docs(marketing): instructions pour l'agent Hermes et mise en conformite des textes de lancement (DeepL retire, 6 moteurs, 107 langues)

docs/marketing/launch/assets-spec.md

@@ -10,10 +10,10 @@
 | 1 | Page d'accueil / hero (drop-zone) | Landing, réseaux | `office-translator-landing-page` |
 | 2 | Upload en cours (fichier chargé, langues sélectionnées) | Démo du workflow | Idem |
 | 3 | **Résultat côte à côte** (original vs traduit, format intact) | Preuve qualité | Backend + dashboard |
-| 4 | Sélecteur de moteur (Google/DeepL/LLM) | Fonctionnalité clé | Idem |
+| 4 | Sélecteur de moteur (les 6 moteurs visibles) | Fonctionnalité clé | Idem |
 | 5 | Interface glossaire | Différenciation | Dashboard |
 | 6 | Dashboard (statistiques, quota) | Crédibilité | `/dashboard` |
-| 7 | Page pricing (4 plans) | Conversion | Landing `#pricing` |
+| 7 | Page pricing (les 5 formules) | Conversion | Landing `#pricing` |
 | 8 | Profil / historique | Confiance | Dashboard |
 
 **Outil recommandé** : Playwright (script de capture automatisée pour les états dynamiques : upload en cours, progression de traduction).

docs/marketing/launch/product-hunt.md

@@ -16,15 +16,15 @@ Google and DeepL flatten your files into plain text. We don't.
 
 Office Translator translates Excel, Word, PowerPoint **and PDF** **in place**: merged cells, formulas, fonts, borders, headers, footers, tables and slide layouts survive intact. Even scanned PDFs are handled via OCR (Mistral), and text inside images is translated via vision models.
 
-Pick the engine per document — 7 providers:
+Pick the engine per document — 6 providers:
 - **Google** (free, fast)
-- **DeepL** (best for business text)
 - **Google Cloud**
-- **LLM engines**: OpenRouter (eco & premium tiers: DeepSeek, Gemini, Claude), OpenAI, Grok by xAI — for complex, context-aware translations
+- **OpenRouter** — eco & premium LLM tiers (DeepSeek, Gemini, Claude)
+- **OpenAI** and **Grok by xAI** — for complex, context-aware translations
 
 Plus:
 - **Custom glossaries** — lock your HVAC, legal or medical terminology across every document
-- **60+ languages**
+- **107 languages**
 - **Private by design** — uploads deleted after 30 min, results auto-deleted within 2 h, never used to train models
 - **API** on the Business plan (10,000 calls/mo) for your own tooling
 
@@ -33,17 +33,18 @@ Pricing: Free (2 docs/mo) · Starter €9 · Pro €19 · Business €49 · Ente
 ## Maker comment (à publier par le fondateur, 9 h EST)
 > We built Office Translator because every other tool broke our layouts.
 >
-> The hard part wasn't translation — it was **re-integrating** the text. Our pipeline parses the document structure (cells, runs, shapes), extracts only translatable content, translates it (your choice of 7 engines), and writes it back into the **same** XML structure. Formulas, merges, fonts, borders: untouched.
+> The hard part wasn't translation — it was **re-integrating** the text. Our pipeline parses the document structure (cells, runs, shapes), extracts only translatable content, translates it (your choice of 6 engines), and writes it back into the **same** XML structure. Formulas, merges, fonts, borders: untouched.
 >
 > What I'd love your feedback on:
-> 1. Which engine do you trust most for business documents — DeepL or an LLM?
+> 1. Which engine do you trust most for business documents — a classic engine like Google, or an LLM?
 > 2. Would you use the API (Business plan) to wire it into your own workflow?
 > 3. PDF (incl. scanned, via OCR) is already in. Which format should we tackle next — IDML, CSV, something else?
 >
-> Free plan = 2 full documents/month, no card. Link in comments.
+> Free plan = 2 documents/month (10 pages/doc, watermark), no card. Link in comments.
 
 ## Checklist de publication
 - [ ] Assets: logo (1200×630 OG + 600×400), 3 screenshots (résultat côte à côte, sélecteur moteur, glossaire)
 - [ ] 5 commentaires amis planifiés (réponses de qualité, pas de upvote-only)
-- [ ] E-mail aux 150+ contacts de la waitlist (J-2) : « On lance demain, votez + testez »
+- [ ] E-mail aux 150+ contacts de la waitlist (J-2) : « On lance dans 2 jours, votez + testez »
+- [ ] Publier le premier commentaire du maker avec le lien dès l'ouverture du listing
 - [ ] Thread X croisé au lancement

