Win
Works across Windows apps
Injects text directly into Outlook, Gmail, Word, Slack, Teams, browsers, CRMs, and any app where pasted text works.
Light, Fast, & Simple
Dilo writes at the speed of thought.
Dilo turns your voice into polished text in any Windows app. Like a feather, it is light, fast, and simple. Different than typing, which eats your time.
Windows-first voice dictation
Tap a hotkey, speak naturally, and place clean writing wherever your cursor is.
- Lightweight Tray App (tens of MBs)
- Two distinct cleanup modes
- Style-learning (STYLE.md)
- Honest, transparent privacy
Dilo Features
Voice typing for people who write all day.
Dilo is tailored for coaches, consultants, lawyers, and writing-heavy knowledge workers who want to replace the typing bottleneck without the overhead of heavy software.
2x
Two cleanup modes
Smart mode polishes and formats paragraphs. Conservative mode preserves your exact words, removing only fillers and adding punctuation.
OK
Faithful by design
Our cleanup pipeline is heavily constrained to improve tone and structure without ever inventing facts or changing your core meaning.
You
Sounds like you
SMART cleanup learns from your writing style. Ingest writing samples so the output adopts your voice and phrasing naturally.
Aa
Preserves your language
Dilo automatically detects the spoken language per recording, outputting in that exact language (optimized for English and PT-BR).
tray
Lightweight background client
A native Windows tray application using minimal system RAM, completely avoiding the bloat of heavy 800MB Electron clients.
Workflow
From thought to clean text in one hotkey.
Dilo runs quietly in the background. It doesn't replace the apps you love—it makes writing in them faster.
Tap Hotkey
Press a system-wide hotkey to start recording (no need to hold it down).
Speak Naturally
Dictate your message, brainstorm a draft, or speak a rough response in your own voice.
Tap Again
Press the hotkey to stop. Dilo captures and processes the audio in under a second.
Auto-Insert
The AI-polished text is immediately inserted at your cursor in the active window.
Tuned for Trust
A desktop helper that respects your intent.
Most dictation apps use a single generic AI prompt for all transcripts. Dilo gives you granular control over how much cleanup is applied.
Switch modes instantly depending on the task: use Smart mode to draft follow-up emails, or Conservative mode when dictating legal terms or precise instructions where every spoken word matters.
Smart Mode Ctrl+Win
Polishes paragraphs, corrects grammar, and structures text while staying completely faithful to your original meaning.
Conservative Mode Alt+Win
Your exact words, retaining precise vocabulary while cleaning up fillers, pauses, and punctuation.
Privacy First
Honest privacy. Plain disclosures.
We don't claim "your audio never leaves your computer" because modern, low-latency AI transcription requires cloud processing. Instead, we are plain about what leaves your device and how we are securing it:
Provider-Only Processing
Your audio is sent to our speech-to-text and cleanup provider (Groq) only to transcribe and polish it — the app keeps no copy. We are putting zero-data-retention terms in place with that provider so your audio is not retained or used for training. (Pending signed ZDR confirmation.)
Local-First History
Your transcription history, dictionary, and tone-learning profiles are stored locally on your machine—never on our servers.
Audio Saving is OFF
Dilo never saves audio recordings to your disk unless you explicitly choose to turn it on for debugging or training your style.
Future Offline Tier
On-device transcription using local models is planned as a future tier for users requiring total offline compliance.
Pricing
Simple pricing for early adopters.
Start free in the beta, or subscribe to Pro for daily, high-volume use. Accounts and billing are handled securely through Stripe.
Early Access
Free Beta
For professionals willing to help test and refine the app.
- Windows-first tray app client
- Smart and Conservative cleanup
- Local transcript history feed
- Custom style-learning profiles
- Direct feedback loop with the builder
Dilo Pro
R$ 39 / month
≈ US$12 / mo
For daily, high-volume professional use.
- High-volume transcription limits
- Provider fallback chain (Groq + Deepgram + Soniox)
- Advanced Custom Vocabulary editor
- Prioritized low-latency server processing
- Multi-device sync (opt-in)
Billed monthly in Brazilian Real. Cancel anytime — secure checkout via Stripe.
Join the Beta
Help build the Windows dictation app you actually want to use.
We are looking for writing-heavy users (coaches, consultants, legal professionals, founders) who spend hours typing drafts and messages every day.
If you are on Windows 11, want to save hours of typing, and can provide honest feedback on accuracy, speed, and formatting—we'd love to have you.
FAQ
Clear answers. No marketing hype.
Does Dilo work in every application?
Yes. Dilo is designed to work system-wide. It uses reliable clipboard and native input simulation methods to inject text directly at your cursor in any active text box, browser, native application, or even elevated windows.
Is Dilo available for download?
Yes — Dilo is in open beta for Windows and macOS. Create your account, then open the download page to grab the installer for your system. It's an unsigned beta for now, so your OS adds one extra confirmation click the first time — we walk you through it.
How does style tone learning work?
You can paste samples of your own writing into the Hub. When you dictate in SMART mode, the system feeds these samples to the cleanup model as a tone reference, teaching it to match your vocabulary and phrasing without altering the facts you spoke.
What AI models are used under the hood?
By default, Dilo uses Groq's Whisper API for ultra-low latency transcription and Groq's Llama 3.3 (70B) for text cleanup. We also build fallback seams for Deepgram, Soniox, and Claude Haiku, ensuring reliability.
How lightweight is the Dilo client?
Dilo is a native Windows tray application written in Python. It consumes tens of megabytes of memory and runs in the system tray, making it significantly lighter than heavy Electron alternatives (which often exceed 800MB of RAM).
Stop typing the first draft.
Use your voice to write emails, notes, messages, and documents across Windows 11.
Join Early Access