Features
Speech-to-well-written text is finally working well
WritingAnt, as a dedicated modern dictation solution, is likely much better than other dictation solutions you might have tried before — such as, for example, the dictation built into AI tools or chats.
Much better text result quality, adapted to the situation
- Keep the text exactly as you spoke it when you want that.
- Get help rewriting your text automatically.
- Personalizable writing styles.
Walk around and think freely
- Connect a wireless headset and walk around while recording your speech to text.
- Or relax on your comfortable couch.
- Record for up to 20 minutes and stay easily in control — no automatic cut-off just because you pause to think, unlike other dictation tools.
One shortcut across every app
Double-press ⌘ Command to start recording. Prefer another trigger? The Globe (fn) key or a spare mouse button work too.
Volume-to-zero option for music lovers
Your speaker volume drops to zero automatically while speech-to-text is recording and returns when it stops.
Be proud of everything you write even with speech-to-text
An automatic writing assistant rewrites your text in a formal or less formal tone — even with speech-to-text — and you can always choose to keep it exactly as you spoke it.
Apply it to text you’ve already selected, or let it run automatically on your speech-to-text results.
Configure which writing style is applied automatically, based on the active program, browser tab, or website.
You can also personalize the writing styles. They quickly learn from the examples where you weren’t happy and rewrote the text — generalizing from them for your future writing, so it soon sounds just like you wrote it.
Our goal is for you to feel proud of everything you write — well-written text that fits each situation and still sounds like you, judged on how well it reads, not only on how quickly you reply.
Choose your privacy level
Local AI means transcription happens on your computer and the writing data stays there.
AI-Enhanced Accuracy
When using a Cloud AI writing style, the speech-to-text result is improved with multiple AI iterations (the latest GPT AI models, or open-source AI models hosted in Switzerland) to achieve well-written text that requires minimal manual edits compared to traditional speech-to-text.
Suitable AI Models
You choose in the app where your text is processed, and each option runs the model that gives the best results for that hosting. Global processing: the latest GPT AI models from OpenAI, currently GPT-6 Luna and, for extra-long texts, GPT-6.1 Terra. Processing in Switzerland & Europe: open-source AI models hosted by PHOENIQS in Switzerland, currently Gemma 4 31B, the open model closest to GPT in result quality and speed. Models change as better ones become available.
Gemma 4 31B is published by Google DeepMind under the Apache 2.0 licence and hosted by PHOENIQS in Switzerland. The GPT models are provided by OpenAI under the OpenAI Services Agreement.
Computer Integration
Write with voice commands or keyboard shortcuts. Results are inserted in the text field where the cursor is located in almost all applications.
Application Aware
Switch to Slack, Teams, Notes, or a specific website and WritingAnt automatically applies your preset language and writing style. It re-applies every time you enter the app, and you can always change it manually.
Fix highlighted text instantly
Highlight any text in any app and double-tap Shift — grammar and spelling are corrected right in place. You can also speak an instruction to rewrite the selection your way.
Smart Text Enhancement
Automatically improves grammar, punctuation, and writing style. Best practice writing rules automatically applied based on writing style.
Your writing style, applied two ways
The AI writing assistant applies your predefined or personalized writing style across any application — highlight text you've typed to rewrite it in place, or let it polish your dictation automatically. The same assistant, working before and after you speak.
Structure & bullet points
When using a Cloud AI writing style, the system automatically adds structure in addition to text enhancements. This includes inserting new lines where appropriate, converting content into bullet points (as dash signs) when suitable, and making other structural improvements.
Basic Speech-to-Text
Trigger basic modern speech-to-text with voice commands for faster results with good quality (Not as polished as the cloud AI writing styles, but probably better than dictation you tried before.). Perfect for writing a few sentences or when data privacy requires offline processing.
Creative Writing
In addition to writing enhancements of what you say, you can also benefit from creative writing. You can give a voice prompt for what to write, and the AI will generate creative content based on your instructions.
Keyword Replacements
For the few names and terms you use extremely often that are not recognized out of the box, set a short spoken keyword such as “my company”, and the exact spelling is inserted every time. WritingAnt deliberately does not boost the detection of vocabulary words, because that makes them appear in place of similar words you actually said.
Supported Languages
Dictate and write in 20 languages, grouped in two accuracy tiers.
English, Spanish, German, French, Italian, Portuguese, and Russian.
Swedish, Danish, Norwegian, Dutch, Ukrainian, Polish, Czech, Slovak, Hungarian, Romanian, Croatian, Bulgarian, and Finnish.
The app's user interface is also available in English, Spanish, German, French, Italian, Portuguese, Russian, Swedish, Dutch, Polish, Czech, Slovak, Hungarian, Romanian, Croatian, Bulgarian, Ukrainian, and Finnish.
Voice Computer Control
We also provide controls for steering the computer using common keyboard shortcut commands triggered by voice. You can turn this functionality on or off. As a bonus feature, voice commands are available in English only.
Modern Transcript Models
Currently we use the AI model NeMo Parakeet for transcription.
Discover even more inside the app
Try it freeFrequently asked questions
Does WritingAnt work in every application?
Yes. It inserts text at the cursor in almost any application on Windows and macOS: email, documents, chat, browsers, CRMs, IDEs and AI tools. Application-aware presets can switch the writing style and language automatically per app.
What is the difference between the Basic style and the polished styles?
The Basic style runs fully on your computer and gives you your words as you said them, near-instantly. The polished styles send the text transcript to the cloud, where the AI model of your chosen processing option (the latest GPT AI models, or open-source AI models hosted on Swiss and European servers) rewrites it into well-written text in the register you chose.
Can it sound like me or my company?
Yes. Define your own style or edit a built-in one, then teach it by example with the Teach button whenever a result is not quite right. The custom writing styles page explains how.
How long can I record?
Up to 20 minutes per recording, with no automatic stop when you pause to think.
Can I edit text by voice?
Yes. Highlight text and speak an instruction such as "fix grammar" or "make it more concise". During a recording you can also say how the text should be finished, for example "add titles and format for Word".
Which languages are supported?
Twenty: English, Spanish, German, French, Italian, Portuguese, Russian, Dutch, Polish, Czech, Slovak, Hungarian, Romanian, Croatian, Bulgarian, Ukrainian, Finnish, Swedish, Danish and Norwegian.
Does WritingAnt upload my audio?
No. Speech recognition runs locally on your computer. Only the text transcript is sent to the cloud, and only when you use a polished style.