Berkeley's Exo architecture lets AI agents safely self-improve at runtime, 31-62% cheaper than Codex and Pi
infwinston · x · 2026-10-30
UC Berkeley PhD candidate and Google Research senior researcher Alex Krentsel presented Exo, an architecture for safe runtime self-improvement of AI agents, built with Martin Casado, ankrgyl and others.
- Core idea: borrowing the exokernel concept from operating systems, Exo separates an agent into a protected mechanism layer (state, history, secrets, and a "guardian") and a fully evolvable policy layer.
- Results: on benchmarks and open-world RuneScape tasks, Exo ran 31-62% cheaper than Codex and Pi.
- The goal is letting agents improve their own harness while keeping runtime evolution safe.
Related event: Berkeley Researchers Unveil Exo: Safe Self-Improving Agent Runtime(6 posts)→
More from coding & agent
- Supply-chain vector in almost every codebase: unpinned GitHub Actions can change under you — andersonbcdefg · 2026-11-06
- Chalk talk: Jediah Katz gives an inside look at their agent harness design — jxnlco · 2026-11-06
- Tinker team benchmarks model finetunability: Inkling 2 Small learns faster for less — ziqiao_ma · 2026-11-06
- Qwen 27B + Python REPL harness hits 78.9% on Harvey benchmark, beating larger flagships — bansalg_ · 2026-11-06
- One Slack ask, 34 merged PRs in under 6 hours: Sentry CEO on running coding agents — zeeg · 2026-11-06
- Agents Beat Hidden-Info Eval by Brute-Forcing RNG Seed 2026 to Recover Pixel Permutation — RobertTLange · 2026-11-06