You upload a product shot, a model in a linen dress against a soft-focus background. The file is hefty, a few megabytes of crisp detail in the weave of the fabric, the subtle gradient of the sky. A moment later, a version appears that is, according to the dashboard, 95% smaller. You lean in, squinting at the screen. The dress still looks like linen. The sky still fades. The page, now unburdened, loads in a blink. This is the central promise of SpeedSize: that you can have the visual fidelity of a photoshoot and the file size of a thumbnail, and that the difference between the two is everything the human eye was never going to see anyway.
Founded in 2019 and based in Netanya, Israel, SpeedSize sells an AI-powered media CDN built on what it calls "psychovisual" compression. The product sits between an e-commerce brand's media library and its website visitors, dynamically analyzing, resizing, and reformatting every image and video. It promises reductions of up to 99% while retaining what it defines as 100% of the perceptible quality. For customers, predominantly fashion and apparel brands, the calculus is straightforward: faster pages convert better, and smaller files cost less to store and deliver. The company has reportedly grown to around 200 customers and an estimated $6 million in annual recurring revenue.
The Neuroscience of the Unseen Pixel
The technical wedge is a proprietary method that applies AI models informed by neuroscience. The system doesn't just compress an image; it attempts to model human visual perception to identify and strip away "unseen pixels", data that the brain cannot perceive under normal viewing conditions. This differs from traditional compression, which may introduce artifacts like blurring or banding in pursuit of smaller files. SpeedSize's claim is that its method preserves "photoshoot-quality" visuals at a fraction of the size. The product then delivers this optimized media adaptively, choosing the ideal format and dimensions for a user's specific device, browser, and screen.
Landing in the Fashion Brand's Stack
SpeedSize's early traction maps directly onto a high-intent segment: online fashion retailers. These stores live and die by visual merchandising. Their images are high-resolution, numerous, and non-negotiable. Yet, page speed is a critical conversion factor. SpeedSize inserts itself as a solution to that tension. Integration is pitched as no-code, with usage-based pricing and a free testing period. The average customer is reported to pay in the "lower end five figures a year," with some mid-market deals reaching six-figure annual contracts. This positions it not as a trivial utility, but as a performance-critical line item with a clear ROI tied to site speed and bandwidth savings.
The Competitive Frame
SpeedSize does not operate in a green field. It faces established players like Cloudinary and imgix, which offer broad-based image and video management APIs, and specialists like ImageKit and Optimole. Its differentiation rests almost entirely on the claimed superiority of its psychovisual AI compression, a patented technology it says is "driven by neuroscience."
| Metric | Value |
|---|---|
| 2024 Revenue | $3M |
| 2025 Revenue | $6M |