Loading SaaS profile...
As of 2026
AI & Automation
A real-time, instruction-guided video editing model that enables targeted transformations of live or recorded video streams with near-zero latency.
Overview
Decart AI's Lucy Edit is a flagship real-time video-to-video editing model that allows users to perform targeted, instruction-based modifications to video streams. By using free-text prompts, users can swap characters, change outfits, replace objects, or transform entire scenes in real time at 30fps. It is designed to preserve the motion, lighting, and spatial structure of the original input, ensuring high temporal consistency and identity preservation without the need for manual post-production or complex masking.
Founded year: 2025
Founder: Dr. Dean Leitersdorf, Moshe Shalev
Team size: 100-200
Popularity: Frontier model in real-time generative video infrastructure
HQ: Tel Aviv, Israel
Status: Active
Funding status: Series B
Revenue source: API Usage Fees
Customer type: B2B
Funding: $450M+
Pricing: Usage-based
Tech stack: Diffusion-based video models, WebRTC, Proprietary Low-Latency Inference
Platform: Web • API
Integrations: fal.ai API, each::labs API, WebRTC streaming
Founder story
Decart AI was founded by Dr. Dean Leitersdorf and Moshe Shalev in Tel Aviv, Israel, to build the infrastructure for the next generation of real-time generative intelligence. Emerging from stealth in 2025 with $21 million in seed funding, the company has rapidly raised over $450 million to push the boundaries of low-latency video inference and world models.
What it does
- Performs instruction-guided, real-time video transformations via free-text prompts
- Supports targeted editing including character/outfit changes, object replacement, and scene style swaps
- Maintains original motion, scene composition, and lighting stability across frames
- Enables live virtual try-on for fashion, commerce, and product visualization
- Operates at 30fps with near-zero latency for live-streaming and interactive experiences
Who it's for
- Live streamers and content creators
- Commerce and retail brands for live shopping
- Advertising and product placement agencies
- Real estate and architectural visualization teams
Why it works
- Achieves near-zero latency, enabling live-streamed, interactive transformations
- Uses advanced identity preservation to ensure the original subject's motion and key features remain intact
- Requires no manual masking or extensive post-production, drastically reducing editing time
- Supports long-form, infinite-length editing without identity collapse or quality drift
- Offers high-fidelity, production-ready output compatible with modern 720p/1080p standards
Growth strategies
- Expanding API- first access through platforms like fal.ai and each::labs for broad developer reach
- Focusing on high- growth commercial use cases like live virtual try-ons for retail
- Building specialized variants like Lucy 2.1- VTON for commerce-specific visual tasks
- Providing scalable infrastructure to support low- latency requirements for edge-computing and interactive media
Alternatives
Comparison overview
- Lucy Edit differentiates itself from standard generative video models (like Runway or Luma) by focusing specifically on low-latency, real-time video-to-video editing
- While other models are optimized for prompt-to-video generation, Lucy is built to edit existing live streams, maintaining temporal consistency and frame-by-frame structural integrity.