Launches Synthetic Data​ Platform To Train AI Models, the leading platform provider for physics-based synthetic data, announced the availability of the Platform as a Service (PaaS) for synthetic data engineers and computer vision scientists.

The first-of-its kind platform is a complete stack for synthetic data including a developer environment, a content management system, scenario building, compute orchestration, post-processing tools, and more. These tools enable users to generate engineered datasets to overcome gaps and bias issues experienced when using real datasets to train AI. has used recent funding from Space Capital, Tectonic Ventures, Congruent Ventures, Union Labs, and Uncorrelated Ventures to deliver a first-ever publicly accessible tool that enables users to combine their own digital sensor models with 2D and 3D content to create simulated scenarios used to create unlimited amounts of synthetic data.

Along with the platform availability,, has released an open-source example application to help people get started with synthetic data. Users can build from that example or reach out to use the wide variety of additional content can provide covering common applications such as remote sensing Earth observations data based on visible wavelengths, hyperspectral imagery, and synthetic aperture radar (SAR). The company has also worked with commercial partners focused on synthetic x-ray data for security applications and with academic teams working in life sciences microscopy.

Synthetic data is rapidly emerging as a required capability to help overcome gaps, bias, and privacy issues with real world data used to train and validate AI systems. Unlike many current synthetic data systems that rely on real world source data as input, has proven that real-world datasets based on simulation of a 3D environment can be used to both increase AI performance and reduce the amount of data used to train AI.

Datasets produced by have 100 per cent accurate labeling and intentionally designed distribution of training features, characteristics which are difficult or impossible to obtain in real datasets with automated or manual labeling.

“Leading global companies have spent years building proprietary synthetic data platforms. brings that capability to the rest of the industry.,” said Nathan Kundtz, founder and CEO, “A PaaS enables our customers to incorporate synthetic data generation as an evolving capability throughout AI workflows, going beyond individual projects to generate single datasets.”

The PaaS is already in use by government, commercial, and academic research customers. The platform includes an open-source SDK for developers to build their unique sensor models, plus a range of 3D and 2D content to get customers started building and configuring their custom scenarios. provides a hosted job management capability that lets customers run nearly unlimited numbers of jobs in the AWS cloud. The platform is team-based and includes a web experience for managing the team and organisation and for enabling data scientists and computer vision engineers to configure dataset generation in a no-code interface.

Having already demonstrated the effectiveness of physics-based synthetic data in customer applications, the new PaaS enables customers to integrate the PaaS as an automated capability in their enterprise AI training and testing workflows.

With engineered datasets produced as an ongoing, tuned capability, simulated or synthetic data will be a critical enabler for faster innovation and validation of safe, effective AI applications.