# Multilingual content integration plan — 12 September 2026

The full plan and editable trackers are in [the integration package](C:/Git/content/chatgpt/speech_multilingual_integration_2026-09-12/START_HERE.md).

Use shared meanings with independently authored native language packs; assess concept selection in mixed contexts separately from familiar exercise execution. Plan covers English, Hindi, Tamil, Telugu, Kannada, Malayalam, Marathi and Bengali, with release order unassigned. The proposed first technical pilot is English/Hindi, matching the enabled live profiles.

The package includes 153 provisional unit mappings, 2,048 target-language applicability rows, 41 dependent implementation tasks, proposed immutable case/trial contracts, native review and media workflows, and 18 core acceptance scenarios plus 16 four-role scenarios. Nothing has been imported or deployed. The live app was verified at https://speech.neurapy.ai; the requested spelling speech.nerapy.ai was not verified.

First reconcile the deployed revision with local HEAD bacf95cf: Native Review exists live but was not found in inspected local routes. Verify local grammar fallback, cross-language progress aggregation, readiness gates, surface-form assembly and locked-version import behaviour against the actual deployed baseline. Reuse existing concepts, constructions, texts, activities, AI Studio and review UI. Keep tenant personal-world content packs separate from global curriculum releases.

See [roadmap](C:/Git/content/chatgpt/speech_multilingual_integration_2026-09-12/IMPLEMENTATION_ROADMAP.md) and [current-state evidence](C:/Git/content/chatgpt/speech_multilingual_integration_2026-09-12/evidence/CURRENT_STATE.md). This is documentation only; app tests and deployment checks are planned, not run.

## Four-view amendment

[Role workflows](C:/Git/content/chatgpt/speech_multilingual_integration_2026-09-12/roles/ROLE_WORKFLOWS.md) explicitly cover admin, practitioner, caregiver and restricted child play, with separate content/clinical/home/media/report decisions. The existing patient-portal caregiver context is not assumed to provide separate guardian identity. Home summaries must retain family-reported provenance and must not invent individual question outcomes. T29–T38 are now first-pilot gates. Live practitioner navigation returned 403 Active organization required; full live role verification remains pending with demo identities. App code and existing app-review document were not changed.

## Latest live staff session

The user opened an organisation session with role sph_admin. Dashboard and patient detail are accessible; the actual Speech Plan link returns missing sp_child_manage. The previous no-org limitation is historical. Live dashboard still leads with games/credits/licences and a /vt prescription shortcut; local parent Dashboard is similarly game-focused. [Foundation-first review](C:/Git/content/chatgpt/speech_multilingual_integration_2026-09-12/roles/PRACTITIONER_LIVE_REVIEW.md) adds T39–T41 for practitioner, patient and parent experiences. T29 now requires matching effective permissions across UI, controller and middleware; central ACL state remains unverified and no grants/sync were performed.

## Content-production handoff

[Content factory](C:/Git/content/chatgpt/content_factory/README.md) now owns language/node authoring folders and media/game jobs. Content model 0.2.0-draft is separate from the unchanged app node graph. App integration must consume approved revisions through the planned adapter, not treat generated drafts or the offline review game as published clinical content.