WorkHub Studio

One image. Every Blueprint.

AI WorkHub ships as one universal image. What a customer gets is decided by a Blueprint — modules, capabilities, workflows, pages, connections, policy — and nothing else. The Studio is where that Blueprint is composed, validated against the image, and provisioned into the artefact a deployment consumes.


What the Studio does

[01]

Compose the Blueprint

AUTHORED HERE

Choose the modules the customer licensed, the capabilities they should reach, the workflows that run, the pages people see, the adapters behind every port, and the policy gates that add to the platform floor. Every option comes from the image's own catalog.

[02]

Validate against the image

ONE IMAGE

The universal image validates the document the way it will at boot: unknown capabilities, unlicensed ones, unconfigured ports and malformed guards are workshop-time errors, not 3am ones. The image identity is a content hash; a fork would show.

[03]

Provision the instance

SPEC EMITTED

Provisioning publishes a Blueprint revision and writes an aiden-deployment-spec/1 artefact: the instance id, the image and schema identity, the adapters selected, and the secrets the deployment must supply by name. A pipeline takes it from there.

The no-fork invariant

Every instance on a train version runs a byte-identical build artifact. The only things that vary per customer are the Blueprint, the secrets, and the data. The Studio never scaffolds, builds or forks anything per customer — if a requirement cannot be met by a Blueprint field, it is a new capability in the platform, never a branch here.