{"version":1,"type":"story","url":"https://digestai.news/story/zhipu-automates-infrastructure-with-glm-5-3s-outer-rsi-loop-in-under-t","json":"https://digestai.news/story/zhipu-automates-infrastructure-with-glm-5-3s-outer-rsi-loop-in-under-t.json","markdown":"https://digestai.news/story/zhipu-automates-infrastructure-with-glm-5-3s-outer-rsi-loop-in-under-t.md","slug":"zhipu-automates-infrastructure-with-glm-5-3s-outer-rsi-loop-in-under-t","headline":"Zhipu automates infrastructure with GLM-5.3’s outer RSI loop in under two weeks","summary":"Zhipu AI, developer of the open-weight LLM GLM-5.3, detailed how it used its own model to accelerate internal infrastructure work. The team deployed an **Infra Agent** powered by GLM-5.3 to optimize GLM-5.3-Flash—a faster, cheaper variant—cutting deployment time to **less than two weeks**. The agent handled analysis, hypothesis testing, and code changes, while engineers set objectives and the experimental environment provided feedback. Through this **outer reinforcement learning (RSI) loop**, Zhipu achieved **tripling end-to-end throughput** compared to the baseline, demonstrating how AI labs can automate their own scaling processes.\n\nThe company emphasized three key principles for effective automation: **localized feedback** (tied to specific system parameters), **inexpensive and timely validation**, and **objective verification** (using reference implementations and metrics). While Zhipu’s engineers now rely on GLM-5.3 daily for coding, they caution that human oversight remains critical for high-level design decisions—though they acknowledge this advantage may shrink over time.","keyPoints":["Zhipu’s Infra Agent, powered by GLM-5.3, cut GLM-5.3-Flash deployment to under two weeks, tripling throughput","Feedback loop involved engineers defining objectives, the agent analyzing/hypothesizing, and experiments verifying results","Company warns human comparative advantage in design may fade as AI systems advance autonomously"],"whyItMatters":"Zhipu’s self-automation proves AI labs can accelerate their own development cycles, reducing manual labor and speeding up model iteration. This trend could lower costs for competitors and force open-source models to innovate faster, but it also raises questions about long-term human roles in AI R&D.","category":{"slug":"research","name":"Research","url":"https://digestai.news/category/research"},"entities":{"companies":["Zhipu AI","Google","Planet","Physical Intelligence","Periodic Labs"],"models":["GLM-5.3","GLM-5.3-Flash","Kimi 2.6","GPT-6 Astra","Claude Fable 5.1","Periodic Neon"],"people":["Michael Levin","Perry Dong","Chelsea Finn"]},"firstPublishedAt":"2026-09-28T12:32:03Z","updatedAt":"2026-09-28T12:32:03Z","sourceCount":1,"hasPrimarySource":false,"sources":[{"outlet":"Import AI","title":"Import AI 474: Platonic mindspace; TPUs in space; Zhipu starts an outer RSI loop","url":"https://jack-clark.net/2026/09/28/import-ai-474-platonic-mindspace-tpus-in-space-zhipu-starts-an-outer-rsi-loop","publishedAt":"2026-09-28T12:32:03Z","type":"newsletter","primary":false,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"Zhipu automates infrastructure with GLM-5.3’s outer RSI loop in under two weeks\", 28 September 2026, https://digestai.news/story/zhipu-automates-infrastructure-with-glm-5-3s-outer-rsi-loop-in-under-t","publisher":"Digest AI","title":"Zhipu automates infrastructure with GLM-5.3’s outer RSI loop in under two weeks","datePublished":"2026-09-28T12:32:03Z","url":"https://digestai.news/story/zhipu-automates-infrastructure-with-glm-5-3s-outer-rsi-loop-in-under-t"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}