Predictive Equations

Predictive Equations (also referred to as Predictive AI) develops machine vision and deep learning models that restore and analyze degraded image and video data. Its enhancement models cover upscaling, deblurring, denoising, relighting, color naturalization and JPEG artifact removal, with additional models listed as in development for underwater clarity and environmental denoising of snow, fog, shadow and glare. The company states its upscaling supports magnification factors beyond the common x2-x8 market standard, describing outputs up to 8K video and 12K images, and in one product page up to x16-x32 and 20K resolution.

Alongside enhancement, the company offers an analysis layer trained toward forensics and scene comprehension. Current capabilities listed include object detection, image and video segmentation, pose estimation, region-of-interest extraction, feature matching for image comparison, object extraction, intelligent zoom, file compression and media size optimization. Facial recognition, optical text recognition and automated text extraction are listed as planned, with behavioral analysis and in-video/in-image clip retrieval described as in research and development. Demonstrations include extraction and upscaling of tiles from a 50,000 x 50,000 pixel Airbus Pleiades satellite image of Marseille, France.

The company's technology originated as legal-oriented software and it states it has been used in courts and by startups, and that it is the first AI company to have gone through a multi-year judicial process from investigation to verdict. Stated target industries include legal and forensics, space and aviation, security, e-commerce, operations, healthcare and digital content. Additional named products include Enterprise Solutions, Predictive Asset Protection, Signalforge and Gamerforge.

Founding story

Predictive Equations grew out of a scholastic online community centered on martial arts, networking and technology, whose members sought to train remotely and applied their skills to building what the company describes as ethical technology, with member-submitted content used to build and train its models. CEO Alexander Borschel designed the AI system, and criminal defense attorney Nurit Maatok co-developed the system for evidentiary enhancement and analysis.

Business model

The company sells access to a Predictive AI platform available via cloud, API and local or on-premise installation, with a desktop application and a team-oriented platform. Business-tier features include shared team workspaces, synchronization for collaborative editing, and a dashboard covering analytics, usage trends and cost controls for AI inference. Free trials and a beta signup are used as entry points, and certain tools are stated to be available only on team-based plans. The company states model improvements are released automatically at no additional cost.

Traction

The company states its technology is already used in courts and by startups and that it has contributed to a defendant's early release. It describes surviving a multi-year judicial process from investigation through verdict. The Predictive AI Platform, spanning cloud and local install, is stated to have been released in 2024; several analysis and enhancement models remain in beta, planned release, or R&D.