# Rubil — Full Site Content

> Voice dictation that formats your speech for the app you're writing in. Available as a Chrome extension and Mac desktop app.

Rubil is a voice dictation tool (speech-to-text) for knowledge workers. It transcribes speech and formats it for the platform you're writing in — Gmail gets email structure, Slack gets concise messages, Notion gets clean prose, and 20+ other apps each get platform-appropriate formatting. Rubil formats your words. It never rewrites or generates them. This "ghostwriter, not generator" philosophy is the core differentiator from competitors like Wispr Flow, Willow Voice, and Aqua Voice.

## Product Details

- **Chrome Extension**: https://chromewebstore.google.com/detail/rubil-%E2%80%93-voice-dictation-f/pdlbogddhgbojodnnpfneoabgcdccnai
- **Mac Desktop App**: https://download.rubil.io/Rubil.dmg
- **Pricing**: Free (7-day Pro trial, then 1,000 words a week) or Pro ($5/month or $39/year founding annual, unlimited)
- **Languages**: All languages supported by Groq's transcription model
- **Changelog**: https://www.rubil.io/changelog — Product updates for Rubil, the voice dictation tool for Chrome and Mac. What shipped, in plain language, from the founder.

---

## Quality: The Post-Edit Rate

URL: https://www.rubil.io/quality

Definition: Post-Edit Rate (PER) is the percentage of voice dictations that need at least one edit before they are fit to send. It scores send-readiness rather than transcription accuracy. Lower is better and zero is the goal. The name comes from post-editing, the term translators use for the cleanup a human does on machine output. A dictation counts against the rate if a reasonable user would fix even a single character before sending.

Current result (August 2026): Post-Edit Rate 0.43%. Per surface: General writing 0.77%, Email 0.69%, Direct message 0%, AI chat prompt 0%. The board is a live status page updated weekly, showing the current rate, a history strip per surface, and a dated run log.

How Post-Edit Rate differs from Word Error Rate: Word Error Rate (WER) counts how many individual words a speech recognizer got wrong against a reference transcript, and it is the right metric for judging a recognizer. Post-Edit Rate counts whole messages that needed human intervention before sending. A dictation can score a perfect 0% WER and still fail on Post-Edit Rate: every word transcribed correctly, but no punctuation, no paragraph breaks, no greeting, and filler left in, so the user still has to clean it up. WER usually ignores punctuation and structure, while Post-Edit Rate counts them as failures. WER requires a reference transcript; Post-Edit Rate grades against published rules instead. The two stack rather than compete: WER is upstream, and Post-Edit Rate is what the user experiences.

What counts as an edit — four categories, and one edit in any of them fails the whole dictation. Content: a fact is missing, invented, or changed. Voice: the speaker's phrasing, hedging, or intensity was altered, so softening "kind of satisfying" to "satisfying" counts, as does toning down "pissed off". Format: the output shape does not fit the surface, such as a dictated list returned as a paragraph. Mechanics: punctuation, capitalisation, greeting comma, sentence boundaries.

Why the bar is asymmetric: Rubil formats the user's words and never rewrites, generates or summarises them, so an output that reads more smoothly than what was said is a failure rather than a bonus. Deleting a hedge like "I think", softening intensity, or tidying a dictated number all count against the score, because each one puts words in the speaker's mouth that they did not say. Numbers dictated as digits are reproduced character-for-character, so an order number or a build number is never silently reformatted.

Why each surface is scored separately: a dictation that is correct for Slack is wrong for email, so averaging them hides the surface a given user actually writes on. Email gets a greeting on its own line, paragraph breaks at topic shifts, and only the closing the speaker dictated; names stay as spoken, so "Hi Sarah" never becomes "Hi, Sarah". Direct messages get chat register — one or two compact blocks with the greeting inline, shorter in shape but never shorter in substance. AI chat prompts keep their structure, so a dictated sequence renders as a numbered list and an instruction stays an instruction rather than being answered. General writing covers documents, notes and posts, where long dictation is broken into paragraphs at natural topic shifts.

Methodology: each weekly run sends the full scenario set through the live production pipeline several times over, because the failures that matter are intermittent and a single clean run hides them. Scenarios are written to be hard, so the result is a stress test rather than a picture of an average day. Grading is deterministic against published rules; an independent AI judge from a different vendor than the formatting model reviews every output as an advisory flagger and cannot change a published number.

Run history (August 2026, most recent first): Value-preservation guard, 0.43% — dictated values reproduced exactly as spoken. Formatting model upgrade, 1.88% — direct messages and AI chat prompts reached zero edits. Fidelity guards, 4.48% — spoken contractions, profanity handling and long-dictation paragraph structure moved from guidance into enforced rules. Baseline, 6.79% — first measurement against the expanded scenario set.

Vocabulary metrics (Glossary terms, @handles, acronym rules) are tracked internally and excluded from the public Post-Edit Rate.

---

## Try Rubil, free voice dictation demo

URL: https://www.rubil.io/try

A free in-browser demo for people who want to test voice typing before installing anything. The visitor speaks one short message and Rubil returns the same words formatted three ways at once, for Gmail, Slack, and Notion, shown side by side. No account is needed for the first few tries. Target phrases: try voice dictation online, voice dictation demo, test voice typing.

The demo processes audio transiently through Groq to transcribe it, then discards it. The transcript is sent to OpenAI to format it, then discarded. Nothing is stored. After a few free tries, the visitor can add their email to keep dictating on the free tier: 7 days of unlimited Pro, then 1,000 words a week. Rubil formats your words and never rewrites or generates them.

