SwoVid

SwoVid (from Sanskrit Swopajna — created by one's own; Videshiya — foreign) is an AI content authenticity platform. It detects whether images, videos, documents, and text are AI-generated or human-created, and it does it where you actually encounter content — in the browser, ambiently, with zero manual action.

Rather than relying on a single detector, SwoVid reads every major open provenance standard simultaneously: C2PA cryptographic content credentials (the highest-certainty signal), IPTC / EXIF metadata, invisible watermarks (Stable Signature from Stability AI and Video Seal from Meta AI), and a proprietary fine-tuned detection model exported to ONNX to run in-browser, in the extension, and on-device with no server dependency.

Every verdict is one of five honest states with a confidence level — Verified Human, Verified AI, Likely AI, Possibly AI, or Unverifiable — rather than a misleading binary. The product ships as a Chrome extension, a web app, and a mobile app from a single Turborepo monorepo, all sharing one DetectionOrchestrator abstraction so any detection layer can be swapped or added without touching application code.

Built solo — architecture, detection-stack research, UI/UX, and go-to-market.