Dragoneye

Dragoneye is a vision AI platform that lets developers and small teams build custom object detection and classification models from plain English descriptions — no labeled images required. Instead of collecting thousands of training examples, you describe what you want to detect, and Dragoneye's zero-shot engine instantly creates a deployable model. Models are served via a managed API, eliminating infrastructure overhead. The platform recently introduced Attribute Detection (beta), which extracts structured attributes from detected objects, and a conversational Model Builder that lets you define models interactively. Dragoneye is Y Combinator-backed and targets use cases like construction safety monitoring, retail inventory, and content moderation. Key features include zero-shot detection from text, instant API deployment, video processing at up to full frame rate (billed at 5 FPS base), Python and Node.js SDKs, and an interactive Playground. The Pay-as-you-go tier starts with $10 free credits and charges $0.046/min of video and $0.005/image. The Scaled tier offers volume discounts, model finetuning (beta), and priority support. Compared to traditional vision platforms like Google Cloud Vision or AWS Rekognition, Dragoneye removes the need for training data and custom model training pipelines. However, it lacks on-premise deployment and advanced enterprise SLAs, making it best for rapid prototyping and low-to-medium volume production use cases.