OpenAI says coding agents cut a 15-hour RNA-seq task to 15 minutes in lab software
rohanpaul_ai · x · 2026-07-29
OpenAI’s field report says coding agents are already helping scientists modernize old research software, speed up slow pipelines, and rebuild tools that were too time-consuming to maintain.
- One RNA-seq workflow dropped from about 15 hours to 15 minutes.
- Another project updated MHCflurry, cutting a training run from about a week to an hour while keeping 99.9% accuracy.
- The report covers 8 projects, mostly in life sciences; 5 used Codex alone and 3 combined Codex with other tools.
- The main limitation is still human judgment: scientists had to define the task, verify outputs, and catch confident-but-wrong agent behavior.
The broader takeaway is that coding agents can remove a lot of maintenance and optimization work in scientific software, but ownership and correctness still remain with researchers.
Related event: OpenAI Report: Coding Agents Reshape Scientific Computing(6 posts)→
More from coding & agent
- ThePrimeagen mocks AI coding: infinite tokens still produce an Electron app — seatedro · 2026-10-17
- Generative UI 101: Building Live Interfaces with Agents on Opus 5.5 and Rivals — EXM7777 · 2026-10-17
- MCP server review: gate tool writes on auth/traversal findings, not test score — SamuelFmedeiros · 2026-10-17
- DeepSeek shows rapid vision gains: agent self-checks screenshots to build 3D scene — cedric_chee · 2026-10-17
- He gave an AI agent phone and email access to collect a $500 unpaid invoice — curious_vii · 2026-10-17
- AI:AM transcript: agent lockouts, hidden valence, and formal proofs for Linux kernel bugs — The Cognitive Revolution · 2026-10-17