PSISTAR LTD

Two model families built in-house: (1) a neuro-symbolic reasoning engine that converts engineering documentation into a queryable operational knowledge graph with full source traceability; (2) a physics-informed forecasting model - Horizon, a model that predicts system state from sensor telemetry, with a vision for zero-shot deployment. The two engines combine into a reliability and optimization layer that runs inside existing control systems: real-time, on-premises, air-gapped, fully auditable. LLMs are restricted to the interface layer; no probabilistic component sits in the decision path. This is the architecture we consider credible for regulated, safety-critical environments where statistical ML and predictive-maintenance approaches fail audit.

Tier 1 (beachhead): defense primes and integrators, critical energy operators (power generation, transmission, offshore O&G), and hyperscale data center operators with air-gap and audit mandates. Tier 2 expansion: pharma, specialty chemicals, nuclear, utilities. Tier 3: metals, mining, pulp and paper. Buyers: heads of operations, reliability, and engineering.

Yochai Pagi - CEO & Founder

Matan Pagi - CTO & Co-Founder