YC-Backed Cardboard Launches AI-Powered Video Editor That Works Entirely in Browser

Summary

Cardboard, a Y Combinator Winter 2026 startup founded by Saksham and Ishan, has launched an agentic video editor that transforms raw footage into polished videos through natural language commands. The platform addresses a common pain point for content creators: hours of manual editing work that turns mountains of raw assets into usable content. Users can simply describe what they want—such as "make a 60s recap from this raw footage" or "cut this into a 20s ad"—and Cardboard generates a first draft that can be refined further.

The technical architecture is particularly notable: Cardboard runs entirely client-side in the browser using a custom hardware-accelerated renderer built on WebCodecs and WebGL2, with no server-side rendering required. Video understanding tasks leverage cloud vision-language models and traditional ML models, while third-party foundational models handle agent orchestration. The editor supports multi-track timelines with keyframe animations, shot detection, beat synchronization through percussion detection, voiceover generation, voice cloning, background removal, and multilingual captions with spatial awareness of subjects in frame.

The founders bring personal experience to the problem: Saksham creates content with approximately 250,000 views per month and found editing took longer than content creation, while Ishan produced launch videos for HackerRank and spent most of his time on cuts and sequencing rather than storytelling. Since November, the team has shipped 13 releases, adding features like Premiere Pro, DaVinci, and Final Cut Pro XML exports for workflow integration. The platform is available for trial at demo.usecardboard.com with no login required, and pricing starts at $60 per month.

Cardboard's roadmap includes real-time collaboration features ("video git") to eliminate inefficient feedback loops, and a prediction engine that learns editing patterns to suggest next actions—similar to Cursor's tab completion but for timeline operations. The founders position their tool as bringing modern developer tool paradigms to video editing, arguing that current video creation tools are stuck where developer tools were in the early 2000s: local-first with zero collaboration and slow feedback loops.

• Founded by creators who experienced the editing bottleneck firsthand, with one generating 250K+ monthly views and the other producing corporate launch videos
• Pricing starts at $60/month with exports to Premiere Pro, DaVinci, and Final Cut Pro for integration with existing professional workflows

Editorial Opinion

Cardboard represents an interesting application of AI agents to creative workflows, but its success will hinge on whether it can deliver "good enough" first drafts that actually save time versus frustrating users with iterations. The browser-based architecture is clever from a privacy and performance standpoint, though it may limit the complexity of AI models that can run locally. Most importantly, the planned "video git" collaboration features and predictive editing engine could be the real differentiators—if Cardboard can become the "GitHub for video" rather than just another AI editing assistant, it might justify its position in an increasingly crowded field of AI video tools.