Sightline – Zero-latency vision verification engine using Yjs CRDTs and WebRTC

Sightline launched a zero-latency vision verification engine that uses Yjs CRDTs and WebRTC to let humans correct low-confidence object detections in collaborative real time, flagging objects below 85% confidence or marked uncertain. The system exposes a read-only CRDT projection that never mutates collaborative state on export, pays a $50 bounty per verified object, and computes all metrics client-side from live CRDT state with zero external analytics and zero server cost. Developers can submit images through a POST endpoint at api.sightline.ai/v1/crdt/ingest with a confidence_threshold parameter of 0.85.

Collaborative Human-in-the-Loop Vision Verification Click a quick verbal instruction or type a patch command to update the selected object in CRDT history: Objects below 85% confidence or marked uncertain Read-only CRDT projection — export never mutates collaborative state. Bounty: $50 per verified object. curl -X POST https://api.sightline.ai/v1/crdt/ingest \ -H "Authorization: Bearer sk live ..." \ -F "image=@warehouse feed 04.jpg" \ -F "confidence threshold=0.85" All metrics computed client-side from live CRDT state. Read-only projection — zero external analytics, zero server cost.