Self-correcting evaluation
Automated loops that detect visual flaws, correct them, and re-evaluate outputs until they meet defined quality criteria.
Research
Our research teams build systems that evaluate, correct, and benchmark image and video AI models across real-world e-commerce use cases.
E-commerce benchmarks
Purpose-built benchmarks for fashion, product catalogues, moodboards, and other commerce workflows.
Taste systems
Research measuring visual taste, aesthetics, and brand fit in generated imagery.
Latest research
PublicationOct 2, 2026
Video, made for 1
Video moves from something made once and sent to many, to something directed once and rendered for each person, often while they watch.
Read the post →
PublicationSep 7, 202628 min read
Building a Self-Improving Visual Evaluation Harness
How we built a closed-loop agent that can judge, repair, re-verify, and learn from failures in e-commerce image generation
Read the post →