---

## The Ghostwriter Philosophy

URL: https://www.rubil.io/ghostwriter

Rubil's ghostwriter position stated plainly. Rubil formats your words and never rewrites or generates them, and it only knows the words you teach it.

---

## About — Your words. Not ours.

Rubil is a voice dictation tool built on a simple belief: AI should format what you say, not rewrite it.

### The problem

Voice dictation has been around for decades, but it still doesn't work for professional communication. You speak a paragraph and get back a mess. Punctuation is wrong. Formatting is missing. The tone doesn't match the app you're writing in. So you spend 10 minutes editing what should have taken 30 seconds to say.

Every other tool in the space is solving this by adding more AI. They rewrite your sentences. They suggest better phrasing. They generate content on your behalf. We think that's the wrong direction.

### Our approach

Rubil is a ghostwriter, not a generator. You speak. We format. Your vocabulary, your tone, your ideas — we clean up the structure, fix the grammar, and adapt the formatting to the platform you're writing in. Gmail gets email structure. Slack gets concise messages. Notion gets clean prose. Same voice input, different output.

Your words stay yours.

### How it works

- **Chrome Extension**: One click to install. Rubil lives in your browser and works in every web app — Gmail, Slack, Notion, Google Docs, and 20+ more. Text appears directly in the field you're writing in.
- **Mac Desktop App**: A lightweight menu bar app with a global hotkey. Works in native Mac apps like Apple Mail, Slack desktop, VS Code, and Claude desktop. Same formatting engine, same Glossary.
- **Your Glossary**: Rubil only knows the words you teach it. Add names, acronyms, technical terms, and text expansions with custom rules. Your Glossary syncs across Chrome and Mac.

### Why we stay focused

Every voice dictation tool in the market is racing to add more AI. More rewriting. More generation. More features that think for you. Rubil goes the other direction. We do one thing — format your speech for the app you're writing in — and we do it better than anyone else. That focus is intentional. It's what lets us ship faster, stay privacy-first, and keep the product simple enough that it disappears into your workflow.

---

## Voice dictation guide

URL: https://www.rubil.io/voice-dictation-guide

The voice dictation guide for knowledge workers: how speech-to-text works in 2026, how to choose a tool, and how to dictate in the apps you already use.

### What voice dictation is

Voice dictation is speech-to-text software that turns what you say into written text in the apps you already use. You speak into a microphone, an AI model transcribes the audio in real time, a formatting layer cleans the output, and the text appears in whatever input field you're focused on. Modern voice dictation tools work in two stages: a transcription model converts speech to raw text, and a formatting layer adds punctuation, paragraph breaks, and platform-appropriate structure before the text lands in your app.

Voice dictation is not the same as meeting transcription. Otter, Fireflies, and Granola record conversations and produce summaries after the meeting ends. Voice dictation produces ready-to-send text as you speak it. It is also not the same as an AI writing assistant. ChatGPT, Claude, and Copilot generate content from a prompt — they write for you. Voice dictation writes what you said. The category got serious in the past 18 months: Whisper-class transcription models from OpenAI and Groq made speech-to-text fast enough and accurate enough to use without thinking about it.

### Why voice dictation is changing

Three things changed at once. First, transcription got fast — models like Whisper and Groq's distilled variants do high-accuracy transcription in under 500 milliseconds, so the lag is gone and speech is now viable for short messages, not just long-form content. Second, knowledge work got chattier: the average professional writes across Gmail, Slack, Notion, Teams, Linear, Jira, and four or five other tools in a single day, most of it short-form. Third, AI started writing instead of helping you write — ChatGPT, Claude, and a wave of AI writing assistants pushed "AI does the writing for you," which created a backlash from people who find AI-generated prose flat. The same wave hit voice tools: Wispr Flow's Command Mode rewrites your sentences, Willow Voice's AI Mode transforms rough notes into polished output, and Aqua Voice promises to refine your words as you talk. These features are useful but are a different category — closer to dictated AI writing. The category is splitting, and before you pick a tool you should decide which side of it you're on.

### The two real choices in a voice dictation tool

Model accuracy and microphone quality are commodities now. The two real choices are philosophy and surface coverage.

Formatting vs generating: Camp one formats your speech — it adds punctuation, fixes grammar, breaks paragraphs, adapts structure for the app, and inserts your words at the cursor, so the output reads like you wrote it because you wrote it. Camp two generates content from your speech — its AI layer rewrites your input into something polished, restructuring sentences and swapping word choices, so it reads cleanly but more like AI than like you. Both are valid. If your output goes to people who know your writing, the formatting camp keeps your voice intact; if your output is rough thoughts you want polished, the generating camp does that work. A quick test: read your last five Slack messages out loud — if you'd be happy seeing those as formatted text, you want camp one; if you'd want the AI to make them sound more "professional," you want camp two.

Browser vs desktop: Most voice dictation tools are desktop-only (Mac or Windows apps that inject text into the focused app), which works for native apps like Apple Mail, Slack desktop, or Linear's native client. A few are browser-only (Chrome extensions) and work in web apps like Gmail, Notion, and Google Docs. Almost no tool covers both, which is a problem for knowledge workers who move between web and desktop all day. The fix is dual distribution: a Chrome extension and a Mac app that share the same engine and the same personal vocabulary.

