Main takeaways

‍

• Payment flows are complex, unwieldy beasts (more on that below)
• They’re generally driven by operational systems + processes and technology, specifically AI, will make an evolutionary impact
• AI models like contextual multi-armed bandits (MAB) or GPT powered solutions already exist in the market
• For every stage of the payments journey, there is an AI opportunity

‍

‍A quick note on AI in this hype cycle

‍

AI is a tool designed to make things faster, cheaper and better. While prohibitively expensive to build and iterate, it’s one of the rare cases where a solution in search of a problem is actually viable. There are lots of problems that are great to solve, but currently unsolvable and care must be taken to identify feasible + viable solutions.

Okay, back to payments.

‍

The Card Payments Ecosystem

‍

It’s complex.

‍

‍

1. The cardholder opens an account in the issuing bank and gets a debit/credit card.
2. Merchants partner with acquiring banks to open merchant accounts with payment processors.
3. The payment gateway accepts the purchase transaction and collects payment information.
4. It is then sent to a payment processor, which uses customer information to collect payments.
5. The acquiring processor sends the transaction to the card network. It also owns and operates the merchant’s account during settlement, which doesn’t happen in real-time.

‍

‍

‍

Where Does AI Fit in?

‍

AI in Action: The Current Landscape

‍

There are two major areas of focus when it comes to payment optimization. The first is improving payment conversion rates and the second is fraud ops optimisation.

‍

Players like Adyen are using Contextual Multi-Armed Bandits (MAB), an AI powered solution, to tackle the first problem. They collect and test contextual information like location, device type, and transaction history to dynamically select the best payment method for each transaction, maximizing conversion rates in real-time.

‍

Stripe, on the other hand, has exhaustive technical documentation used for its GPT4 powered Stripe Docs to enhance its products allowing users to manage fraud and increase payments conversion.  

‍

‍

CorgiAI offers an easy plug & play AI SaaS that simplifies the traditional fraud prevention process, usually a combination of an internal fraud team and legacy machine learning models plus a third party.

By generating new rules to maximise fraud detection precision and recall and deploying to continuously retrain and update the model - CorgiAI catches over 40% of fraud and recovers >50% of revenue where false rejects would lead to blocked transactions.