VisionFleet
Where every device is, what it runs, and how far a rollout has got.
What it is
A central operations console for fleets of VisionEdgeOS appliances, smart cameras and site gateways, and for the VisionAI and VisionML software they carry. It answers four questions: where each commissioned device is assigned; its health, connectivity and open incidents; the exact operating system, application, runtime, model, manifest, recipe, firmware and driver versions and digests it reports; and how far a rollout has progressed across countries, sites, lines, stations and governed groups.
Locations are hierarchical — tenant, country, site, line, station — because that is how a fleet is actually addressed.
Who it is for
Global IT and the software teams behind the fleet: the people who are asked what is deployed where, and whether it matches what was intended.
How it is governed
Tenant context is derived on the server, not taken from a request field. Access tokens are validated against an OIDC provider through a provider-neutral connection model. The local development identity bypass is visible in the interface and cannot start outside the development environment. Live observations and seeded demonstration rows are labelled, and neither is ever treated as deployment authorization.
What it does not do
The first increment grants no production authority of any kind: no production approval, no qualification, no retention or evidence-disposition decisions, and no general remote shell. Browser sign-in, activated customer roles and review gates, identity-provider role bindings and production-qualified administrative audit are named as later work.
Status
Design and a development-only increment. Every development-control response reports that the deployment is not eligible, and the packaged development deployment says the same.
Related
VisionEdgeOS produces what the devices run and reports what they observe. VisionAI and VisionML are the software and model versions the inventory tracks.