Idomoo introduces Strata – the inaugural AI foundation model designed for layered video.

      The Israeli video personalization company is introducing Strata, a foundation model that claims to produce distinct, editable layers for text, animation, footage, and actors, as opposed to a single flat file. This directly challenges the architectural constraints of diffusion-based video generators.

      Currently, every AI video model available generates a flat file. While you can view, share, and trim it, if you wish to modify the font, adjust an animation, or change the background, you have to start over.

      This limitation has kept AI video from fully integrating into professional production workflows, where videos are traditionally constructed in layers, featuring separate tracks for text, motion graphics, footage, and audio that can be independently adjusted right up to the final render.

      Today, Idomoo, the Israeli enterprise video personalization company, has announced Strata, a generative AI foundation model it claims is the first designed specifically for layered video output. Rather than creating pixels, as co-founder and CTO Danny Kalish describes, Strata generates structure: independent layers encompassing typography, animation, motion paths, and synchronized audio, all integrated into what the company refers to as a “production-ready video blueprint.”

      The distinction is architectural. Typical diffusion models merge everything into a single tensor during production, with the spatial and temporal relationships embedded within the pixels.

      Idomoo asserts that Strata addresses a different computational challenge: it designs the entire composition, defining placement, contrast, movement, timing, and pacing across all layers at once while also adhering to brand guidelines. The final output is editable at the layer level, similar to how a professional would operate in Adobe After Effects.

      Strata is part of Lucas, Idomoo’s AI video agent, which operates on top of the company's existing Next Generation Video Platform. One of its more specialized features is brand awareness: Lucas analyzes a company's approved content to extract what Idomoo calls Brand DNA, encompassing design, narrative, and assets.

      Strata then applies this specification to every video generated through the platform, ensuring typography, motion cues, color values, and tone of voice are maintained across outputs at scale. The goal is to eliminate the template workaround that most "AI wrapper" products currently rely on, where generated footage is limited to fitting predetermined layouts.

      Idomoo argues that forcing content into templates results in a noticeable visual compromise; instead, Strata creates custom blueprints for each video.

      Personalization is the other key aspect. Since Strata’s output is layered and not flat, individual data fields such as names, account details, transaction histories, and product images can be incorporated into specific layers of the video composition in real time.

      This principle is central to Idomoo’s existing business model: the company currently serves clients like JPMorgan Chase, Verizon, and American Airlines, producing personalized videos at scale for customer communications, onboarding, and marketing.

      Theoretically, Strata enhances that personalization significantly because it functions at the composition level rather than merely as an overlay on a pre-rendered clip.

      The company is being cautious about the launch's extent. An early access version is undergoing testing with several of its major customers. It is available now through the Lucas AI Video Agent, but Idomoo has not revealed which customers are part of the early access group, the benchmarks the model has been evaluated against, or how it measures up to standard diffusion models on quality metrics.

      Idomoo’s claim of being the “first foundation model purpose-built for layered video” is their own assertion and has not been independently verified. The technology behind Strata is currently patent pending.

      It is notable that Idomoo’s platform documentation previously indicated that it employed off-the-shelf foundation AI models rather than proprietary ones. With Strata, the company is shifting its positioning: from being a user of AI in video to developing foundational AI for video.

      Whether the underlying architecture fulfills this promise will become clearer as enterprise customers transition from early access to production setups.

      Founded in 2007 by Yaron Kalish, Danny Kalish, and Assaf Fogel, the Ra’anana-based firm has secured a total funding of $27 million, which includes a $9 million Series A in 2013 and an $18 million Series B in 2019. It has been developing its personalized video platform for enterprise clients for nearly two decades, which provides it with a unique amount of structured video production data for training.