Boosting Pagos’ Engineering Capacity with AI

At the most recent Pagos company offsite in Washington, DC, our Engineering team came together from across the globe. We’re a remote-first team spread across 10 countries, so dedicated in-person co-working opportunities are rare—and extremely valuable. Making the most of this time together, we prioritized an engineering Code Retreat: interactive sessions designed to encourage experimentation and inspire new ideas. This time, we designed the session around our ongoing integration of AI.

A Traditional Code Retreat Format—With a Twist

Code Retreats typically involve a series of paired programming sessions focused on solving a single problem—Conway’s Game of Life. The goal isn’t to correctly identify perfect solutions, but to emphasize principles like simple design, test-driven development, and pairing. In each of our retreat sessions, engineers tackled the given problem under new constraints designed to force creative thinking.

When we facilitated a similar activity at a past offsite, it helped engineers from different subteams get to know each other better and bond. This time, we added a new goal: experiment with and push the capabilities of AI.

AI at Pagos

As a small and nimble team with an industry-disrupting suite of products, we’re always looking for ways to punch above our weight. At Pagos, we’ve embraced AI as a force multiplier for our Engineering team in particular. With AI tools, we’re operating more efficiently, iterating faster, and delivering value at scale—without compromising on quality. We know human oversight makes AI powerful, and our team of engineers and payments experts are bridging the gap between machine-generated code and real-world production needs.

WIth this in mind, we’re actively using Cursor, an AI-powered agentic IDE, and are delighted with the measurable impact it has on individual productivity. This retreat was about taking that one step further—working together to understand not just what Cursor can do, but how it can break down or surprise us when pushed harder than usual.

Exploring AI-Powered Development

The first part of the day followed the standard Code Retreat schedule: engineers paired together, accomplishing the goal of continuing to build relationships. After lunch, however, pairing was over. Cursor was everyone’s pair now.

Session 1: Speed Coding with Cursor

For our first Cursor-paring session, we asked our engineers to implement additional features for the Game of Life as quickly as possible. Once completed, each engineer walked through their AI-generated solutions. The goal wasn’t actually speed—it was to prompt reflection on AI’s strengths and weaknesses. Specifically, while it may be easy to generate new code with these tools, that code can often be hard to revisit or maintain. Fast and easy doesn’t always mean good. 

Session 2: The One-Shot Prompt

Next, we introduced a constraint: engineers could only give Cursor a single prompt. They could write tests within the repository and define Cursor rules, but they could only interact with Cursor once. This pushed the team to explore guiding AI beyond basic prompting. While some created rules to influence Cursor’s behavior, most focused on refining their single prompt through iteration. It was a valuable exercise in prompt engineering and system design.

Session 3: Production-Ready & a Little Weird

In the final session, we dialed up the creativity with a fun challenge: write a production-ready implementation of Game of Life without referencing the game’s unique terminology. Even more, the definition of “production-ready” was left intentionally vague.

This was the final session of the day, and things got a little silly. One team member repeatedly asked Cursor to “make it more fun,” resulting in a Game of Life using emojis where the cells could have parties and went through market cycles which changed the rules for when cells were born, died, or lived. The exercise drove home that AI can be a co-creator, especially when prompted with humor, abstraction, or unexpected constraints.

Reflections and Looking Ahead

We didn’t walk away from the Code Retreat with a perfect process or AI playbook—that wasn’t the goal. Instead, it inspired fun teamwork and relationship building while giving us all a better understanding of how to better use tools like Cursor with intention. AI’s a part of how we work now, and this retreat gave us the space to test it together, break things, and build trust in each other and our tools.

Looking forward, we’re excited to keep integrating AI into even more of our workflows—always with a critical eye and an emphasis on human expertise. With AI by our side, we’re ready to build smarter, faster, and better.