Without evals, recursive agent self-improvement is just LOC with no value
omarsar0 · x · 2026-11-16
Elvis Omarsar (omarsar0) distills weeks of conversations with agent startup builders into one thesis: without evals, the recursive self-improving loop is just more lines of code without clear value — 'you can't effectively automate what you cannot properly measure.' He argues any product used for agent automation should have first-class support for building and measuring evals, and is inviting founders to chat about collaboration opportunities.
More from coding & agent
- Hawkeye brings hardware-aware kernel optimization to coding agents, at NeurIPS — simonguozirui · 2026-11-16
- Same analytics agent task: Sonnet 5.5 costs ~12x more than Haiku 5.5 — sh_reya · 2026-11-16
- Jiff 1.0 released: Rust datetime library already powering uv, Cargo and more — charliermarsh · 2026-11-16
- Berth v0.6.0 leases one RTX 4090 across 23 engines, 143 lease-hours a week — Distinct-Pie2389 · 2026-11-16
- Agent notices Mac battery at 3% and pauses work to warn user — adonis_singh · 2026-11-16
- Three classes of programming languages in the AI era: natural language, performant code, and specs — dbreunig · 2026-11-16