Global Today Hub and one-program workspace shell model for logged-in DOL English

DEC-0095 - Global Today Hub and one-program workspace shell model for logged-in DOL English

Superseded on 2026-03-27 by DEC-0096_resume-last-program-and-selective-cross-program-aggregation-shell-model.md. This record is retained as archive for the retired Global Today default-shell model.

Context

The current logged-in product has enough capability breadth to create navigation confusion: discovery, cross-program portfolio visibility, and deep execution inside courses/practice/vocabulary all compete inside the same shell. The user explicitly requested a new direction that treats navigation as workflow control, not as a flattened sitemap.

Decision

Logged-in DOL English adopts a two-mode shell:

• Global Today Hub
  • the only cross-program execution surface,
  • used for continue flow, due today, upcoming classes/deadlines, active programs, and recommended next step,
  • not treated as a program context.
• Program Workspace
  • one real program at a time (IELTS, SAT, TOEIC, …),
  • used for deep execution inside the learner’s current program.

Program context policy:

• Program switcher lists real programs only.
• Tất cả chương trình must not behave as a peer program context inside the program switcher or deep learning shell.

Workspace destination baseline for desktop:

• Hôm nay
• Khóa học
• Luyện tập
• Từ vựng
• Tiến độ

Logged-in discovery policy:

• Khám phá does not stay in primary logged-in navigation.
• Cross-sell/discovery can surface through:
  • Global Today,
  • Thêm chương trình,
  • recommendation cards,
  • explicit discovery entry points outside the primary execution shell.

Canonical object ownership policy:

• Every deep object must have a single canonical home section.
• Minimum ownership fields:
  • program,
  • object_type,
  • source,
  • canonical_section.
• Summary cards in Global Today or other surfaces are shortcuts only; they do not create duplicate object ownership.

Default entry policy after login when no valid returnTo exists:

• no active program -> onboarding/discovery,
• one active program -> direct into that program workspace,
• multiple active programs -> Global Today Hub.
• valid returnTo still overrides default entry resolution.

Cross-program deep-link policy:

• notification/search/deep-link may switch current program only when target object belongs to another program,
• switch must be explicit in URL/state and visible in UI context cues,
• user must not feel that the app silently changed program beneath them.

Decision Value

• Removes the pseudo-program problem caused by treating All Programs like a real learning context.
• Reduces daily scan cost for retention-first learners.
• Clarifies destination boundaries between next-action orchestration and detailed learning execution.
• Creates a stable base for object trust, because the same assignment/test/vocab object keeps one owner section.

Rationale

This model matches the learner’s primary jobs-to-be-done more closely:

• first decide what to do now,
• then work inside one program,
• then open the deep object in its owner section.

It also keeps discovery available without letting discovery compete with daily execution in the logged-in shell.