Meta-Harness R&D: Enterprise-Grade Self-Improvement for Long-Horizon AI Workflows
OpenAI News / Jun 23, 2026
- 検証と監査
- 段階的デプロイ
- ヒューマンインザループ
Matched posts: 2
Building self-improving tax agents with Codex
OpenAI News / May 27, 2026
- Processed ~7,000 returns in pilot
- Saved ~1/3 practitioner time; drafts up to 97% accurate
- Codex converts production corrections into evals and candidate fixes
Previous1 / 1Next