Hey DEV community!
Here's my workflow now: I press a hotkey, say what I need, and release. That's it. I'm done.
If I'm writing a message — AI cleans up my messy speech into polished text.
If I need information — AI goes to the internet, finds the answer, and pastes it where my cursor is.
I don't type it. I don't edit it. I don't open a browser. I just speak and move on.
That's Whisper by Remskill. The tagline is "Just ask. Consider it done." — and after using it daily, I can tell you it's not marketing. It's literally how it works.
The core: speak and AI does the rest
This isn't a transcription tool that gives you a raw dump of what you said. That's step one, and honestly, that's where most voice tools stop.
Whisper does the full job:
- You speak — messy, unstructured, with filler words, whatever
- Speech is transcribed — locally on your device or via OpenAI
- AI immediately improves it — cleans grammar, removes filler, restructures into professional text
- Result is pasted — directly into whatever app you're working in
One hotkey. One action. Speak and forget.
What "speak and forget" actually looks like
A Slack message
You say:
"so basically um I looked into the performance issue and it turns out the problem was that we were running the query without an index on the user ID column and uh once I added the index the response time dropped from like 800 milliseconds to about 40 and I already pushed the fix to staging"
What appears in Slack:
Investigated the performance issue — root cause was a missing index on the
user_idcolumn. After adding it, response time dropped from ~800ms to ~40ms. Fix is already on staging.
You spoke for 10 seconds. Didn't think about structure. Didn't edit anything. The AI kept the technical details, dropped the filler, made it concise. Done.
A commit message
You say:
"fix the bug where authentication tokens weren't being refreshed properly when the session expired and the user had to log in again every time"
What appears:
Fix token refresh on session expiry — users no longer forced to re-login
A standup summary
You say:
"john said the API migration is done and it's deployed to staging and sarah is working on the android push notification bug and we decided to push the release to next wednesday"
What appears:
Standup Summary
- API migration complete, deployed to staging (John)
- Android push notification bug under investigation (Sarah)
- Release pushed to Wednesday for additional testing
Speak once. Get structured output. In the app you're already in.
Need information? Just ask.
This is the second half of the equation. Whisper's AI has access to the internet — so when you need an answer, you just ask out loud.
Say your AI keyword before a question:
"Hey Whisper, what's new in React 19?"
Whisper searches the web, pulls data from multiple sources, synthesizes it into a structured answer with cited sources, and pastes it right where your cursor is. ~2 seconds. Zero browser tabs opened.
More examples:
"Hey Whisper, how do I set up a many-to-many relation in Drizzle ORM?"
"Hey Whisper, compare Bun vs Node for a new backend project"
"Hey Whisper, what's the current S&P 500 price and YTD performance?"
"Hey Whisper, how do I fix a segmentation fault in Rust?"
Each time: you speak the question, the answer appears where you're working. You never leave your editor. You never open Chrome. You never copy-paste from a search result.
Just ask. Consider it done.
Custom AI instructions
You're not limited to "clean up my text." You can tell the AI exactly what to do:
- "Summarize this in 3 bullet points"
- "Translate to Spanish"
- "Write this as a formal email"
- "Make this more concise"
- "Convert to a bullet list"
- "Write a git commit message from this"
Save presets for the ones you use often. Switch with one click.
Where the AI runs — your choice
Cloud (OpenAI): GPT-4o or GPT-4o mini. Paste your API key, done. ~$0.003/min for transcription, zero markup from us.
Local (Ollama): Run Llama, Mistral, or any model on your own machine. Fully private, fully offline. Whisper connects to Ollama automatically.
Same for transcription — run it via OpenAI's API for max accuracy, or locally via faster-whisper with 5 model sizes (75 MB to 3 GB). Mix and match however you want.
The app itself is ~25 MB. Built with Tauri v2 (Rust backend, not Electron).
Works in every app
One global hotkey. Every application. No plugins, no extensions, no setup per app.
- Editors: VS Code, JetBrains, Vim/Neovim, terminal
- Communication: Slack, Discord, Teams, Outlook, Gmail
- Docs: Notion, Obsidian, Word, Google Docs
- Browsers: Chrome, Firefox, Edge
- Anything with a text field
Press the hotkey in Slack — works. Press it in VS Code — works. Press it in a random form field on a website — works.
Privacy if you want it
In local mode:
- Voice never leaves your device
- Audio processed in RAM only — never written to disk
- Discarded immediately after transcription
- Zero network traffic
- No telemetry
Disconnect your internet. It still works. That's the proof.
This matters for proprietary code, medical work (HIPAA-safe), legal work (attorney-client privilege), and air-gapped environments.
Dev-specific things
Custom vocabulary. Add your stack's terms: tRPC, Bun.js, Drizzle ORM, FastAPI, PostgreSQL, Kubernetes. Case-sensitive, always spelled correctly.
Documentation by voice. Nobody likes writing docs. Speaking them is 3x faster. The AI makes it read well.
90+ languages. Auto-detection. Speak in one language, get output in another.
Configurable everything: hotkey, recording mode (push-to-talk or toggle), theme, paste method, AI model, custom prompt, overlay position.
Pricing
- Monthly: ~$9.99/mo
- Yearly: ~$79.99/yr (~$6.67/mo)
- Lifetime: $99 one-time — pay once, own forever
7-day free trial. Team plans from $5/user.
System requirements: Windows 10+ or macOS 12+. 4 GB RAM. ~25 MB install. No GPU required. Linux coming soon.
Press a hotkey. Speak. AI turns your messy voice into clean text — or goes to the internet and finds the answer you asked for. Result appears wherever your cursor is. One action. Done.
Just ask. Consider it done.
whisper.remskill.com — 7-day free trial, 2 minutes to set up.
Happy to answer questions in the comments.
Speaking is 3x faster than typing. But speaking + AI in one step? That's the part that changes everything.
Top comments (0)