### Voice dictation by app

Different apps need different formatting. Gmail needs full email structure (greeting, paragraph breaks, complete sentences, sign-off). Slack needs short, direct, conversational messages with the right @mentions, acronyms, and project names — no greetings or sign-offs. Notion, Google Docs, and document-style writing need clean prose with proper paragraph breaks and full sentences. Microsoft Teams sits between Slack chat and email. Linear, Jira, GitHub, and engineering tools need code references in code formatting, intact ticket numbers, and exact spelling for technical terms — handled best by a personal Glossary. Apple Mail, Apple Notes, and native Mac apps require a desktop app, not just a Chrome extension. Claude desktop, ChatGPT, and AI assistants want short, direct prompts with no greeting or sign-off.

### How to choose a voice dictation tool

Four questions, in order. (1) Do you want your words formatted or rewritten? Formatting-camp tools include Rubil, Voice In, and the default mode of most apps; generating-camp tools include Wispr Flow with Command Mode, Willow Voice with AI Mode, and Aqua Voice with natural language editing. (2) Where do you write? Mostly web apps means a Chrome extension covers most of it; native Mac apps mean you need desktop coverage; both means you need dual distribution. (3) How important is personalization? A flat custom vocabulary list versus structured personalization that handles names, @mentions, and platform-specific behavior (an acronym that expands in email but stays short in Slack). (4) What's your privacy posture? Check whether the tool stores audio, stores transcripts, reads screen content, uses your data to train models, and names every subprocessor. A tool that can't answer clearly is one to skip. Most knowledge workers end up wanting a formatting-camp tool, dual distribution, structured personalization, and a privacy architecture that doesn't store their speech.

### How Rubil approaches voice dictation

Rubil is a ghostwriter, not a generator. It formats what you say without rewriting it — verbal hedges, sentence structure, vocabulary, and cadence are kept, and the formatting adapts to the app you're in (email structure for Gmail, concise structure for Slack, clean prose for Notion and Docs, conventions for Linear and Jira). Two surfaces — a Chrome extension for any web app and a Mac desktop app for native Mac apps — share the same backend, formatting rules, and personal Glossary. You teach the Glossary your vocabulary by adding entries as you speak, and Rubil can suggest an entry when you correct the same word twice. Privacy is the architecture: audio is processed transiently and discarded, transcripts are formatted and discarded, and nothing (voice files, transcripts, or formatted output) is stored on servers. Your Glossary is stored in our cloud database (Supabase), isolated to your account with row-level security and encrypted in transit and at rest, so it syncs across your devices. Multilingual support comes from Groq's transcription model: 50+ languages including Spanish, Portuguese, French, German, Japanese, Chinese, Korean, Arabic, and Hindi.

### Voice dictation FAQ

- Speech-to-text is the underlying technology (an AI model converts audio into text); voice dictation is the product category that uses speech-to-text plus a formatting layer to insert ready-to-use text into the apps you write in.
- Voice dictation works in Gmail and Slack if the tool supports them — some are browser-only, some desktop-only, a few cover both via dual distribution.
- It is faster than typing for most knowledge workers: people speak ~150 words per minute and type ~40–60, though heavy reformatting gives back the gain.
- Modern voice dictation is accurate enough for professional work — Whisper-class models hit 95%+ on clear speech, with a good personal Glossary fixing specialized vocabulary.
- It works in 50+ languages via OpenAI Whisper and Groq, including Spanish, Portuguese, French, German, Italian, Japanese, Chinese, Korean, Arabic, and Hindi.
- A low-volume or whisper mode lets you dictate in a quiet or shared office.
- Voice dictation turns what you say into text in your own words; AI writing assistants generate content from a prompt in the AI's words — they're complementary, not competing.
- Whether tools store your audio depends on the tool; the honest ones name every subprocessor and document retention on a public Trust or privacy page.
- Meeting transcription (Otter, Fireflies, Granola) records multi-person conversations and summarizes after; voice dictation produces formatted text from a single speaker while you speak.
- Voice dictation is worth paying for if you write more than two hours a day across multiple apps; otherwise free tiers usually cover your usage.

---

## Features — Everything Rubil does

Voice dictation that adapts to where you're writing and never stores your audio. Rubil only knows the words you teach it.

### Platform-specific formatting

Same voice input, different output depending on where you're writing. Rubil reads the URL in Chrome or the active app on Mac, and formats accordingly. Gmail gets email structure with subject lines and greetings. Slack gets concise, direct messages. Notion gets clean prose with headers. Google Docs gets document-quality writing. 20+ platforms and growing.

- **Email**: Gmail, Outlook (web + desktop), Apple Mail
- **Messaging**: Slack (web + desktop), Microsoft Teams, Discord, WhatsApp, Telegram, iMessage
- **Documents**: Google Docs, Notion, Apple Notes, Word, Confluence, Coda, Dropbox Paper
- **Code & PM**: Jira, Linear, GitHub, GitLab, Asana, Monday, ClickUp, Trello, VS Code, Figma
- **Social**: LinkedIn, Twitter/X
- **AI**: Claude desktop, ChatGPT

### 4-category personal Glossary

Most dictation tools have a basic custom vocabulary list. Rubil has a structured glossary system with four categories, each designed for a different kind of personalization.

- **People**: Teach Rubil your team's names, nicknames, and how they're spelled.
- **Acronyms**: Control when acronyms expand vs stay abbreviated, per platform.
- **Terms**: Lock the spelling of product names and technical terms.
- **Expansions**: Map trigger phrases to fixed text blocks you use every day.

### Whisper mode

Optimized for low-volume dictation. Use Rubil in open offices, shared workspaces, coffee shops, or anywhere you don't want to project your voice. Speak softly and Rubil still hears you.

