Why This YC Startup Just Raised $4M to Catch Fraud Before It Happen

what if AI could do the most tedious parts of fraud detection

chasing false positives

analyzing transactions so teams could focus on what really matters

that's what today's guest Satya Tumati is working on

his startup Socratic AI

creates specialized AI agents that automate the grunt work

Satya has raised over $4 million

in order to reinvent the way the world fight financial crimes

you literally used to connect your laptop to a car

and make real time changes while it was driving

so here I am like you know

writing some code and like

seeing a 2 ton robot respond to it in real time

operations on the other hand hasn't seen much slower

that's where like you know

the shared vision of building AI co workers for fraud and this teams

you don't need confidence to go for it because like

you know what's the worst that could happen

the earliest demo was basically a proof of concept or

or like you know

very small MVP that we hacked together in less than a week

you just need enough for people to believe you can actually delivers

what you're promising

in terms of prioritization

like we had to do that through our customers

like you know

what do our customers value at this point of time

early weeks like we were doing everything the hard way

like you know

cold outreach on LinkedIn emails

like you know

tapping into every connection

once you start building your own company

it's it's a bit hard to imagine working on someone else's dream

hi my name is Satya

I'm the co founder and CEO of Socratic AI

and this is founders in motion

quick thing before we get started

we have a lofty goal this year of hitting 1,000 subscribers

in order to help more people build really great companies

so if you enjoy the content

learn something new the best way to support us is by subscribing

okay let's get into it

so you grew up in a small town in India

um and the story goes that you were inspired to come to the US

by a TV show

yeah Silicon Valley

watching the show like

you know that idea that

a few people with conviction

could turn the impossible into inevitable

it just stuck with me

I also wanted to be in a place where that kind of ambition is like

a baseline and not seen as something crazy

and the perfect place is the Silicon Valley

you did end up at UCLA for your masters

and then you work for

sort of some of the biggest tech names in the world

so cruise on autonomous vehicles

LinkedIn obviously

I read that you literally used to connect your laptop to a car

and make real time changes while it was driving

this was when I was at like

you know hell may I like

you know it was a very small startup back then

there's this thing called like you know

Half Moon Bay in in the uh

in the Bay Area which is like an incredibly like

you know beautiful spot like

you know by

by the ocean

but the road to it from the highway is a little curvy and all

and we were gonna we we took that route

so here I am like

you know writing some code and like

seeing a two ton robot respond to it in real time

I mean it

it was pretty surreal it's one of those moments where you realize like

okay this is not a simulation like

you know the physical world is your integration test

and yeah

like it was equal parts terrifying and a little bit addictive too

yeah yeah

I can't imagine like

you're really at like the cutting edge of technology

okay let's connect all the dots

um after this really cool and incredible journey that you've had

um you landed on building AI co workers for fraud and risk team

yeah

at a high level in very simple terms

how would you describe what you're working on at Socratic's AI

over the years right

like if you look back

the engineering world has benefited a lot from the co pilots

auto suggest or seamless tooling

like you know

since everything starts with code

like you know

people are gonna build something that

that's gonna help the engineering world first

uh when I

when I'm like you know

reading or like you know

doing some research

operations on the other hand hasn't seen much love

like especially with today's AI like

you know I felt that there's a real opportunity over there

and since early this year

like

when I was exploring problems in fintech and other ops heavy domains

where AI could create the real lift

which has not been possible

my co founder

she lived the problem that that we are tackling on that's where like

you know

the shared vision of building AI co workers for fraud and risk teams

came like you know

while solving today's problem

you also wanna build for something that's gonna last uh

in the future with AI coming in

we we

we want to reduce the routine and amplify the judgement

I wanna double down a bit on the kind of validating the problem space

mm hmm um

so when you came about it

so obviously your co founder had this issue in her previous role

but when you came into doing those like customer interview

tips and tricks or learnings from that experience

that really help you validate the problem

very quickly

going into customer calls

we should try to listen as much as possible

most of the times like

you know especially I've seen like

you know some of my colleagues

other other founders or people who wanna be wanna be founders

we think this is a problem

we go and try to build it

but when we talk with customers

like you know

the entire perspectives change like

you know what we think is difficult

but is needed may not be what the customer want like

you know they might want a very simple thing which is easier to build

you need to listen very carefully

not bringing in assumptions and asking them to validate

but identifying what their pain points or what their problems are

from the engineering world like

you know people dread like

you know customer calls or like

you know interfacing with people outside the companies

but treating it as a interpersonal conversation rather than okay

like it's a transactional

was this always the product you landed on

or were there kind of earlier iterations that you guys canned

from these conversations with customers

it was more or less the same like I I wouldn't say like

you know we we had to pivot uh a lot but

in terms of prioritization

like we had to do that through our customers like

you know what do our customers value at this point of time

and like you know

