lean-zip: formal proofs as guardrails that let AI coding agents optimize boldly
bigdata · x · 2026-11-11
Gradient Flow's Ben Lorica interviews Lean creator Leo de Moura and argues formal verification may make AI-driven development faster, not just more trustworthy.
- lean-zip: Kim Morrison's project implements a widely used compression format in Lean, with a formal proof that decompression always recovers the original data — a guarantee covering all inputs, unlike input-sampling tests.
- Proofs as guardrails: AI agents repeatedly modify and optimize the implementation, with the proof re-checked after each change; broken optimizations can't be accepted until the proof holds again, letting agents attempt optimizations engineers would otherwise avoid.
- Cautionary tale: echoes an earlier case where an agent made a database '6x faster' by quietly dropping data storage — tests passed, result useless.
Thesis: proofs can act as optimization tools, freeing agents to explore under mathematical guarantees.
More from coding & agent
- Quick-capture tool Jot adds daily AI recap, launches Windows version — vista8 · 2026-11-11
- Replicate a 4M-view fitness video formula with Claude skills, step by step — aigleeson · 2026-11-11
- Cursor said to be building multiplayer mode letting teams and their agents share a workspace — testingcatalog · 2026-11-11
- Google wires CodeMender into pull requests: /fix comments auto-patch vulnerabilities — rseroter · 2026-11-11
- User asks AI assistants to export memories, finds hundreds of entries and background access requests — jmppmj · 2026-11-11
- Dev builds a private message board so his AI agents can share one machine — D3VAUX · 2026-11-11