Mihawk delivers spatial AI that combines real‑time perception with long‑term temporal reasoning. By fusing feeds from multiple cameras, it creates a persistent, time‑indexed 3D map of environments, tracking geometry, people and objects across views and occlusions. This structured memory lets downstream agents query past events, test hypotheses, and generate alerts that rely on both immediate visual data and historical context, enabling more accurate situational awareness and decision‑making.
Funding
Funding not disclosed
Founders
Product
Problem
Current computer‑vision systems either provide instant perception without context or maintain static records without real‑time insight, limiting the ability to understand dynamic environments over time.
Solution
Mihawk delivers a Spatial AI platform that fuses real‑time perception with temporal reasoning to create a persistent, time‑indexed 3D map from multiple camera feeds. The system aligns cross‑camera geometry to reconstruct a shared 3D scene, then continuously recognizes and tracks people and objects, building a searchable history of locations, movements, and interactions. Reasoning agents can query this world model to generate hypotheses, reconstruct scenarios, and validate counterfactuals against recorded evidence. This combination enables immediate alerts based on current observations while preserving long‑term identity and event context for deeper analysis.
Target Audience
Primary customers are enterprises that require continuous situational awareness, such as security and surveillance operators, smart‑city infrastructure managers, and industrial or retail analytics platforms.
Features
- Multi‑camera integration that produces a unified, persistent 3D map with cross‑camera alignment and scene reconstruction
- Real‑time detection and tracking of people and objects, generating timestamped event histories and movement paths
- Spatial relationship indexing that records where entities were relative to each other and to the environment
- Queryable world model allowing scenario reconstruction, hypothesis testing, and counterfactual validation under physical constraints
- API for downstream applications to retrieve spatial queries, identity timelines, and alert triggers