what is something they would want in a couple of months

so we need we had to

be very flexible and dynamic in terms of how we organize our roadmap

because we're we're building something pretty powerful and generic

like AI coworkers for fraud and the stream

so whatever we are planning

is something that the customers would want at some point of time

but the hard part is what do they want now and what

what gets this product into their door like

you know what

and what gets an immediate value for

for our customers so yeah

like you know

this is a process that we keep refining and like

you know try to come up

you met your co founder RIA through YC Co founder matching

yeah um

I hear it's it's

it's it's kind of a mix back

it's a little bit like dating app for founders

so what was that process like

yeah it's like a dating app without the selfies

but but yeah

I'm just kidding

yeah like it's

it's a it's an amazing platform like

you know quite underrated when I share this with other people like

oh really

does that work

there are several times that it won't work

like I met some really interesting people

some amazing folks

but there was kind of like a mismatch on what we wanna build or like

in terms of how we wanna build it

but when when I met with Riya

like you know

the alignment was pretty obvious on why we want to start a company

and what winning means our skills are pretty complementary

like you know

she has this huge insane product and domain expertise

and I bring the engineering skills

we feel like you know

this actually balances the company and it

it it was a structured search for someone uh

with the same reasons to build and different strengths to ship for me

and yeah like it's been going well

really well so far

when you think about kind of success within YC

Co Founder matching or any other co founder matching program

what do you recommend or kind of some key questions that people should

um ask each other to understand where the alignment is

if we did this 50 co founder matching questionnaire

like you know

that's available online it's

it's pretty brutal like

you know you it

it takes at least a couple of hours to write it down

but if someone is serious enough like to actually partner with you

they they would do it

the there are questions around what do you wanna do

when would you wanna exit

what are your motivations

so you kind of truly understand

why a person is trying to partner with you

and like what you can expect from this

and I think like you know

that's the first step and once you know that okay

your values are aligned

and you can bring in this complementary skill set

having a work trial definitely helps like

you know at least a few weeks seeing how

how you respond how the other person responds to feedback or like

you know alternate suggestions because like you

you also want to be able to like

you know speak freely and make sure there's

there's still a healthy relationship keeping it in the YC family

what's even crazier is that you guys applied to

Ycombinator with just an idea

no product no website not even a name

you guys kind of just met each other

so yeah what gave you the confidence to go for it

and what do you think made you two stand out

you don't need confidence to go for it because like

you know what's the worst that could happen right like

you know they would they would reject you

so so so like I have this saying like

you know if there's when you have nothing to lose like

you know why not give this a try

so that's one reason and I

I mean my co founder is the same way

so yeah

when we started the batch

we noticed a lot of teams are in a similar position okay

like you know

they just had some idea or like they have built similar prototypes

I wouldn't say a lot like

you know there are significant other teams like

you know in a similar stage like us

so yeah I I wouldn't say we we were special

but but yeah like

you know

so what I gathered is like

or what I my intuition says is like

you know YCC looks at three simple things is the problem huge

is this gonna be a big big uh

company if if they are successful is a big market

and the second thing is can this team deliver the product

can it does this team have the technical chops

or like track record or intuition to build this

and can they sell it

do they know how to talk to customers

so in the end like

you know that that's what you want right

you want someone to build and someone to sell

this is a big enough problem that needed someone to work on it

and we we were even prepared to work on this like

you know regardless of YC so yeah

yeah I love that

um and I think that's true

right like

I think the sentiment around it is

it's so much more about the team

especially at the pre seed level

like do we think that this is like a 1%

I do want to talk a little bit about the product now

so first version of the product

an MVP or a minimal viable product can mean a lot of different things

so yeah

what so there is a rhetoric of like

sell before you build and sell while you're building

yeah in your opinion

when is an MVP actually enough to start selling

every founder faces this like

you know when

when do you uh

when when you go to the market with your product right

so our earliest demo was basically a proof of concept or m or like

you know very small MVP that we hacked together in less than a week

a scrappy prototype that showed like

you know how AI could automate a lot of manual work

analysts were doing it wasn't fancy

but it it worked

and alongside that we also designed

like

a small prototype version of what the full platform would look like

and which like you know

we did it by putting a lot of thought into the workflow and like

you know how the experience would be

because my co founder has been in this space

so like you know

it was easier like

you know to build this vision product

in b to B in in

in my opinion an MVP isn't about perfection

it's it's about credibility and conviction

you just need enough for people to believe you can actually deliver

what you're promising

something that is live and something that that can say okay yeah

like you know

if they could expand this or if they could keep working on it for

for a month or couple of months

I could see definitely like

you know being a really good quality product

because in the end like

you know you have to build something people truly want

