In September of 2025, I implemented Velocity Labs for my team in order to manage new and ongoing production incidents occurring on our legacy platform every week and provide ourselves a chance to develop revenue-driving software for Affirm in the new year.
Over the span of that quarter, we achieved 99.9% availability (up from as low as 99.7%) and worked through what must have been 10 production incidents (one of them being a blocker to launch with hundreds of thousands of new merchants with Intuit), all while managing the collaborative effort to revamp the backend architecture of the merchant portal and laying groundwork toward a 99.99% availability target.
Velocity Labs had a simple objective: get engineers on my team developing more with less, and embracing AI-driven development. Coupled with 1:1s with each of my direct reports, we did the work required to work effectively with AI and change our mindsets from “I can't, I'm still fixing this bug” to “I think we can actually get this done in two days.” And I did this by going to San Francisco for two weeks, changing my mental model for how AI works under the hood, and learning that AI is about organizational intelligence. The reason we weren't getting as much done before wasn't because we had adopted a brittle legacy system; it was because, organizationally, we did not have a growth mindset.
Shortly before departing Affirm, I initially gained traction with engineers in the larger org to talk about how we could find new ways to grow the engineering practice using AI. And as part of a company-wide AI Paved Paths week, I started to develop an idea for an application that would allow Affirm to expand its merchant platform for in-store mobile use, managed by merchant users.
I'll end with this:
“The intelligence age is helping us move faster, but the real gains in adopting AI technology will help us unlock products that make it easy to try new things and extend trust with customers.”