### Inline dictation

No floating window. No copy-paste. Text appears directly in the field you're writing in. On Chrome, the extension injects text inline. On Mac, the app pastes directly into the focused field. Zero friction between speaking and seeing your text.

### Privacy by architecture

Audio is processed transiently and discarded. Transcripts are formatted and discarded. We never store your audio, your transcripts, or your formatted text on our servers. Every data processor is named on our Trust page. Your Glossary is stored in our cloud database (Supabase), isolated to your account with row-level security and encrypted in transit and at rest, so it syncs across your devices.

### Dual distribution

Chrome extension for the browser. Mac app for everything else. Both call the same backend, share the same Glossary, and follow the same formatting rules. Install one or both — your Glossary syncs across surfaces.

### Generous free tier

Seven days of unlimited Pro on free, starting on your first dictation. After that, 1,000 words a week on the same engine paid users get. No credit card required. Pro is $5/month, or $39/year on the founding annual plan, for unlimited.

---

## Trust & Privacy — How Rubil handles your data

Rubil is a voice dictation tool. You speak, we format, the text appears in your app. Here's exactly what happens to your data at every step — and what doesn't happen.

### What happens when you dictate

1. **You speak** — Your microphone captures audio in your browser (Chrome extension) or on your Mac (desktop app). Audio stays in local memory.
2. **We transcribe and format** — Audio is sent to Groq for transcription, then the transcript is sent to OpenAI for formatting. Both transmissions use encrypted HTTPS. Neither Groq nor OpenAI stores your data — both process transiently and discard.
3. **Text appears** — The formatted text is inserted directly into your app (Gmail, Slack, Notion, etc.).

### What we store

| Data | Stored? | Where | Retention |
| --- | --- | --- | --- |
| Audio recordings | Never | — | Discarded after transcription |
| Transcripts | Never | — | Discarded after formatting |
| Formatted text | Never on servers. In-memory on Mac app (last 50, cleared on quit) | Local device only | Session only |
| Glossary entries | Yes | Your Glossary is stored in our cloud database (Supabase), isolated to your account with row-level security and encrypted in transit and at rest, so it syncs across your devices. | Until you delete them |
| Account info | Yes | Supabase | Until you delete your account |
| Usage counts | Yes | Supabase | Aggregate only, no content |
| Anonymous analytics (Chrome + Mac) | Yes | PostHog | No transcripts, no content, no IP addresses |

### Our data processors

- **Groq** — Transcription. Receives audio, returns text. Zero data retention. Based in the US.
- **OpenAI** — Formatting. Receives transcript text + Glossary context, returns formatted text. API usage with data retention disabled. Based in the US.
- **Supabase** — Authentication, Glossary storage, usage tracking. Row-level security enforced. Based in the US (AWS).
- **PostHog** — Anonymous product analytics for the Chrome extension and Mac app. Receives event names and metadata only (sign-in completed, recording completed, etc.). Never receives transcripts, formatted text, Glossary content, or IP addresses. Users can opt out in Settings. Based in the US.
- **Google** — OAuth authentication only. We use Google Sign-In. Google does not receive any dictation data.
- **Vercel** — Hosts the Rubil backend (API routes) and website. Processes API requests transiently. No persistent storage of user content. Based in the US.

### What we will never do

- **Never store your audio.** Recordings are processed in memory and discarded. There are no voice files on our servers, ever.
- **Never store your transcripts.** The text version of what you said is used for formatting and then discarded. We don't keep transcript logs.
- **Never use your data to train AI.** Your dictations are not used to fine-tune, train, or improve any AI model — ours or anyone else's.
- **Never sell your data.** We have no advertising, no data brokerage, no third-party data sharing beyond the processors named above.
- **Never track what you type.** We don't capture URLs, page content, window titles, or anything from the apps you use. The only context we read is which platform you're on (Gmail vs Slack vs Notion) so we can format correctly.

---

## FAQ

**How is Rubil different from other voice dictation tools?**
Most voice dictation tools do one of two things. They either transcribe exactly what you said (including the mess), or they rewrite it in the AI's voice. Rubil does neither. It formats your words for the platform you're writing in. The same voice input becomes a structured email in Gmail, a concise message in Slack, or a formatted ticket in Jira. It also knows your world: your team's names, your @handles, your acronyms, your jargon. No other voice tool offers platform-specific formatting or a personal Glossary this deep. And it's a Chrome extension, so there's nothing to install on your desktop.

**Does Rubil work with Google Docs, Slack, and Gmail?**
Yes. Rubil is a voice dictation Chrome extension that works in any text input on the web, including Gmail, Slack, Notion, Google Docs, Jira, Linear, Asana, Salesforce, HubSpot, Confluence, and more.

**Does Rubil add punctuation and formatting automatically?**
Yes. Unlike basic speech to text tools, Rubil automatically adds punctuation, bullet points, and proper formatting based on the platform you're writing in. Voice to text with formatting is what Rubil was built for.

**How does the personal Glossary work?**
You teach Rubil the names, acronyms, jargon, and shortcuts you use every day. You add entries by speaking naturally ("Sarah Chen is our design lead, her Slack handle is @sarah.chen") and Rubil automatically categorizes and stores them. When you dictate, it applies your Glossary in real time, correcting transcription errors, inserting @handles on the right platform, expanding acronyms in formal emails but keeping them short in Slack, and preserving exact spelling for brand names and technical terms. The more words you add, the better Rubil matches your vocabulary. Your Glossary is stored in our cloud database (Supabase), isolated to your account with row-level security and encrypted in transit and at rest, so it syncs across your devices.

