Our lab explores hard problems where AI must interact with evidence, authority, constraints and real-world consequences.
A process-based approach to authorship integrity that records the evolution of writing and produces a portable, tamper-evident record.
A reusable architecture for answering customer and employee questions across invoices, payments, shipments, cases and other current records—with query validation and audit logging.
An authorization layer that evaluates whether an agent remains properly scoped, authorized, auditable and safe before a consequential action proceeds.
An architecture combining structured domain knowledge, document interpretation, simulation and deterministic validation for complex logistics work.