and the sooner you test that

the faster you can iterate and

cut through a lot of assumptions that people put that in their heads

as uh

Peter Thiel says like brilliant thinking is rare

but courage is even in short supply

like yeah

like you have to be courageous enough to

show your customers the not so fancy product and uh

and be able to like you know

polish it for what you're building

it's operation for quite highly sensitive data materials

so when you think about bridging the gap from like

showing the prototype

they're engaged to actually having like a signed price agreement

how do you think about like maintaining the connection

is it like through a design pilot

and how would you structure that to ensure that kind of

they're also incentive aligned

depends on the size of the customers that we are selling to yeah

for example uh

if it's another startup like

you know it

it's a different ball game

like you just go knock up

uh go knock knock

knock their door and like hey

you know I built this like

you know are you gonna use it

but if it's a public company

it's it's

it's quite different or it's a bank

it's different so in our case

we know that we're gonna be working with larger institutions like beat

fintechs beat banks or credit unions or or

or or

or like you know other market places

so what we thought is we need to build product with security and uh

the future long term scalability in mind

and that's how we approached and like

you know beat like

you know getting these compliance

uh done like

you know from

from the very first couple of weeks since we started the company

so we were like you know

thinking in those directions and uh

in terms of whether we wanna do pilots

paid pilots or design partnerships

I think it depends like

you know whatever gets your product in

uh be used by the

by your customers like whatever it takes like

you know do they want a design partnership

do it do they want a pilot

and do they want do it like

you know do they want to test it on production data

do it do they want to backtest on historical data

do it so whatever gets your product to

whatever gets your customer use your product

like you know you

you gotta do it because if

if you are confident enough on the problem

and confident enough on the product that you build

they're definitely going to continue buying this

so your your optimization should be for the long term

and not for the short term

so in the in the

in the end like

you know if there's a real problem and if the product is good enough

people would continue using it

so yeah

the optimize yeah

like you need to look in long term

so that's what I would say

yeah yeah

that's a really great piece of advice

so if we shift gears a little bit to the selling side

so walk me through the moment when you got your first yes

how did you land the first customer

yeah like

uh in early weeks

like we were doing everything the hard way

like you know

cold outreach on LinkedIn emails

like you know

tapping into every connection and like

you know my co founder has in in the fraud and risk space

and I I even white coded a small outreach product that writes

customized email to larger VC companies like

you know saying hey

you know we are in this VC batch

and you know

we noticed you launch this product like

you know for fintech companies and

you know you could have fraud and risk

and like we could yeah

like you know

we did everything to our surprise

our first customer is actually

has actually come from a completely cold message like

you know we noticed they're trying to scale their fraud operations

and we restarted saying

what if we could help you with our AI agent

and we are lucky or we are fortunate that they saw a demo like

you know they loved it like

you know the

what we are building and our intuition

like really aligned or resonated with what they need at that point

and in less than a week

we showed the demo to a broader team and we got buying and

and our champion such such an amazing person like

you know she moved the wheels for us really

really quickly and we were like in production pretty soon after that

I'm actually witnessing that

you know you got this contract and you're signing it

like it's such a great moment of validation

oh man like this is super real OK

hundred percent we are doing it

like you know people are actually

I think that yes you know

this is what we need and like

this is what we want so yeah

like that validation is pretty helpful

how did you figure out like what to charge your first customer

yeah like pricing is always a tricky one it

it it still is and we are still refining it

so I'll try to stay a little diplomatic without revealing much

but what

what I can say is like AI has changed the game completely

right like you see everything from token based to seed based and

and and they all make sense depending on the use case

but we Learned that in our space

it has to be more of like a hybrid approach

like whether it is like a part time

part value based part outcome based

or part part seed based with some thresholds

and we experimented quite a bit

bit in the early

and found out that the best approach is to stay flexible and to align

uh on pricing with how customers actually realize value

how they budget and whatever makes it easier for them

so yeah like for startups to be rigid on pricing

uh

not not in terms of the amount

but to be rigid in terms of how you structure your pricing uh

is uh

is not a good idea in my opinion

so you have to be flexible and

and and like

you know

work with customers on how how they actually are used to because it

it makes a better story for them

like to actually sell it to their finance teams

or when they're budgeting

yeah and you

you want that to be as seamless as possible

yeah yeah

yeah I mean

to be quite fair like at a startup

you really don't know what to price

you kind of have to play around with it and see what

cause pricing is all about like what they're willing to pay right um

yeah and it

it's kind of bridging that gap

um but so it

it's actually super awesome

cause if you think about it

you're living your Silicon Valley uh

dream from a few years back

so yeah looking back on this whole journey

what has been the most valuable lesson that you've Learned

as a founder something you wish maybe someone has told you earlier

I'm still early in my founder journey