**Do I need to train Rubil or set it up?**
No. Rubil works out of the box with zero configuration. Install the Chrome extension, click any text field, and start talking. It automatically detects whether you're in Gmail, Slack, Jira, or any other web app, and formats your speech accordingly. You can optionally build a personal Glossary over time to teach it your team's names, acronyms, and jargon, but it's not required. Most users start dictating within 30 seconds of installing.

**Does Rubil work in languages other than English?**
Yes. Rubil supports 50+ languages including Spanish, Portuguese, French, German, Japanese, Chinese, Korean, Arabic, Hindi, and many more. Dictate in your language and get the same formatting, punctuation, and Glossary features.

**Is my voice data private and secure?**
Absolutely. Rubil is private speech to text software. We never store your voice recordings or transcripts. Your Glossary is stored in our cloud database (Supabase), isolated to your account with row-level security and encrypted in transit and at rest, so it syncs across your devices. Your data is processed transiently by our AI providers and never used for model training.

**What happens when I run out of free words?**
Every new account starts with 7 days of unlimited Pro, beginning on the first dictation. After that the free plan is 1,000 words a week on a rolling seven-day window, so words free up again as older days age out. The dashboard shows the date they come back. You can also upgrade to Pro for $5 a month for unlimited words.

**Can I switch between Free and Pro anytime?**
Yes. You can upgrade or downgrade anytime with one click. No commitment.

**Do you offer refunds?**
If you're not satisfied with Pro in your first 30 days, we'll refund you in full. No questions asked.

---

## Blog Posts

### Voice dictation for sales reps: clear the follow-up pile without changing what you promised

URL: https://www.rubil.io/blog/dictation-for-sales

A sales rep's calendar leaves two minutes between calls, and in those two minutes they are supposed to write the follow-up email, log what the prospect said, brief the solutions engineer in Slack, and update the close date. So it gets deferred, and by end of day there are six calls of context to reconstruct and the note reads "good call, sending pricing." Generic voice dictation should fix this but breaks on the vocabulary that matters: the champion's name spelled three ways, "SOC 2" rendered as "sock two," a product's Growth tier written in lowercase like a generic noun.

Rubil is voice dictation built for people who write across many apps all day. It runs as a Chrome extension for browser work and a Mac desktop app for native apps, with the same voice input producing different output for Gmail vs Slack vs Jira. It formats your words. It never rewrites or generates them. Your ghostwriter, not your ghost-thinker. For a sales rep that carries specific weight: when an AI generates the final wording of a follow-up it can put words in your mouth that you never said, and in sales the words in your follow-up are the ones the buyer holds you to. A rewrite that turns "I'll check whether we can do that" into "we can do that" has changed a commitment.

Post-call example: a messy voicemail-style dictation ("hey um Priya wanted to follow up on on the call today so we we talked about the Q3 rollout and you mentioned the the security review is the blocker um I'm going to loop in Marcos on our side he he owns the SOC 2 questionnaire and uh I said I'd send over pricing for the fifty seat Growth tier by Thursday") becomes a formatted Gmail draft with a greeting, paragraph breaks, correct prospect name (Priya), correct colleague name (Marcos), correct acronym casing (SOC 2), correct product tier capitalization (Growth), and the seat count as a number (50-seat). Nothing is added: no invented pleasantry, no manufactured urgency, no closing line the rep did not say. Rubil also distinguishes a fresh compose from a reply, so replying in an existing thread produces shorter output without the greeting. The same dictation dropped into Slack lands as a single channel-shaped line with no greeting or sign-off.

The apps with purpose-built formatting are the ones sales work runs through: Gmail, Outlook, and other webmail, Slack channels and DMs (handled differently from each other), Microsoft Teams, LinkedIn posts and comments, Google Docs and Notion for account plans and proposals, Jira and Linear, and AI chat tools.

Where Rubil does not have a dedicated format yet, stated plainly: there is no purpose-built formatting profile for Salesforce, HubSpot, Outreach, Apollo, or Gong. Dictation works in those apps because Rubil works in any text field in the browser, and the Glossary still applies so names, acronyms, and product tiers land correctly, but output falls back to document-style formatting rather than something shaped for a CRM activity note. LinkedIn posts and comments each get their own formatting; LinkedIn messages and InMail do not have their own profile yet and are treated as posts. There is no Windows app yet and no iOS or Android app yet, both on the roadmap, so a Windows-based sales floor can use the Chrome extension for browser work and nothing for native desktop apps.

The Glossary is the personal vocabulary layer. Four categories: People (champions, economic buyers, procurement contacts, internal counterparts across open opportunities), Acronyms (MEDDIC, ARR, POC, MSA, SOW, QBR, with platform-aware behavior so POC can stay POC in a Slack message to an SE and expand to "proof of concept" in an email to a buyer), Terms (strict casing so product tiers like Starter, Growth, and Enterprise stay capitalized as product names), and Expansions (short triggers that drop a whole block, such as a post-demo follow-up structure, a mutual action plan outline, a standard security-review response, a renewal recap, or a pricing block). No entry limit. Entries sync between Chrome and Mac. Rubil auto-suggests entries when you correct the same word twice.

Why not the alternatives: Wispr Flow, Willow Voice, and Aqua Voice use AI to rewrite voice input into a more polished version, which is a real trade with real upside if you want the tool to improve your wording. For sales the risk is that an AI-rewritten follow-up changes a commitment made verbally, and the buyer holds the written version.

