Shopify/liquid: Performance: 53% faster parse+render, 61% fewer allocations

Shopify CEO Tobias Lütke used the "autoresearch" agentic coding pattern — an agent brainstorming and running ~120 semi-autonomous experiments — to achieve 53% faster parse+render and 61% fewer allocations in Shopify's 20-year-old Liquid template engine, resulting in 93 commits. The key enabling factors were a robust 974-test suite and a benchmarking script, which turned "make it faster" into a concrete, agent-executable goal. Simon Willison highlights this as a model for how coding agents unlock high-ROI optimization work that's impractical to do manually, and how they're enabling people in high-interruption roles to contribute meaningfully to codebases again.