like in a few months

and it's been incredible so far like

you know it's fortunate that I have an amazing co founder

incredible customers and the support of VC and pair or lead investor

I mean uh

I I wouldn't say it's one specific lesson

but it's something more I've grown to appreciate over time

in the past few months

you know

like there's a lot of advice out there about startups and it's true

but only in certain context

so people miss miss that asterisk

so you know

you hear move fast and break things the Facebook way or uh

build something unbreakable

which stripe and apple are known for

this is this is very contrasting evidence

Amazon you know

it's persistent that helped Amazon

but knowing when to pivot helped

you know Slack

YouTube and Instagram so yeah

like there is a lot of advice like

you know which applies only at a point of time

so what I Learned is to take everything with a

with a pinch of salt and like

you know lean more on my intuition

at the end of the day you

your company doesn't succeed because

you know you followed someone else's playbook right

like you know

it succeeded or failed

because you had the courage to trust your own judgement

when no one else could see the answer

and you have to own own this part of the company

building both the mistakes and breakthroughs

and that's what makes it yours like

you know be it success or failure

like you know

it has to be yours and like

there's only so much advice out there

that you can apply it for every point of time

and yeah and for me like

you know that's the real job of a founder like

you know you need to listen deeply to all these advices

but you need to also think independently

and bet on your own insight when it matters

hundred percent I think everyone loves giving advice um

but advice is only applicable

if it's in the right context

um exactly yeah

and yeah we've

I've heard through kind of the show

I've heard so many founders saying like

oh we got like this advice

blah blah

and then we built our product this way

and then we had to scrap it all over again

um yeah

so like this point definitely resonates kind of across the board

since you're kind of the CTO

and I don't really get to talk to CTO super often

I'm so curious to see your perspective on how you think

the way people build AI products will shift in the future

so kind of there's this two conflicting trends of like um

of using kind of general purpose models

that will be great for everything

or using very specific models

and like having a bunch of these together to

in order to create a very robust

um kind of answer based on whatever they query

um yeah

I would love to get your take on like

where you where you see the world of AI going more towards the latter

because there are only few big players like

you know

that have the dollars that can build these general purpose models

startups or like you know

teams that are ambitious enough to solve for a particular vertical

that are gonna like

you know fork these or like

you know take these and build a like

you know fine tune or more narrower version of these

like you know that work well in a particular use case

we we could see a lot more of that

like I I mean

I I could quote one example

like we we know that there is

there are a lot of you know

text to speech models

one of my VC batchmate

they're building a text to speech

but their selling point is like

you know it will be more emotive

it's like you know

humans like you know

they are training on the podcast data or like

you know real conversations so that the text to speech has emotions

I would say I'm seeing a lot

lot more of initiatives or a lot more startups coming up in this space

like trying to build something for narrower use cases

and with AI it has become easier to build these because like

you know the time to product and time to market has

has dropped like you know

by an insane amount and

you know you can test these ideas quickly

you can get some feedback from customers

you worked at really massive and established companies

and now you're building from zero

maybe if we flip the script on the question

what's one thing you miss about big company life

and one thing you would never ever go back to the co workers part

this this could be so like

you know if uh

I get a chance to work with them again in

in this in in my company or we hire some incredible folks in

in our company as well so yeah

like that's one thing uh

that I miss and I don't know if there's a single thing like

you know I wouldn't go back

but once you start building your own company

it's it's a bit hard to imagine working on someone else's dream

I mean unless it aligns so deeply that it

it feels like your own

the sense of ownership

and the sense of creating something from scratch is so

yeah yeah

and yeah and so beautiful

so that's why we see so many founders

like they exit their company and a few years later

they're like oh

I'm building something new again

alright before I let you go

um I

how we like to end this is we like to end with a few

like rapid fire questions

so just say the first thing that comes to mind

don't really need to explain

you can you can if you want to yeah OK

would you rather debug a million line of legacy code

or sit through an 8 hour investor workshop

okay who's

who's giving the workshop

um an investor that you don't love too much

I love learning things even if it's a person that I don't love

I think a different perspective always helps

so probably the investor yeah

okay yeah

this is a non typical answer from an engineer

but yeah like people can give

give you a lot more than you know

than either money or work right

like you know

you get to learn from their experiences and perspectives

yeah hundred percent

and legacy code is just painful

what was the first thing you ever built with code

probably a HTML website like yeah

it's it's dumb yeah

yeah I mean

like I I haven't had access to a computer since I was in college

so it's it's

it's not like I I

I'm a wiz kid coder like I

I'm late to the coding game

so that's why it's probably just a simple HTML page

better late than never than never

Tia this has been super fun

yeah thank you so much for inviting

that's a wrap if you like this episode

please hit the subscribe button

it helps us bring on more awesome guest

level up production and drop new series you wanna watch

see you in 2 weeks