OpenAI says its agents can now complete well-defined research tasks that would occupy a skilled researcher for several days, meeting its internal definition of an automated research intern. By mid-August, its research organisation reportedly consumed 3.1 agent workdays for every human workday, although more than half of successful four-to-eight-hour tasks still required human intervention. OpenAI — 6 September 2026

• What changed: Agents have progressed from code completion toward running experiments, troubleshooting infrastructure and executing multi-day research tasks.
• Why it matters: Human effort is shifting toward task definition, evaluation and deciding which results merit further investment.
• Practical takeaway: Track estimated human effort, agent runtime, human interventions, accepted result and rework time for every substantial agent task.
• Analysis: These are internal measurements using OpenAI’s classifiers; they do not establish equivalent performance on ordinary enterprise repositories.