured this 80-year-old man two weeks ago in our merchant event who was selling a thousand bucks a month Priyati and now they sell.
$30,000 a month and we're 75% of those sales.
And it's insane.
It's like, you know, 300K plus a year, like in the Amazon, by the way, down in the Amazon forest.
Like it's just like out in the Amazon.
So in five years time, we need to be able to do that at even higher scale, two, three orders for Magnet, more and in more countries.
That's the first thing that, that it's a very unique combination.
We're not just a fintech or a retail commerce.
It's just that combo.
The other piece is we're not only AI.
First, we're probably mostly AI, AI native.
I was talking to my head of human resources and I said, we should think about calling you the head of human and agentic resources.
Like, is there a different way where you're now in charge of the productivity of the people who produce?
And the people who produce are people and agents.
So one of the things that I get most excited about is I get to play at, even within Addy in this tiny Colombian company, we get to play on the forefront of some of these things.
And can we, so in five years' time, we should have agents running all around the place.
And that'd be, and then the third piece is, in five years' time, we also need to be the place where people have a great time and do the best work of their lives because they're driving economic empowerment, inclusion, growth.
They're playing with great technology and great technology enabling them.
And they keep going, right?
You know, we were talking before the podcast that I'm a rare traveler.
That's because I have a young family.
And these jobs and these companies are so demanding that you just need to make sure you're doing them for the right reason, but you're having fun.
So in five years time, like, are we, like, I go to work, I'm so giddy.
And every day there's, as Angela knows, like, there's always something.
But I am so giddy.
So in five years' time, we need to be giddy.
Like, there are a few hundred people, because it'll be a few hundred people, by the way.
I don't see how it'll be a thousand people.
You know, Angela doesn't know this because we haven't yet had our summer board meeting, but we're running 150 heads below our budget while exceeding the growth.
Like, 150 heads.
Wow.
Your cost to serve is already...
Our cost to serve is nuts.
And we're running this because we're seeing...
And it doesn't mean, by the way, that we are cutting heads.
It just means that for the growth rate we had, right, we were growing this year again at that rate.
We, even with AI, we're like, oh, we got to make a few hires here or there to support the thing.
And we're running behind.
So we cut that thing and we're going to keep cutting it again.
So the velocity and efficiencies we're seeing, they're real, but five years down the line.
Like imagine what we can do.
And, you know, at that point we may be talking about, you know, taking this interplanetarily, right?
Like I always ask, and this is something I tell Daniel, I'm like, part of the reason you build this company is because you always want to be pushing the envelope a little bit.
I think the day I can't push the envelope, maybe the day that I'm like, I may look for a different job, but you get to, and by the way, you don't have to be in San Francisco to push the envelope.
Like we're in Bogotá, we're pushing the envelope in interesting ways.
different ways, obviously, than the foundation labs.
But yeah, in five years, we should still be pushing the envelope.
And if the last five years are an indication, I mean, we'll be doing crazy stuff with jetpacks or whatever it is that will be there.
But maintaining that sense of joy and wonder, I think it's going to be critical.
And I'd love to see that again in five years.
Awesome.
Santi, thank you so much for joining us.
Thank you, guys.
Thank you, Santi.
Thanks again for listening, and I'll see you in the next episode.
As a reminder, the content here is for informational purposes only.
Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast.
For more details, including a link to our investments, please see a16z.com forward slash disclosures.