Quick tips for the first week: add the ten people you email most and the eight acronyms you say every day, then let auto-suggestions handle the rest. Dictate the CRM note before the email, since it is the writing reps skip most often and the writing that costs the most later.

Privacy for named accounts and deal terms: audio is processed transiently through Groq and discarded after transcription. Transcripts are sent to OpenAI for formatting and are not stored on Rubil's servers. The Glossary is encrypted locally and in the cloud. Rubil never watches your screen or takes screenshots; it detects which app you are in from the browser URL on Chrome or the app bundle ID on Mac, not from anything you write. Every data processor is named on the privacy page. The free tier is 7 days of unlimited Pro, then 1,000 words a week, always on the same engine paid users get.

### Voice dictation for consultants: get billable time back without losing your voice

URL: https://www.rubil.io/blog/dictation-for-consultants

A consultant's calendar is a wall of client calls, and the space between them is writing: status emails to three engagements, Slack threads for the internal team, deliverable edits in a Google Doc, and follow-up notes to partners waiting on answers. Every hour spent typing is an hour not billed. Generic voice dictation should fix this but breaks on names and acronyms: Vercel becomes Versal, Project Odyssey becomes Odysseus, EBITDA becomes "a bit dot," and the correction pass eats the time it was supposed to save.

Rubil is voice dictation built for people who write across many platforms all day. It runs as a Chrome extension for browser work and as a Mac desktop app for native apps, with the same voice input producing different output for Gmail vs Slack vs Jira. It formats your words. It never rewrites or generates them. Your ghostwriter, not your ghost-thinker. That distinction matters more for consultants than for almost any other role, because a client can tell when an email was written by an AI, and it changes what they think of you.

Tuesday morning example: dictating a client recap into Gmail. Rubil turns a messy voicemail-style dictation ("Hi Priya um wanted to close the loop on the the OKR workstream we agreed to move the the North Star metric review to next Thursday and I I asked Marcos to pull the the APAC numbers from the Q3 dashboard before then also uh regarding the pricing tier analysis I think we can we can push the draft to Friday if you're okay with that let me know") into a formatted email with a greeting, paragraph breaks, correct client name (Priya), correct acronym (OKR, APAC), correct colleague name (Marcos), and correct project workstream name. The same words in Slack would land as a shorter chatty version without greeting and closing; the same words in Notion would land as document-style prose. Rubil detects the app you are in and formats accordingly.

Why consulting communication is different: a consultant is on three or four engagements at once, each with its own private language of client codenames, industry acronyms, contact names, format preferences, and account-lead tone. Rubil is different because you teach it your world instead of hoping it guesses.

The Glossary is the personal vocabulary layer that makes voice dictation viable for client-facing work. Four categories: People (partners, engagement managers, client contacts, cross-functional counterparts), Acronyms (EBITDA, SG&A, TAM, and the ten others specific to your practice, with platform-aware behavior so OKR can stay short in Slack and expand to "Objectives and Key Results" in a client email), Terms (strict spellings and casing for product names and technical terms), and Expansions (short triggers that drop a whole scaffold, e.g. "weekly update" inserts your standard weekly client update block). No entry limit. Every entry syncs between Chrome and Mac. Rubil auto-suggests entries when you correct the same word more than once.

Chrome extension covers Gmail, Google Docs, Slack in the browser, Notion, Confluence, Jira, Linear, Asana, and any other web app. The Mac app covers the native Slack client, Apple Mail, Microsoft Word, Notes, and any other Mac app with a text field. Same Glossary, same formatting engine, same backend.

Why not the alternatives: Wispr Flow, Willow Voice, and Aqua Voice take the opposite approach and use AI to rewrite voice input into a polished version. For a consultant whose credibility is tied to how they sound in writing, that is a different bet.

Quick tips for the first week: add the ten people you email or Slack the most, three per current engagement, and the five acronyms you use every day. That gets you to about 80 percent accuracy on your typical week.

Privacy for client-confidential work: audio is processed transiently through Groq and discarded after transcription. Transcripts are sent to OpenAI for formatting and are not stored on Rubil's servers. The Glossary is encrypted locally and in the cloud. Rubil never watches your screen or takes screenshots; it detects the app you are in from the URL in Chrome or the app bundle ID on Mac, not from your content. Every data processor is named on the privacy page. Consultants working under NDA or on sensitive engagements can adopt Rubil without introducing a new place their work is stored.

### Dragon NaturallySpeaking alternatives for 2026

URL: https://www.rubil.io/vs/dragon-naturallyspeaking-alternatives

Dragon NaturallySpeaking earned its reputation over decades, but its users are shopping for alternatives because it costs several hundred dollars for a one-time Windows license, asks you to set up and correct a voice profile before it reaches peak accuracy, and dropped its Mac version in 2018. Since Microsoft acquired Nuance in 2022, product development has slowed. The bigger shift is that Dragon transcribes and leaves formatting to you, while modern tools transcribe and format, so the same spoken input becomes a structured email in Gmail, a short message in Slack, and document prose in Notion.

