— ContextTwo markets, opposite needs, one studio.
Creators want to generate; journalists and platforms want to verify. We built one tool that serves both, with shared infrastructure and a single billing surface. Honesty about provenance is the product.
— ApproachOne pipeline, two modes.
- Generation mode: prompt → image / video with provenance metadata baked in.
- Detection mode: upload → ensemble score with confidence bands.
- Voice-over generation with Coqui-TTS, in the same API.
- Frame-level analysis for video — not just thumbnail-thinking.
— OutcomeA small studio with confident answers.
99.1%AI detection accuracy (in-distribution)
~3¢avg. cost / inference
0.8sp50 latency
generate · prompt panel
detect · drop zone
score breakdown · per detector
video timeline · frame scores