#agent-testing

Shiplight's Pivot to Agent-First Testing: Lessons from a Year in the AI Trenches

A year ago, Shiplight bet on humans authoring tests. Now? AI agents rule, with Plugins turning brittle scripts into spec-driven, self-healing verifications. Here's the gritty why and how.

AI Tools

Sculpting AI Agents with Precision Evals: Deep Agents' Blueprint

Imagine AI agents that don't just pass tests—they master real-world chaos. Deep Agents shows us how evals aren't checkboxes; they're the chisel carving tomorrow's intelligence.