
Turn regular mobile photos into professional product visuals
Turn this teardown into a decision-ready prompt for ChatGPT, Claude, or your agent.
If you only have a few minutes to spare, here’s what investors, operators, and founders should know about ClipDrop (W21).
ClipDrop began with a striking promise: point a phone at an object, cut it out with computer vision, and move it onto a desktop in one gesture. Paris-based Init ML turned that viral 2020 prototype into a broad image-editing suite spanning background removal, cleanup, relighting, upscaling, uncropping, and image generation. Stability AI acquired Init ML in March 2023 after reporting that ClipDrop had reached more than 15 million users.[1]
This was not a shutdown story. It was a two-step transfer of a useful product and team. Stability bought an application layer for its image models, then sold Init ML to Jasper in February 2024 as it narrowed its focus to open models. Jasper kept the team, standalone product, and API path. Both prices remain undisclosed. ClipDrop still operates under Jasper, although its developer path now points customers toward Jasper's image API.[2][3]
Init ML was founded in Paris in July 2020 by Cyril Diagne, Jonathan Blanchet, and Damien Henry. Diagne became CEO, Blanchet CTO, and Henry COO. All three had worked at Google or Google Arts & Culture, giving the founding team experience at the intersection of machine learning, product design, and visual culture. Y Combinator accepted ClipDrop into its Winter 2021 batch and lists the company as acquired.[4]
The trigger was unusually concrete. In March 2020, Diagne published an “AR Cut Paste” demonstration in which a phone camera isolated a physical object and transferred it to a computer. The clip reportedly drew five million views within days. The interaction compressed an 11-step phone-to-desktop workflow into two steps, turning image segmentation from a technical capability into a legible product gesture.[5] Diagne later wrote that when he published the prototype, “the one user it addressed was… myself.”[10] That origin helps explain why the demo felt unusually direct: it solved a workflow its maker already understood.
That spectacle contained a practical wedge. Marketplace sellers and e-commerce operators repeatedly need isolated product images, clean backgrounds, and usable packshots. ClipDrop made those jobs accessible without requiring Photoshop fluency. The company raised a €1.2 million seed round in 2021, led by Air Street Capital, to add research and product staff.[5]
The public record does not establish how the three founders decided to start the company together beyond their shared Google backgrounds. It does show that the product thesis survived both sales. When Jasper bought ClipDrop, Henry said the companies shared “a clear focus on creating the best multimodal platform for marketers.”[2] The quote describes the endpoint more clearly than the missing founding account: a self-made interaction prototype became infrastructure for enterprise marketing production.
The original product joined phone capture, object segmentation, and desktop transfer. A user pointed a camera at an object, let the software identify its boundary, and pasted the isolated object into a desktop document. The novelty was not background removal alone. ClipDrop packaged several actions into one visible gesture, making a difficult edit feel like moving a physical item.
Read the complete post-mortem, the rebuild playbook, and the exact reasons ClipDrop is still worth studying now.