ResearchBenchmarksDirectReports on Production DirectReports (test)Follow79.5F1 ScoreINIT75.23676.34377.4578.557Jun 29, 2026Evaluation ResultsMethodMethodLinksF1 ScoreDeployment Delta (∆Deploy)INITDescription type=Initi...Description type=Initial LLM description (iter-0)2026.0679.5—HUMANDescription type=Human...Description type=Human-written2026.0679.4—NEWPipeline=automated des...Pipeline=automated description optimization pipeline, Optimization=trainmax, standard2026.0678.63.2OLDDescription type=Pre-e...Description type=Pre-existing system baseline2026.0675.4—