What's New in Pluely: November 2026 (v1.2.0)

Pluely 1.2.0 speaks your language, stays with you through a bad connection and gets out of the way of the app you are presenting. Update from the app or download the latest installer (how to update). Your existing license works as is.

Languages

• 22 languages. English, Spanish, Portuguese (Brazil), French, German, Italian, Dutch, Polish, Swedish, Czech, Romanian, Turkish, Russian, Ukrainian, Arabic, Indonesian, Vietnamese, Thai, Japanese, Korean, Chinese (Simplified) and Chinese (Traditional). Pluely starts in your system language; change it at the top of App Settings. The overlay and the dashboard switch together, menus and error messages included. Dates, times and numbers follow your language, and Arabic runs right to left.
• Language in the sidebar. The dashboard sidebar shows Language: Deutsch; click it to jump straight to the setting.
• One language everywhere, if you want it. Two switches under the language picker: Also use it for live transcription and Also answer in this language. Prefer to choose per feature? The Listen language menu and the Response Language setting both offer Same as app language, and a change in one place is picked up everywhere. See Transcription languages.
• Website and docs too. pluely.com and docs.pluely.com are in the same 22 languages, with a language switcher in the footer.
• The translations are new. If a sentence reads wrong in your language, tell us at pluely.com/feedback.

Emails (Limited Pro and Unlimited Pro)

• Meeting summaries by email. When a Listen session ends, Pluely can email a summary with an AI-written title, action items and key points, plus your notes and the full transcript if you want them. Meetings shorter than 2 minutes are skipped by default.
• Email this. A button on chat and meeting pages sends that conversation to your saved addresses.
• A new Emails page in the dashboard sidebar: turn it on or off, send after meetings, up to 5 addresses, a subject template, what to include, the minimum meeting length and a test email. See Emails.

Listen

• A dropped connection no longer ends the session. When the network blips or the server restarts, the composer shows Reconnecting… and Pluely keeps capturing audio while it reconnects, for up to 45 seconds. The most recent seconds are sent through once the connection is back, so the transcript carries on. See Real-time transcription.
• Honest offline wording. When your computer has no internet, Pluely says you are offline. When a service does not answer, it says that instead. Nothing is dressed up as the other, and the app tries again by itself when you are back online.
• Intel Macs. Starting Listen no longer crashes Pluely on Intel Macs.
• A long pause ends the session. A session left paused for more than 30 minutes stops with a note, so an idle pause never runs down your transcription hours.

Stealth, per operating system

• Showing the overlay never steals focus, on any operating system. On macOS the overlay is a non-activating panel; on Windows every window is excluded from captures and never takes the foreground; on Linux, where no system can hide a window from screen capture, App Settings shows a note that explains it and suggests sharing a single window. Details in Stealth mode and screen-share privacy.

Overlay

• To bring the overlay back, open Pluely again from Applications, the Start menu or your launcher. On macOS, clicking the Dock icon works too.
• Always opens on your primary tab.
• Compact is a small pill. In Listen, the Compact view is now a Compact pill in the transcript header.
• Smaller Follow-ups row, and you can collapse it.

Dev Pro

• Guided setup in a new order. Dev Space now starts with Speech (your Deepgram key or another speech-to-text provider), then AI (a CLI or an API key), then Files. The Home page shows how far along you are.
• Reading uploaded files. Mistral is recommended and reads every document type: PDF, Word, PowerPoint, Excel and scans. You can also use the AI key from step 2 when it is OpenAI, Anthropic, Gemini or OpenRouter (reads PDFs), or pick Another PDF reader (Gemini, OpenAI, Claude or OpenRouter). Text files and images need no key. See Dev Space.
• Everything runs on your own keys. Summaries, translations, meeting recaps, smart follow-ups, prompt generation and push-to-talk all use the AI and speech providers you connected. After activation the app only checks your license, shows announcements and looks for updates.

All Conversations

• A Your data dialog. Download all and Delete all moved into a Your data dialog next to Refresh. Both ask before acting and follow the filter you have selected.

Providers and models

• Perplexity is built in. Pick it under AI Providers and paste your key: it runs on Perplexity's Agent API with web search, on perplexity/sonar or on frontier models such as openai/gpt-6-luna and anthropic/claude-sonnet-5-5.
• Current model lists. The example and default models for every provider and CLI match what the providers serve today, including the GPT-6 family, Claude Sonnet 5.5, Gemini 3 and Qwen3-Coder.

Models

• A model is selected the moment Pro activates. Nobody lands on "Select a model" after paying. A model you pick yourself always wins. See Model selection.
• An answer always comes. If the hosted model you picked has not started answering within a few seconds, Pluely retries on a fallback before anything reaches your screen, so an answer never mixes two models.
• Failing models are reviewed every day and taken out of the list until they work again.

Permissions on macOS

• A notice when something is denied. If Microphone, System audio or Screen capture is off, the dashboard shows an amber notice with an Allow button for each one. The first click asks macOS; after that the button opens the exact System Settings pane. It disappears by itself once everything is allowed. See macOS permissions.

Buying from inside the app

• An upgrade that fails to activate keeps your current license active instead of leaving you without one.
• The messages during an in-app purchase are plainer, and a Dev Pro purchase takes you straight to the setup steps.

Fixes

• Custom speech-to-text providers no longer repeat transcript lines, and the last sentence before Stop is saved.
• Removing a provider key now sticks, and each speech-to-text provider remembers its own key.
• Meeting titles use your language and date format.
• JPEG screenshots are sent with the right image type, which some models need to read them at all.
• The compact transcript labels speakers Speaker 1, Speaker 2 instead of raw labels.
• Dev Pro: your own Deepgram key uses the current diarization model, and models that need a different API are not offered.
• CLIs: an error exit is reported instead of an empty answer, a stuck CLI stops after ten minutes, and on Linux CLIs start from your login shell, so tools installed with nvm are found.