MSBAI announced it has secured a Direct-to-Phase II Small Business Innovation Research (SBIR) contract to mature OrbitGuard, its hybrid-intelligence copilot for space domain awareness. The opportunity originated from the Department of Defense Chief Digital and Artificial Intelligence Office (CDAO), was competitively selected by the Office of the Secretary of Defense (OSD), and has been executed by the Air Force Digital Transformation Office (DTO). Valued at $1.2 million over 18 months, the award will propel OrbitGuard toward operational deployment, addressing the exponential surge in orbital traffic projected to exceed 17,000 active satellites by 2026.

OrbitGuard transforms the space data deluge into decision dominance. Integrated with DoD enterprise data services like the Unified Data Library and running on MSBAI's GURU platform, it detects anomalous satellite behaviors in near-real time and identifies maneuvers with exceptional precision. Rather than relying on a single generative AI, OrbitGuard employs a hierarchical neuro-symbolic architecture that fuses multi-source orbital data including infrared, CelesTrak, and electro-optical feeds with symbolic checks, JEPA-based learned world models, and reinforcement-learning planning agents. 

This enables operators to move from alert → investigation → visualization in seconds, while delivering reliability by design, not by prompt. Peer-reviewed evaluations demonstrate 94-98% accuracy in anomaly detection and classification across ~15,000 on-orbit objects, ensuring robust performance in dynamic, mission-critical environments.

"Space operators are overwhelmed by the volume and velocity of orbital activity in a contested domain," said Allan Grosvenor, MSBAI founder & CEO. "OrbitGuard isn't 'just generative AI.' It's a hybrid-intelligence copilot, combining symbolic rules for guaranteed precision, JEPA-based world models for predictive understanding, and multi-agent reinforcement learning for adaptive planning to compress timelines from detection to decision, empowering Guardians to act with confidence and maintain superiority."

Under this contract, MSBAI will:

• Scale the processing pipeline to handle over 20,000 resident space objects with ~two-minute end-to-end latency and ≥99.9% uptime.
• Advance AI models using Graph-JEPA and Patch Time-Series Transformers, introducing multi-agent reinforcement learning for threat assessment and predictive analytics, targeting ≥95% operational accuracy.
• Strengthen explainability and safety through blackboard-style logging, symbolic constraint checks, and audit-ready artifacts to facilitate Authority to Operate (ATO) certification.
• Ensure seamless integration across DoD systems via secure APIs, standardized data formats, and cross-domain workflows (Air Force, Army, Navy).
• Deliver an ATO-ready prototype incorporating NIST SP 800-53 controls and comprehensive security documentation, with support from academic partner Dr. Faisal Kaleem at Metro State University.

The effort builds on MSBAI's lab-validated results (≈94-96% maneuver-detection accuracy) and recent demonstrations, including a live showcase at the U.S. Space Systems Command's Apollo Project Demo Day at the SDA TAP Lab in Colorado Springs on August 13, 2025, where OrbitGuard demonstrated autonomous integration with KeepTrack for orbital tracking, analysis, and visualization. Under the hood, GURU's hierarchical agent society blends symbolic constraints, learned world models, and reinforcement-learning planners—ensuring every alert includes an audit trail and rule checks before reaching the operator. A program kickoff is scheduled for September 2, 2025, with DoD stakeholders, positioning OrbitGuard for dual-use in commercial space traffic management.

Click here to learn more about MSBAI's Technology