The best alternatives are Rubil, Superwhisper, Wispr Flow, Willow Voice, and Aqua Voice. All use modern AI transcription that works from the first word with no voice training, cost a fraction of Dragon's price, and run on Mac. Rubil formats your words and never rewrites or generates them, available as a Chrome extension and Mac app at $5/month (or $39/year founding annual) with a free tier of 7 days unlimited then 1,000 words a week, and it is the only tool on the list with platform-specific formatting across 20+ apps plus a Glossary for names, @handles, acronyms with platform-aware expand or keep rules, and expansion triggers. Superwhisper is the closest to Dragon's local-processing model: Mac-only, offline-capable, with nothing sent to the cloud. Wispr Flow is polished and cross-platform (Mac, Windows, iOS) at $15/month and leans into AI generation with its Command Mode. Willow Voice is $15/month and its AI Mode generates polished messages from rough notes. Aqua Voice is fast and affordable at $8/month using a floating-box-then-paste model tuned for technical vocabulary. Apple Dictation and Windows Voice Access are the free built-in baseline, honest about their formatting and vocabulary limits.

When Dragon is still the better choice: full voice command-and-control of the computer (a real accessibility need none of the modern formatting tools replace; Talon is the open-source power-user option), fully offline or on-premises requirements in regulated environments, and Dragon Medical One for clinical documentation, which is its own category.

Privacy: Rubil processes audio transiently through Groq and discards it, never storing audio or transcripts; the Glossary is stored in Supabase encrypted in transit and at rest; every processor (Groq, OpenAI, Supabase) is named on the privacy page. Superwhisper's local, on-device processing is a real privacy advantage for anyone who valued Dragon's offline model. Wispr Flow, Willow Voice, and Aqua Voice process audio in the cloud, each under its own privacy policy.

### Best Willow Voice Alternative for Knowledge Workers (2026)

URL: https://www.rubil.io/vs/willow-voice-alternative

Willow Voice and Rubil are both voice dictation tools for professionals. They both convert speech to text and both claim context awareness. The disagreement is what should happen to your words after you speak them. Willow Voice has a feature called AI Mode that takes brief verbal notes and transforms them into polished, complete messages, plus a "smart writing style memory" that learns your tone so the AI's output sounds more like you. Rubil formats your words instead. It fixes grammar, removes filler words, adds punctuation, and structures the text for the platform you are in, but it never rewrites what you said, never generates new content, and never swaps your phrasing for what an LLM thinks is better. Because Rubil formats your words rather than generating new ones, it cannot hallucinate sentences you never said.

Platform coverage: Willow covers Mac, Windows, and iOS. Rubil runs on Chrome as the primary extension and Mac as a desktop app, with no Windows or iOS yet. Rubil's Chrome extension is browser-native and purpose-built for web apps like Gmail, Slack, Notion, Google Docs, Teams, Jira, and Linear, while Willow has no Chrome extension and uses its desktop app to dictate into browser tabs.

Privacy: Rubil's architecture is built around transient processing with no voice files, no transcript history, and no recordings saved on servers. It stores only encrypted Glossary entries and aggregate usage metrics, and it never reads your screen. Willow treats transcript history as a feature, storing past dictations so they are searchable, with an optional Privacy Mode to disable it, and offers SOC 2 Type II compliance at the Team and Enterprise level.

Personalization: Willow's smart writing style memory adapts its tone across app categories so the AI's rewrites get closer to how you sound, and it has a personal dictionary for custom vocabulary. Rubil's Glossary ensures accuracy and consistency instead of adapting tone, locking in colleagues' names and @handles, acronyms that expand in email but stay short in Slack, and project names with exact casing.

Pricing: Willow Voice is $15/month or $12/month annual, with Team at $12/user/month ($10 annual) and custom Enterprise pricing. Rubil is $5/month or $39/year on the founding annual plan. Willow's free tier is unlimited in volume but runs a weaker speech-to-text model (Frontier Mini) with limited personalization and 20 Scribe uses per week, while Rubil's free tier is 7 days of unlimited Pro and then 1,000 words a week, always on the same engine paid users get. The trade is volume against quality.

When Willow is the better choice: anyone who wants AI to transform rough notes into polished messages, anyone who needs Windows or iOS support, organizations that require SOC 2 Type II for procurement, anyone who wants offline dictation, and anyone who prefers a tool that learns and matches their writing style over time.

When Rubil is the better choice: knowledge workers writing across Gmail, Slack, Notion, Google Docs, and Teams all day who want their words formatted for each platform without being rewritten. Inline dictation removes the paste step, platform-specific formatting adapts output per app, and the Glossary keeps names, @mentions, and acronyms correct. If your messages need to sound like you wrote them rather than like AI wrote them for you, that is the signal.

### Best Wispr Flow Alternative for Knowledge Workers (2026)

URL: https://www.rubil.io/vs/wispr-flow-alternative

Wispr Flow and Rubil are both voice dictation tools, but they make opposite bets about what should happen to your words. Wispr Flow is a well-built, cross-platform product with decent accuracy and an expanding set of AI editing commands — its Command Mode lets you say "make this more professional" and the AI rewrites your sentence. Rubil goes the other direction. It cleans up grammar, adds punctuation, and structures your speech into paragraphs, but it never rewrites what you said. The hedging, emphasis, and casualness that make you sound human stay intact. Rubil fixes how it reads; it never touches what it says.

Platform formatting: Wispr Flow offers four style categories (email, work messaging, personal messaging, and other). Rubil formats per app — Gmail gets a greeting and sign-off, a Slack DM stays short and direct, a Notion page gets structured sections — across 20+ platforms including Gmail, Slack, Notion, Linear, GitHub, Jira, and ClickUp. You don't configure any of this; Rubil detects where you are and adapts automatically.

