Workspace B · commercial plant
Home is Workspace A. Workbench B is the SaaS surface: one environment, many domains. The customer does not buy an AI. They buy a system that takes an action, measures error against a reference, and uses that experience to choose the next action.
Diagnostic validation and the audit desk live on this workbench, not as separate products. Music remains the first vertical. Intelligence is Δe, not model prose.
HOME A Goal / Task / namer of y │ ▼ WORKBENCH B Environment · SaaS plant Run | Diagnostic | Audit | Experience │ ▼ measured z → error e → Self() next move → repeat Organization → Workspace A → Workbench B → Run → Experience → Marketplace Metric: Δe per Run
A Run records
organization, workspace, workbench, objective, actions, measurements, error reduction, experience, outcomes, permissions, usage/billing.
Hard rule
Core loop = protected. Commercialization = additive. Do not commercialize the AI. Commercialize the closed-loop outcome. Build the intelligence once; change the environment.
Workbench B is the environment. A Run is how a customer operates it.
No Run yet. Pick a pack above.
Product: Aethel Node Outcome Engine. First Workbench: Node. Safety stays an evidence-gate demonstration, not a diagnosis. Marketplace is an Experience Marketplace: validated Δe becomes a reusable asset.
Acceptance: Home A and Workbench B still work. A user can start one Run. The Run records y → z → e. Self() names the next action. Diagnostic and Audit open from this plant. No PHI. No unsupervised clinical decision.