Trust slider over Clairic: 102 manual reviews catch 74 of 76 extraction errors
michellechen · x · 2026-10-21
A developer built a trust slider on top of Clairic, System One's new class of extraction models. Dragging a confidence threshold across 1,000 answers extracted from 200 messy invoices controls how much is auto-filed vs. sent to a human. At 0.9, 898 answers were filed automatically, 102 routed to review, and 74 of the 76 mistakes were caught — near-complete error capture for a small review cost.
Related event: Clairic's trust slider routes document extraction by confidence(2 posts)→
More from coding & agent
- LukeW maps the five UI problems every AI-native company faces with agents — LukeW · 2026-10-21
- Claude Code over ADB guts a bloatware-laden Android TV, freeing ~90% more RAM — edwinarbus · 2026-10-21
- Developer uses Claude to fix his adware-slowed Xiaomi TV, releases tool for any Android TV — edwinarbus · 2026-10-21
- Open-weights Ornith-1.5-122B debuts with 84.5% SWE-bench Verified, runs on a single DGX card — _akhaliq · 2026-10-21
- People are using Claude Code to debloat ad-ridden smart TVs — edwinarbus · 2026-10-21
- crossSessionInbound=accept flagged as the key Claude Code primitive behind hive swarms — kieranklaassen · 2026-10-21