We're not shy about it — AI tools are a core part of how we work. Not as a replacement for thinking, but as a multiplier for execution. Here's our honest breakdown.
Cursor for Day-to-Day Coding
Cursor is our primary editor. The inline chat, the codebase-aware completions, and the ability to ask questions about your own code without leaving the editor — it's genuinely changed how fast we move. We use it for everything from boilerplate generation to debugging obscure TypeScript errors.
Claude for Architecture and Review
When we're designing a system or reviewing a complex PR, Claude is our thinking partner. It's particularly good at spotting edge cases, suggesting alternative approaches, and explaining tradeoffs clearly. We paste in code, describe the context, and have a real conversation about the design.
GitHub Copilot for Repetitive Patterns
Copilot shines at pattern completion — writing the 10th similar function, filling in test cases, generating TypeScript types from a JSON example. It's not always right, but it's right often enough to save significant time.
What We Don't Use AI For
We don't let AI make architectural decisions. We don't ship AI-generated code without reading it. And we don't use it as a substitute for understanding the problem. The best AI-assisted code still comes from developers who know what they're building and why.
The Bottom Line
Teams that learn to use AI tools well will outship teams that don't — not because AI writes better code, but because it removes friction from the parts of development that don't require deep thought. Use that freed-up time for the parts that do.