Device Model
Rather than architecting everything around a singular developer or operator machine, the system uses a scalable, fleet-managed node structure with cloud orchestration.
Fleet Model Components
- QiLabs: The canonical local workstation root (
C:/QiLabs/) on physical devices. - Device Node: An enrolled machine in the fleet.
- Local Agent: The installed thin control service, reporting health and executing assigned jobs locally.
- Local Runtime: The machine-local execution layer for OCR, extraction, file chunking, and vectors.
- Watch Assignment: A configuration assigned from the hosted admin instructing a particular node to monitor specific directories or pipelines.
- Drop Zone: Specified ingress pathways (e.g. specialized folders) defined and assigned by the hosted admin.
- Primary Dev Node: A device node designated specifically for primary operator deployment, configuration, and structural adjustments.
- Ingest Edge Node: A specialized device node strictly responsible for watching paths/drop zones, reading local files, and ingesting them into the archive pipeline without administrative overhead.
- Infrastructure Node: A user-controlled server (e.g.,
infra.backup_server) providing local AI runtimes, secondary data services, and public application hosting under governed control. - Fleet Control Plane: The hosted Admin platform authorizing and orchestrating action across the fleet of device nodes.
- Trust/Auth Boundary: Access control and credentials managed securely separating local worker execution from cross-tenant operations.