Recreate lost PSDs: AI that extracts layered gradients/styles from screenshots

Many design teams, agencies and SMBs regularly inherit screenshots or flattened legacy assets that take hours to recreate as editable PSDs, costing time and slowing UI modernization—an addressable market of roughly 2 million businesses at an assumed $3,000 ACV (~$6.0B) suggests a large recurring demand. This problem is particularly acute for teams consolidating design systems or trying to reuse vintage gradients and textures, and while the market score is 95/100 and revenue potential 88/100, competition is at a medium level and technical complexity is the principal barrier to entry. A practical product would be an AI-powered service and plugin suite that ingests screenshots and outputs layered, named PSDs and Figma files with reconstructed gradients, blend modes, masks, vector primitives and replicated textures, plus confidence scores and a human-in-the-loop editor for refinement. The offering should include a web app, Figma/Photoshop plugins, an API for automation, and an on-prem/private-cloud option to meet enterprise privacy needs. Under the hood this combines layer decomposition, style extraction, generative fill for occluded geometry, and a curated training set of historical UI examples to capture retro/skeuomorphic detail. The timing is favorable because recent advances in image understanding and style transfer make layer-level reconstruction increasingly feasible, organizations are actively consolidating design systems, and nostalgia for retro UI styles creates a niche desire to preserve original assets; together these trends support the $6B market and the 88/100 revenue outlook. To stand out you must deliver measurable fidelity (target ≥80% automated correctness on common patterns), tight integrations into existing design workflows, enterprise-grade privacy controls, and a clear IP/legal policy—these are realistic differentiators but also the core challenges that will determine if the idea can scale.