Built on Google Cloud
LinguistPro is deliberately hybrid: local-first storage and deterministic language data establish the learning record; Google Cloud handles selected heavy workloads under the user's own key.
Built on Google Cloud
- Gemini API
- Cloud Text-to-Speech
- Cloud Translation v3
In production today
Selected cloud processing routes use these services; the browser and the local learning core have separate responsibilities.
| Google Cloud service | Where | What it does |
|---|---|---|
| Gemini API | LinguistPro | Hebrew learning tables, PDF/image OCR, audio/video transcription and A1–B2 graded retelling. Runs under a per-user API key stored in the browser; the selected source is processed under the user's Google account. |
| Cloud Text-to-Speech | LinguistPro | Niqqud-aware Hebrew audio for lines, examples and study material, with content-addressed caching. |
| Cloud Translation v3 | LinguistPro & HDLE Premium | Neural translation of Hebrew source text. The official v3 client, with budget guards and usage tracking in the desktop product. |
What stays deterministic and local
AI does not get authority over the learner's memory or Hebrew grammar. Browser-local OPFS + SQLite WASM hold the primary workspace; FSRS-6 schedules review from an append-only event trail; a shipped Pealim dataset supplies 9,279 inflection paradigms; and the morphology resolver exposes ambiguity and provenance instead of asking a language model to invent a confident answer.
Optional server layer
Account sync, Telegram/Mini App and OAuth/MCP agent connections are separate from the local workspace. Each uses explicit consent, scoped access and revocation. Connected-agent and Telegram surfaces remain an owner pilot rather than generally available product features.
Planned / in evaluation
Our second vertical — EgorGenom, rare-disease genomics — is where we expect to scale on Google Cloud next. This is explicitly forward-looking, not yet in production.
| Service | Vertical | Intended use |
|---|---|---|
| Vertex AI | EgorGenom · LinguistPro | Phenotype-to-gene scoring models; fine-tuned Hebrew morphology / translation models. |
| BigQuery | EgorGenom | Querying population-frequency datasets (gnomAD) at scale instead of static local dumps. |
| Cloud Storage | EgorGenom | De-identified genomic data and long-read archives. |
| Document AI | EgorGenom | Parsing clinical reports and lab PDFs that are currently handled manually. |
Why Google Cloud
For selected workloads, model quality for Hebrew and multilingual work: Gemini handles Hebrew and mixed RTL/LTR source material well across extraction, transcription, translation and controlled rewriting. The architecture calls cloud models for those bounded workloads while keeping the durable workspace and deterministic learning logic local. This preserves graceful degradation when a provider is unavailable and makes the cost-bearing step visible to the learner.
The privacy design
LinguistPro keeps its primary workspace in browser-local OPFS + SQLite WASM. BYOK requests are governed by the user's Google account. Optional sync and integrations can create server-side learner records or consented text replicas, so they are disclosed and controlled separately. See theprivacy page for the current data-flow summary.