docs/marketing/launch/reddit-posts.md

@@ -15,9 +15,9 @@
 >
 > **What I did instead:** the pipeline parses the document structure, extracts only the translatable content, translates it, and writes it back into the **same** structure. Merges, formulas, fonts, borders, headers/footers, table styles — untouched. Text inside images is handled by vision models.
 >
-> **Tech:** FastAPI (async), openpyxl / python-docx / python-pptx, 7 translation engines the user picks per document (Google, DeepL, Google Cloud, OpenRouter éco & premium, OpenAI, Grok/xAI), Stripe billing, Docker/Postgres/Redis.
+> **Tech:** FastAPI (async), openpyxl / python-docx / python-pptx, 6 translation engines the user picks per document (Google, Google Cloud, OpenRouter éco & premium, OpenAI, Grok/xAI), Stripe billing, Docker/Postgres/Redis.
 >
-> **Pricing:** Free (2 docs/mo, no card) → Starter €9 → Pro €19 → Business €49 (API included).
+> **Pricing:** Free (2 docs/mo, 10 pages/doc, watermark, no card) → Starter €9 → Pro €19 → Business €49 (API included).
 >
 > I'd love feedback from anyone who deals with international documents. What's the nastiest format you've tried to translate?
 >
@@ -58,8 +58,8 @@
 > - **Human translators:** great quality, 50–100× the cost and a multi-day turnaround
 > - **Google/DeepL:** cheap and fast, but they flatten the file — merged cells, formulas, slide layouts get destroyed, so someone has to rebuild the document
 >
-> We built **Office Translator** to kill that middle step: it translates the document **in place** and gives you back a file that looks like it was written in the target language from the start. You pick the engine (7 options, from free Google to premium LLMs), and you can lock your company terminology with a custom glossary so "unité de froid" is never suddenly "cold unit".
+> We built **Office Translator** to kill that middle step: it translates the document **in place** and gives you back a file that looks like it was written in the target language from the start. You pick the engine (6 options, from free Google to premium LLMs), and you can lock your company terminology with a custom glossary so "unité de froid" is never suddenly "cold unit".
 >
-> Free plan: 2 documents/month, no card. From €9/mo after that. Uploads are deleted after 30 minutes and results within 2 hours.
+> Free plan: 2 documents/month (10 pages max, watermarked), no card. From €9/mo after that. Uploads are deleted after 30 minutes and results within 2 hours.
 >
 > If you handle international documents, what's your current workflow and where does it break? Curious what I'm missing.

docs/marketing/launch/show-hn.md

@@ -12,10 +12,10 @@
 >
 > - **In-place translation** for .xlsx / .docx / .pptx / .pdf. The pipeline parses the native structure (cells + merges + formulas, paragraph runs + styles, slide XML + shapes), extracts only translatable content, translates it, and writes it back into the *same* structure. Output looks like it was authored in the target language.
 > - **Scanned PDFs work too**: image-only pages are recovered with OCR (Mistral) before translation — most competitors (DeepL, Azure) reject those outright.
-> - **7 engines, user's choice per document**: Google (free), DeepL, Google Cloud, OpenRouter LLMs in two tiers (DeepSeek / Gemini / Claude), OpenAI, and Grok (xAI). Cheap engine for drafts, premium LLM for client deliverables.
+> - **6 engines, user's choice per document**: Google (free), Google Cloud, OpenRouter LLMs in two tiers (DeepSeek / Gemini / Claude), OpenAI, and Grok (xAI). Cheap engine for drafts, premium LLM for client deliverables.
 > - **Custom glossaries** to lock technical/legal/medical terminology across documents.
 > - **Vision translation** for text inside images (paid plans).
-> - **Privacy**: uploads deleted after 30 min, results auto-deleted within 2 hours, zero data retention, content never used for training.
+> - **Privacy**: uploads deleted after 30 min, results auto-deleted within 2 hours — no long-term retention — and content never used for training.
 > - **API** on the Business plan (10k calls/mo) — the web workflow is fully mirrored: submit file, poll job, download.
 >
 > Stack: FastAPI (Python 3.11), openpyxl / python-docx / python-pptx, PyMuPDF, Postgres + Redis, Stripe, Docker, Next.js 15 frontend.

docs/marketing/launch/x-thread.md

@@ -20,7 +20,7 @@ Every translator you've used works the same way:
 The "hope" step is where your afternoon goes.
 
 **T3 (notre approche)**
-Office Translator works in place:
+Excel, Word, PowerPoint or PDF — Office Translator works in place:
 
 • Parse the real structure (cells, runs, shapes)
 • Extract ONLY the translatable content
@@ -35,10 +35,9 @@ Merges, formulas, fonts, borders: untouched.
 Same file. Different language. Zero reformatting.
 
 **T5 (multi-moteurs)**
-You pick the engine per document — 7 options:
+You pick the engine per document — 6 options:
 
 • Google — free, fast
-• DeepL — best for business text
 • Google Cloud
 • DeepSeek / Gemini / Claude via OpenRouter — eco & premium LLM tiers
 • OpenAI, Grok (xAI)
@@ -54,9 +53,9 @@ Lock "unité de froid → chiller", "clause de résiliation → termination clau
 Your documents are not our training data.
 
 • Uploads deleted after 30 min, results within 2 h
-• Zero data retention
+• No long-term retention, never used for training
 • Encrypted in transit (TLS)
-• No human ever sees your content
+• Processed only by the engine you pick — no human review on our side
 
 **T8 (API)**
 Wiring it into your own stack? Business plan includes API access — 10,000 calls/month.
@@ -66,13 +65,13 @@ Submit file → poll job → download. Same workflow as the web app. Docs at /do
 **T9 (prix)**
 Pricing (yearly = −20%):
 
-• Free — 2 docs/mo, no card
+• Free — 2 docs/mo (10 pages/doc, watermark), no card
 • Starter — €9
 • Pro — €19 (AI translation, glossaries)
 • Business — €49 (API, 5 seats)
 • Enterprise — custom
 
-A page translated costs a fraction of a cent. A human does the same page for €1–2.
+A page costs €0.06–0.10 with us. A human charges €1–2 a page — before fixing the layout.
 
 **T10 (pour qui)**
 Built for:
@@ -82,7 +81,7 @@ Built for:
 • Anyone who's ever re-built a translated Excel by hand
 
 **T11 (preuve sociale + CTA)**
-60+ languages. 7 engines. 100% format preservation.
+107 languages. 6 engines. Format preserved — merges, formulas and styles intact.
 
 Free plan = 2 full documents/month, no card required. Test it on your ugliest file.