Set a goal, reference the affected entity or work item, attach permitted context, and see the plan that supports it. A goal is more useful when its owner, priority, progress, and dependencies remain visible.
Context-aware AI coordination
Give AI a business objective. Keep the human decision.
Spine Twin is designed around an intention grounded in work: a customer renewal, an operational exception, a delivery commitment, or a decision requiring evidence.
Goals attached to real context
Specialized assistance, one human interaction
Context assembly, schema understanding, triage, domain analysis, and execution orchestration can contribute to the same work. Users should not need to operate a separate agent chat for every internal function.
Trace the work
Inspect plan steps, status changes, delegated responsibilities, outputs, pending approvals, and execution receipts. The plan and its evidence should be separate from confident-sounding narrative.
The authority ceiling stays in place
The Twin may prepare and coordinate within its permission envelope; it cannot create a new authorization merely by delegating a step to another agent.
Continue with context
Move from understanding to the right next step.
Historical recall does not grant authority. Scoped access and explicit approval still apply.
Capability availability varies by tenant, host, and profile. Marketing describes design intent unless a page states verified evidence.