MSBAI has officially confirmed the news pf securing a Direct-to-Phase II Small Business Innovation Research (SBIR) contract to mature OrbitGuard, which happens to be the company’s hybrid-intelligence copilot for space domain awareness.
For better understanding, this particular funding opportunity came 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).
More on the same would reveal how, valued at $1.2 million over 18 months, the stated award should now be able to propel OrbitGuard towards operational deployment, and therefore, address the exponential surge in orbital traffic projected to exceed 17,000 active satellites by 2026.
Talk about MSBAI’s OrbitGuard on a slightly deeper level, we begin from how it arrives on the scene as integrated with DoD enterprise data services like the Unified Data Library and running on MSBAI’s GURU platform (the company’s hybrid-intelligence platform which accelerates high-stakes engineering and operations for defense and industry). As a result, it can detect anomalous satellite behaviors in near-real time and identify maneuvers with exceptional precision.
Markedly enough, taking a departure from a single generative AI, OrbitGuard also leverages the prowess of a hierarchical neuro-symbolic architecture which, on its part, 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.
Such a mechanism, like you can guess, makes it possible for operators to move from alert investigation to visualization in a matter of few seconds. This they can do while simultaneously delivering reliability by design, not by prompt.
In fact, peer-reviewed evaluations, conducted so far, have also went on to showcase 94-98% accuracy in anomaly detection and classification across 15,000 on-orbit objects, ensuring robust performance in dynamic, mission-critical environments.
Anyway, turning our attention what all MSBAI’s new contract will unlock for the company, it will facilitate, for starters, the scaling of its processing pipeline to handle over 20,000 resident space objects with ~two-minute end-to-end latency and ≥99.9% uptime.
Next up, it will effectively 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.
Another detail worth a mention relates to strengthening of explainability and safety through blackboard-style logging, symbolic constraint checks, and audit-ready artifacts to facilitate Authority to Operate (ATO) certification.
Hold on, we still have a couple of bits left to unpack, considering we haven’t yet touched upon potential for seamless integration across DoD systems via secure APIs, standardized data formats, and cross-domain workflows (Air Force, Army, Navy).
In case that wasn’t enough, the contract will also come in handy to 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.
Among other things, it ought to be acknowledged that this particular effort delivers a rather interesting follow-up to MSBAI’s 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 would blend symbolic constraints, learned world models, and reinforcement-learning planners to make every alert accommodate an audit trail and rule checks before reaching the operator.
“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.”
