{ "schema_version": "paper_public_manifest_v1", "paper_id": "qianxun-general-brain-data-loop_2026_07", "slug": "qianxun-general-brain-data-loop", "title": "Qianxun Intelligence: A Double Bet on a General-Purpose Brain and a Closed-Loop Data System", "authors": [], "source": { "arxiv_id": "", "pdf_url": "", "project_url": "https://mp.weixin.qq.com/s/jx2Ne2RlQnnqEha6jAaXLQ", "github_url": "", "huggingface_url": "", "original_source": "https://mp.weixin.qq.com/s/jx2Ne2RlQnnqEha6jAaXLQ" }, "site": { "post_url": "/posts/qianxun-general-brain-data-loop", "canonical_url": "https://haiguangboy.com/posts/qianxun-general-brain-data-loop", "cover_image": "https://static.haiguangboy.com/papers/qianxun-general-brain-data-loop/cover.webp" }, "taxonomy": { "domain": "embodied_ai", "track": "vla", "tasks": [ "embodied_ai", "vla", "action_generation", "robotics", "Embodied Intelligence", "Robot Data", "General-Purpose Robots" ], "related_topics": [ { "paper_id": "wx_星河频率_20260718_2026_07", "title": "Robots 'Standing in the Light': Lingchu Intelligence Enters Optical Module Production Lines", "url": "https://haiguangboy.com/posts/lingchu-optical", "relation": "contrast", "summary": "★Route Bet: Betting on a general-purpose brain rather than a vertical industry brain—'The only opportunity in the era of large models is a general-purpose model' contradicts Route Bet: Specialize first, then generalize—a data flywheel that penetrates a single high-value scenario before migrating to adjacent tasks", "strength": "strong" }, { "paper_id": "latepost_xuhuazhe_202603_2026_03", "title": "latepost_xuhuazhe_202603", "url": "https://haiguangboy.com/posts/latepost_xuhuazhe_202603", "relation": "same_track", "summary": "★Route Bet: Betting on a general-purpose brain rather than a vertical industry brain—'The only opportunity in the era of large models is a general-purpose model' validates Route Bet: AI-native triple negation—not robotics/not autonomous driving/not prehistoric deep learning", "strength": "strong" }, { "paper_id": "sunday_blog_20260717_2026_07", "title": "sunday_blog_20260717", "url": "https://haiguangboy.com/posts/sunday_blog_20260717", "relation": "same_track", "summary": "★Distributed data collection network scale: 100+ cities, over 400,000 collection points, 1,000+ staff validates Recipe's asset structure: sensorized human data + self-developed collection hardware/screening systems/processing pipelines + fleet closed loop", "strength": "strong" }, { "paper_id": "wx_界面新闻_20260605_2026_06", "title": "wx_Interface News_20260605", "url": "https://haiguangboy.com/posts/wx_界面新闻_20260605", "relation": "same_track", "summary": "Core judgment: The bottleneck lies not in hardware but in the brain; the robot market size is only about 1% of the automotive market validates Route Bet: Build the brain, not the body—betting on a foundation model company identity, rejecting the motion capability track", "strength": "strong" }, { "paper_id": "orca_2026_07", "title": "π0.5 Shivers in Place When Failing to Grab a Spoon, While Orca Goes Further with Physical Intuition Learned from Watching Videos", "url": "https://haiguangboy.com/posts/orca", "relation": "same_track", "summary": "The key to a world model is a readable state", "strength": "medium" }, { "paper_id": "t_rex_tactile_reactive_dexterous_manipulation_2026_07", "title": "T-Rex: Why Touch Should Be Modeled Separately", "url": "https://haiguangboy.com/posts/t_rex_tactile_reactive_dexterous_manipulation", "relation": "extension", "summary": "T-Rex: Why Touch Should Be Modeled Separately", "strength": "medium" } ] }, "ruling": { "importance_score": 3.0, "one_sentence": "Qianxun Intelligence's closed-loop data bet" }, "asset_base_url": "https://static.haiguangboy.com/papers/qianxun-general-brain-data-loop", "assets": [ { "type": "public_brief", "object_key": "papers/qianxun-general-brain-data-loop/public_brief.md", "content_type": "text/markdown; charset=utf-8", "upload_status": "uploaded", "bucket": "paper-assets", "url": "https://static.haiguangboy.com/papers/qianxun-general-brain-data-loop/public_brief.md", "role": "public_brief", "size_bytes": 4629 }, { "type": "public_brief", "object_key": "papers/qianxun-general-brain-data-loop/public_brief.en.md", "content_type": "text/markdown; charset=utf-8", "upload_status": "uploaded", "bucket": "paper-assets", "url": "https://static.haiguangboy.com/papers/qianxun-general-brain-data-loop/public_brief.en.md", "role": "public_brief_en", "size_bytes": 5740 }, { "type": "public_manifest", "object_key": "papers/qianxun-general-brain-data-loop/public_manifest.en.json", "content_type": "application/json; charset=utf-8", "upload_status": "uploaded", "bucket": "paper-assets", "url": "https://static.haiguangboy.com/papers/qianxun-general-brain-data-loop/public_manifest.en.json", "role": "public_manifest_en", "size_bytes": 5318 }, { "type": "public_manifest", "object_key": "papers/qianxun-general-brain-data-loop/public_manifest.json", "content_type": "application/json; charset=utf-8", "upload_status": "uploaded", "bucket": "paper-assets", "url": "https://static.haiguangboy.com/papers/qianxun-general-brain-data-loop/public_manifest.json", "role": "public_manifest", "size_bytes": 5798 } ], "published_at": "2026-07-21T22:35:35+08:00", "created_at": "2026-07-21T22:35:35+08:00", "updated_at": "2026-09-03T12:13:52+08:00", "analyst_take": { "type": "author_opinion", "text": "The most important thing in this article is not the number '400,000 collection points,' but that it packs the collection end, processing end, training end, and deployment end into the same closed-loop data system. What it truly bets on is: as long as this loop can keep shortening, the general-purpose brain can achieve stronger reuse efficiency than vertical routes; if it cannot shorten, then the scale numbers are just scale numbers and will not automatically become a moat.\n\nIt is the direct opposite of 'specialize first, then generalize,' but the divergence from other general-purpose brain routes is not about 'whether data is needed,' but about 'whether the data forms a closed loop.' I value the speed of feedback from collection to deployment more than the scale of collection points alone; only when the success rate of new tasks, the required data volume, and the iteration cycle all continue to improve can this route truly be considered viable." } }