Incognito PrivoAI

Incognito PrivoAI — browser tools for transcription, translation, temporary chat and reading. Independently developed and maintained. The extension is free, ad-free, with no developer tracking or subscription. Gemini API use may be billed. ──────────────────────────────── 🆕 What's New (v28.18.0) ──────────────────────────────── • Chinese transcripts default to a saved Traditional Chinese preference. Spoken language and translation target are separate, with the actual recognition service shown. Flash-Lite / Flash receives a Traditional Chinese instruction; dedicated Transcribe still auto-detects Chinese and does not guarantee the script. Enable translation for another output language; local language packs depend on Chrome availability. • Basic recording and imported audio default to Economy: Gemini 3.5 Flash-Lite. You can select Flash or dedicated Transcribe. Economy mode never automatically switches to a more expensive model. Cloud audio goes to Google and may incur charges; recognition can be incomplete or incorrect. • The default AI model, Lite, has been upgraded to Gemini 3.5 Flash-Lite for fast everyday questions and translation • Flash has been upgraded to Gemini 3.8 Flash, balancing speed and quality for images, documents, and complex tasks ──────────────────────────────── ✨ Key Features ──────────────────────────────── 🌐 Webpage Translation (in-place / screenshot) Translate webpage text in place or use screenshot OCR when needed. Supports bilingual reading and selected translation ranges, with controls to switch languages or restore the page. Layout preservation and results depend on the site and extraction method; not every page, image, ebook or PDF can be processed. DOM-based translation can use your configured AI endpoint, while Gemini screenshot OCR requires cloud access. 🎙️ Voice transcription Record up to about 59 minutes 50 seconds or import audio up to 200 MiB. Gemini Transcribe accepts files up to 60 minutes; longer files must be split or processed with Flash under that model’s limits. Basic recording transcribes after Stop; Gemini Transcribe Live displays text while you speak and can leave an incomplete transcript if interrupted. Recognition can contain errors or omissions. Optional translation uses finalized text. Chrome on-device dictation (Testing) depends on the country/region, user, Chrome version and language pack, and has no cloud recognition fallback; enabling translation may send recognized text to your configured AI provider. Copy or export the transcript, open it in Reader, or download/retry the latest recording while this page remains open. Audio kept in page memory is cleared when the page closes. 💬 Temporary chat This feature does not turn on Chrome Incognito mode. The extension does not persist the chat transcript after the chat window closes; Google or your configured endpoint may still process or retain data under their policies. Four tabs are available: • Lite (Gemini 3.5 Flash-Lite) — fast everyday text questions and translation • Flash (Gemini 3.8 Flash) — balanced speed and quality, supports images and documents • Pro (Gemini 3.1 Pro, preview) — for reasoning and detailed document analysis; access and quota depend on Google • Local — text chat through your configured Ollama / LM Studio or compatible endpoint (localhost or LAN; not necessarily offline) The default Lite suits many text tasks. Includes a searchable prompt library. Lite/Flash/Pro chat and attachments still go to Google even in Local mode; choose the Local tab for local chat. 🧰 Text & image translation Paste text or upload an image screenshot for OCR and translation, optionally with an AI summary, opened in the Reader. 📖 Reader Read extracted content in a clear layout and select paragraphs to translate. Choose a language to read aloud, and adjust font size, line spacing and background color. Browser-provided voices may be local or remote. 🔊 Summaries & read-aloud (Beta) Summarize retrievable YouTube captions, audio files, webpages, PDFs or pasted text, then read the summary aloud. Text summaries can use your configured model in Local mode. Audio, PDF processing and WAV generation require Gemini. YouTube uses captions, not full-video transcription. Download format is WAV; downloaded speech uses only the first 5,000 characters. Gemini TTS requires cloud access and may incur API charges. ──────────────────────────────── 🔒 Privacy by Design ──────────────────────────────── • No developer relay: Gemini content and your API key go directly from the browser to Google for processing and authentication • Developer data handling: no developer-run service receives your inputs or outputs; cloud content is still subject to Google’s processing and retention terms • Local storage: API key and prompt library use extension local storage without Chrome Sync; other preferences may sync • Configured AI endpoint: Local text chat and text/DOM translation can use localhost or a LAN server. Choose Local chat explicitly; Lite/Flash/Pro tabs and attachments still go to Google • Local mode disables automatic Gemini OCR, image/PDF processing, cloud voice and TTS. On-device dictation may be available (Testing); optional translation follows the configured provider • Google data terms: Unpaid Services may use inputs/outputs for product improvement and human review; do not submit sensitive, confidential or personal information. Paid Services do not use prompts/responses for product improvement but have limited safety retention. In the EEA/Switzerland/UK, Paid Services data rules also apply to AI Studio and unpaid API quota; see the current terms ──────────────────────────────── 🌏 Multilingual Interface ──────────────────────────────── 13 interface languages: Traditional Chinese, English, Japanese, Korean, German, Spanish, Arabic (with RTL), Russian, Brazilian Portuguese, Turkish, Ukrainian, Vietnamese, Czech. ──────────────────────────────── 🚀 Getting Started ──────────────────────────────── 1. Install the …