Scott MacVicar, Head of Developer Infrastructure at Stripe, has one big recommendation for engineering leaders who are rolling out AI with their teams: accelerate.

"My advice for anyone thinking about trying AI coding is, move faster on this," he said.

Stripe itself adopted Cursor quickly but carefully. The company is known for 99.9999% API uptime and products that set the bar for developer experience. They couldn't sacrifice either.

The key was getting all of Stripe's more than 3,000 engineers started within the right guardrails. Stripe preloaded Cursor on every machine, used Cursor Rules to provide codebase context, and adapted code review practices to maintain quality at higher velocity. Along the way, they also discovered something unexpected about who was getting the most value from AI.

#Setting engineers up for success

Stripe had high conviction early on that AI would fundamentally change how its engineers work. The question was how to manage that shift across more than 3,000 developers. The approach they took: Make Cursor work well for everyone from day one.

"When people start, they've got VS Code, IntelliJ, and Cursor all installed on their machines, and they go through a lab that teaches them how to use the Stripe developer environment," said MacVicar.

Everything is preconfigured. Stripe uses Cursor Rules to give Cursor context about Stripe's codebase and coding standards. They've empowered teams to add their own rules too, while calibrating as an organization on how comprehensive to make them. This approach has allowed Stripe's developers to become productive with AI coding tools on day one and then go even further from there.

Cursor adoption spread through an influencer system. Power users shared their workflows in lunch-and-learns, showing teammates how to run multiple agents in parallel or write better prompts.

#Adapting code review for AI coding

AI allowed engineers to write more code. But that wasn't the goal. Stripe didn't want higher velocity to mean lower quality. Their solution was adapting code review to keep pace.

Stripe now uses LLMs to help reviewers work more efficiently. AI flags complex methods or risky files, directing reviewers' attention where it's needed most. Engineers have also gotten more comfortable pushing back on code that isn't ready, whether it came from a person or an agent.

Stripe engineers have embraced this new way of working: On an internal survey, developer sentiment recently hit a five-year high. MacVicar said, "People are really excited about the tools they're getting."

#The tenure advantage

Stripe is now focused on helping engineers get the most value they can from Cursor. The early findings have surprised MacVicar. He expected junior engineers to benefit most, using AI to compensate for limited experience. Instead, he's seen the longest-tenured engineers make the biggest productivity gains.

In response, MacVicar is exploring whether new hires should spend their first weeks learning the codebase before getting Cursor access. They could use that time to build up the context needed to get the most out of AI tools.

It's one example of why MacVicar advises other engineering leaders not to wait to have everything figured out with AI. Many of the most important questions they'll need to answer only emerge once engineers are actually using the tools. He sees getting started now as the best way to optimize for both present possibilities and the future of software engineering.

If you're interested in accelerating your engineering team's adoption of agentic coding workflows, please reach out to our team to get started with a Cursor trial.