Quickly customize a vision model without a dataset to start. Just describe your inputs and a dataset will be generated on the fly, model trained in a few minutes. We’ve started with classifiers but planning to expand to other tasks and APIs to fine-tune on your data.