CorgiAI is a B2B SaaS company that provides payment fraud prevention solutions using proprietary AI. The founder previously developed a fraud prevention solution at Stripe that improved performance by 78%. CorgiAI's solutions aim to bring more balance between preventing fraud and allowing legitimate revenue by using lightweight, end-to-end, customizable, and explainable AI-based models that integrate directly with payment providers. The company is currently processing $14 million in monthly GMV and aims to raise $2 million to expand its team, customer base, and global integrations.
2. FAE
Founder
Saif Farooqui
Former APAC Lead Data Scientist
for Stripe Radar
Discovered shortcomings in fraud
prevention AI
Invented solution improving
performance by 78%
3. Fraud in payments is
hard to solve, adversarial
and constantly evolving
Details
9. GTM Strategy
We grow by integrating directly with
payment providers
Evidence of reduced fraud + increased revenue for individual
businesses builds the pathway for these partnerships
12. “CorgiAI can help us solve fraud problems for
the underloved SMEs segment”
Product lead at global payment provider
“CorgiAI is poised to become the de facto
payments firewall for the industry”
Investor
13. Raising $2 million
Build Team
Engineers, Data Scientists, Sales
GTM
50 customers, $500k MRR by 2025
Gateway Integrations
Checkout.com, Paypal, Shopify, Airwallex
Global Expansion
US, EU, Japan, Korea, Australia
15. Top of Fraud Funnel
Digital payment flows are complex, involving consumers, payment processors,
issuers, acquirers and merchants
Payment fraud usually occurs after the merchant has received the payment, and
stems from disputes
Digital Payments + Fraud
Commerce Portal Payment Processor
Acquirer
Card Network
Issuer
Merchant Account
Customer Account
Merchant
Customer
Merchant delivers goods/services to customer
Payment
Debit
Verification
Verification
Fees
Payout
Disputes
Friendly fraud, card testing, account takeover, refund fraud
Chargebacks
~50%
probability
Interception fraud / Triangulation fraud
Cardholder files dispute (~2% probability)
Credit
Debit
Fees
16. Fraud Funnel
Dispute Chargeback
Fraud Actor
Consumer
Merchant
Fraud Loss
(Good/Service)
Refund
Payment
Fraud Gain
(Good/Service)
90 days
Fraud Detection + Prevention by
Payment Provider Risk/Fraud Teams
90 + X days
(X could be ∞)
The current fraud detection + prevention funnel for digital payments is extremely long, and in some cases endless (which also means
uncapped fraud losses)
Consumers have up to 90 days to dispute charges, and especially in the case of stolen credentials, it can take even longer to detect fraud
After chargeback occurs, payment provider internal fraud/risk teams take significant time + effort to identify fraud, and then work on
primarily retrospective mitigation methods, all this while other fraud actors are concurrently looking for exploits
17. Fraud in payments is hard to solve, adversarial and constantly
evolving
What makes it worse is that most merchants are unaware of fraud
exposure + losses
For e-commerce, scam + fraud in payments leads to 3x losses
(monetary chargebacks, lost goods/services, penalties)
Compounded by human capital cost of dealing with disputes +
fraud, meaning productivity losses
This is nothing new, fraud has been around as long as money has,
but there are still no robust solutions
Fraud is a Problem
>$3.00
Every $1 of fraud transactions
costs a store >$3
Source
80%+
Real-world data estimate
percentage of chargebacks that
are fraud-related
Source
18. Merchants and businesses are being squeezed due to inflationary pressures
With profit margins shrinking, preventing fraud is more important than ever to reduce losses
Payment providers have had trouble tackling fraud in newer markets (APAC)
Emergence of new payment providers (especially eWallets) throughout APAC, without the
resources to combat fraud and the long-term implications of stunted business growth due to
customer trust concerns
We can do the dirty work and fight the bad actors, leaving merchants to grow their companies and
win back those profits
Why Now?
19. The largest problem currently facing automated payment + merchant fraud solutions
is the reliance on legacy modelling approaches,1
and this inertia makes it difficult to
adapt known working methods to new markets and new problems2
They tailor the problem to the solution. It should be the other way around.
The net effect? AI/ML solutions that make mistakes in both directions, not catching
enough fraud (low recall) and blocking too much good revenue (low precision)
We’ve already seen error-strewn models our clients use with ~10% precision (only
10% of the predicted fraud is actually fraud) and ~15% recall (only 15% of the actual
fraud out there is caught), significantly worse than a coin flip (50%)
1
Models trained on payments data from original markets (US, EU) are assumed to maintain effectiveness in newer markets, but the in-built
data bias leads to poorer results.
2
For instance, current fraud prevention efforts consistently struggle in dealing with bad actors from Vietnam.
Limitations of Current Solutions
20. Manual
Large parts of the processes (>75%
in some cases) are dependent on
human solutions (labelling,
evaluation, validation)
Limited Scope
Current add-on solutions focus on
endpoints of the fraud intelligence
spectrum, don’t consider the complete
fraud user journey
Capital-intensive
Cost of human labelling, costs
associated with implementation and
forward deployed engineering, not to
mention significant server costs
Legacy Dependence
Payment providers index heavily on
solutions like 3DS1
which essentially adds
an extra (costly) layer of checks, Not a
comprehensive fraud solution.
1
3D Secure, technical standard that adds a layer of security in online credit and debit card transactions. Similar to 2FA.
More details at https://business.ebanx.com/en/resources/payments-explained/3d-secure
Limitations of Current Solutions
21. Merchants find these black-box solutions difficult or impossible to understand, so
when it works, it works, but when it fails they have no idea how to fix it
They end up with no solution and are forced to compromise, by either being very
aggressive on fraud leading to lost revenue, or not being aware of fraud and incurring
huge fraud losses
Customers are Lost and Confused
Fraud Losses
“We know our fraud rate is very high, but all the
solutions we’ve tried were blocking too way much
revenue.” - e-commerce CEO
Chargeback rate was 0.8%, translating to
~$960k lost per year
Every $1 of fraud transactions costs a merchant
>$3, so actual fraud loss was ~$2.88m
Lost Revenue
“We’re scared of chargebacks. We paid for ML
models, set strict rules, and ran extra verifications.
Keeps our fraud rate low.” - marketplace CFO
The super-strict fraud prevention rules led to
an additional 5% potential revenue blocked
Based on projected revenue, this translated to
annual lost revenue of ~$1.3m
22. 1
Clustering Optimized Rule Generation Intelligence (patent pending)
2
More details in Technical Deck
Introducing CorgiAI1
Corgi is a user-centric API-based end-to-end fraud detection + prevention
suite, built on a core of customizable and explainable AI2
We make it easy for merchants to understand and prevent fraud, and
unblock lost revenue
Lightweight
End-to-end
Simple
Customized
Transparent
Upstream data processing + filtering optimizes ML runtimes
Integrates directly with your payment provider through an API
Intelligent clustering adapts the algorithm to the problem space
Automated fraud detection and insights all the way to blocking
Explainable AI + observable performance metrics