Spherical Defense provides API security through unsupervised deep learning, creating real-time positive security models that adapt to evolving application traffic without the need for predefined rules. This technology protects APIs from malicious traffic and misuse while ensuring data remains within the user's network, eliminating false positives and minimizing performance impact.
Funding
Funding not disclosed
Founders
Product
Problem
Traditional Web Application Firewalls (WAFs) and first-generation API security tools rely on manually defined rules and signatures, which are difficult to maintain and often generate false positives. These systems struggle to adapt to evolving application traffic and new attack vectors, leaving APIs vulnerable to misuse and malicious traffic.
Solution
Spherical Defense provides API security through unsupervised deep learning, automatically building real-time positive security models that adapt to evolving application traffic without predefined rules or signatures. The technology protects APIs from malicious traffic and misuse by learning normal application behavior and identifying deviations indicative of attacks. By creating a dynamic model of expected API behavior, Spherical Defense minimizes false positives and ensures data remains within the user's network. The solution integrates within existing infrastructure, whether on-premise or in a private cloud, and operates without requiring third-party access to data.
Target Audience
The primary users are organizations that require robust API security, including those with internal networks, service meshes, and applications deployed on AWS.
Features
- Unsupervised deep learning to build a positive security model of API traffic in real-time
- Automatic adaptation to application development and changing user behavior
- Rapid deployment without the need for rule or signature creation
- Easy integration with existing on-premise or private cloud infrastructure
- Secure and confidential operation, ensuring data stays within the user's network
- Unattended learning, dynamically building models without user intervention
- Transparent operation with little to no performance degradation
- Session-level analysis, holistically monitoring sequences of interactions
- Tree-based data analysis for complex JSON and XML objects, minimizing false positives