Prompts Stored in GitHub for Workflow Automation

At Every, Dan Shipper's team has developed a powerful approach they call "compounding engineering" - a methodology that ensures each unit of work makes future work easier and more efficient.

In a world where AI agents like Claude Code are handling most of the actual coding, the engineering team's role has shifted toward writing detailed PRDs (Product Requirement Documents) that specify what needs to be built. Rather than accepting this as their new job description, the team saw an opportunity to create a virtuous cycle of increasing productivity.

The core principle is simple but profound: for every unit of work you complete, you should make the next unit of work easier to do. As Dan explains, "You could just be like, 'Okay, that's my job now, I'm going to just write PRDs.' Or you could spend a little bit of time being like, 'There's a sort of platonic ideal of a PRD, and what I'm gonna do is write a prompt that can take my rambling thoughts and then turn that into a PRD.'"

The team maintains a GitHub repository of slash commands for Claude Code and other AI tools. These commands are essentially specialized prompts that transform rough ideas into well-structured PRDs, automate code reviews, and handle other repetitive tasks. By investing a small amount of time upfront to create these prompts, they dramatically reduce the effort required for all future similar tasks.

This approach has enabled a remarkably small team - just two engineers plus "fifteen Claude Code instances" - to build and ship Quora (their AI email assistant) with 2,500 active users processing millions of emails. The compounding nature of these efficiency gains means that each new feature or product becomes progressively easier to build, creating a powerful multiplier effect on the team's productivity.

The beauty of this system is that it creates a culture of continuous improvement, where engineers are constantly looking for ways to automate their own workflows. Rather than just completing tasks, they're building a system that makes them exponentially more productive over time - a perfect example of how AI can be leveraged not just to replace work but to fundamentally transform how work gets done.