Engineering first: measurable, regression-tested and rollback-ready. Capabilities that apply across multiple stacks are described here as shared foundations for the three core stacks.
Agent orchestration foundation
Shared by the agent stack, the robot product, the Business Agent and the Freight Agent. Roles, workflows, Skills/Tools governance, model routing and approval flow are configured once and reused across products.
Drawing analysis foundation
Text, vector primitives, raster images, and tables are weakly aligned in one project space; cross-page references use frame IDs and layout trees to avoid plausible but wrong cross-document bleed.
Edge and cloud-device deployment
Vision models export to ONNX / TensorRT. Edge and center sync metrics and sample fingerprints only; raw images stay on premise by default. The same deployment discipline is applied to robot and platform products.
Governance and security
Private knowledge bases can run fully offline. When public LLM calls are required, we support field-level data masking and a configurable "zero egress" mode with a local small model as fallback.
