Adrien
200 posts

Adrien
@adzeb_
Data science and AI @airwallex



There’s a tension in fintech that most companies avoid addressing directly, but @scott_shannon, CPO of @Airwallex, leaned into it. If you are building infrastructure and applications at the same time, aren’t you inevitably competing with your own customers? That assumption only holds if you believe the value sits in the product layer, but what became clear in this conversation is that the real leverage is shifting underneath, into the infrastructure itself. If your infrastructure is truly differentiated, then restricting access to protect your own applications becomes a limiting strategy, because no single company can build the full range of products that an open ecosystem can create on top of the same foundation. This is the AWS analogy playing out in fintech. Amazon did not weaken itself by opening AWS; it expanded the market by turning internal capabilities into external infrastructure, allowing thousands of companies to build on top of it while still continuing to build its own products. The same logic applies here. Airwallex offering expense management does not eliminate partners like Brex; it establishes a baseline while enabling others to go further, faster, and in more specialized directions. The real shift is from vertical control to horizontal enablement, where the goal is no longer to own every use case, but to become the layer others rely on to build theirs. And that changes the surface area of fintech entirely. Because once financial infrastructure becomes as accessible as cloud infrastructure, every software company can embed financial services without becoming one, which is where the next wave of expansion actually begins. That’s why this conversation matters 👇 youtu.be/g19KZzvmPWk?si…

I wrote about how I'm thinking about DS transformation at @airwallex . tl;dr: DS Analytics becomes systems of decision. DS Algorithm becomes systems of action. The job shifts from running analyses to building the systems that produce reliable decisions and actions at scale. Read more in link below



Nevertheless, we still think it won’t be long until we see AI middle-managers. This version of Claude had no real training to run a shop; nor did it have access to tools that would’ve helped it keep on top of its sales. With those, it would likely have performed far better.

I generally like Anthropic: but the more they paint a dystopian future where AI “manages” people (“AI middle-managers”) the more I am starting to think they are losing their marbles. LLMs is a tool humans should use. The tail should not wag the dog; Anthropic should know better