Privacy: Wispr Flow's context-aware features read text from your screen and process it on their servers to improve formatting accuracy. This is enabled by default and can be toggled off in settings. Rubil's architecture doesn't capture screen content at all. Audio is processed transiently and discarded — no screenshots, no transcripts stored, no voice recordings on servers. Your Glossary is stored in our cloud database (Supabase), isolated to your account with row-level security and encrypted in transit and at rest, so it syncs across your devices. Every data processor is named on the privacy page.

Personalization: Wispr Flow has a flat custom dictionary. Rubil has a structured Glossary built from the words you teach it, with phonetic hints, aliases, and platform-specific handles.

Pricing: Wispr Flow is $15/month with roughly 8,000 free words per month. Rubil is $5/month, or $39/year on the founding annual plan, with a free tier of 7 days unlimited and then 1,000 words a week. Wispr Flow ships as a desktop app that works system-wide including browsers; Rubil ships as a Chrome extension plus a Mac desktop app, covering both browser-native and desktop use.

When Wispr Flow is the better choice: anyone who wants AI to actively help them write rather than just clean up dictation — rewriting messages, expanding short prompts into longer drafts, adjusting tone, summarizing, or translating. Those are real features Rubil deliberately does not compete on.

When Rubil is the better choice: knowledge workers who spend the day across Gmail, Slack, Teams, Notion, and Docs and want to dictate their own words, formatted correctly for each app, without AI rewriting them. For people tired of AI that speaks for them and want a tool that keeps up with how fast they think without changing what they think.

### Best Aqua Voice Alternative for Chrome and Mac (2026)

URL: https://www.rubil.io/vs/aqua-voice-alternative

Rubil and Aqua Voice are both voice dictation tools for professionals. The core difference is philosophy. Aqua Voice is direct about what it does: its homepage promises to "refine your words as you talk," and the product is designed to take your raw speech and add an AI layer on top — smoother sentences, cleaner structure, tighter phrasing. Aqua also offers natural language editing commands like "make this a list" or "rewrite this paragraph."

Rubil does something different. Rubil formats your words. It fixes grammar, adds punctuation, and structures your text for the platform you're writing in. But it never rewrites what you said. No synonym swaps, no sentence restructuring, no AI-generated phrasing. Your vocabulary, your cadence, your way of saying things.

Platform coverage: Aqua covers Mac, Windows, iOS, and a Chrome extension. Rubil runs on Chrome (as the primary extension) and Mac (as a desktop app). No Windows or iOS yet. Rubil's Chrome extension is the primary product and is purpose-built for web apps like Gmail, Slack, Notion, Google Docs, Teams, Linear, and Jira.

Privacy: Rubil's architecture is built around transient processing. No voice files, no transcript history, no recordings saved on servers. Rubil stores only encrypted Glossary entries and aggregate usage metrics. Rubil never reads your screen — it knows you're in Gmail by reading the URL or focused app. Aqua processes audio through its proprietary Avalon model and does not store audio after processing, but Aqua does read your screen content to improve transcription accuracy.

How they work: With Rubil, formatted text appears inline at your cursor — no intermediate window, no copy-paste. With Aqua, you dictate into a floating box that sits on top of your screen, review the AI-refined output, then paste it into your destination app. Rubil's formatting changes automatically per platform (email vs Slack vs Notion); Aqua produces one output regardless of where you paste it.

Personalization: You teach Rubil's Glossary your social graph: people's names, @mentions, acronyms, technical terms, expansions for email and address. The Glossary has no entry limit and suggests entries based on your corrections. Aqua's Avalon model is tuned for technical vocabulary (framework names, programming terms, CLI commands), and Aqua's Pro plan includes an 800-term custom dictionary.

Pricing: Aqua Voice Pro is $8/month or $96/year. Rubil Pro is $5/month or $39/year on the founding annual plan. Aqua's free tier is a one-time 1,000-word allotment to evaluate the product. Rubil's free tier is 7 days of unlimited Pro and then 1,000 words every week, forever. Aqua offers a 70% student discount on annual with a .edu email; Rubil does not currently offer a student discount.

When Aqua Voice is the better choice: developers and technical writers who need its developer-tuned vocabulary, anyone who needs a native Windows desktop app or iOS app, anyone who wants AI to actively refine and restructure text (not just format it), and anyone who prefers a preview window before text lands in the destination app.

When Rubil is the better choice: knowledge workers writing across Gmail, Slack, Notion, Google Docs, and Teams all day who want their words formatted for each platform without AI rewriting them. Inline dictation removes the paste step. Platform-specific formatting adapts output per app. The Glossary holds the names, acronyms, and terms you teach it. The Chrome extension means one click to be live in every web app you use.

### Voice dictation for Gmail: how to dictate emails that are ready to send

URL: https://www.rubil.io/blog/voice-dictation-gmail

Gmail has no built-in voice dictation, and most voice tools dump raw text into the compose window with no greeting, paragraphs, or sign-off, so you spend minutes reformatting. Rubil adds voice dictation directly inside Gmail and formats the output like an actual email. It is available as a Chrome extension and a Mac desktop app that share the same backend, the same Glossary, and the same formatting.

Open Gmail, click compose or hit reply, and start talking. Rubil's mic button appears next to the text field, and formatted text lands directly in the compose window with no floating box and no paste step. A rambling spoken sentence becomes a structured email: greeting at the top, proper paragraph breaks, punctuation and casing handled, and a professional sign-off at the bottom.

What makes Gmail formatting different is that Rubil detects the app you are in and applies email-specific rules. The same spoken words produce a structured email in Gmail, a concise message with no greeting or sign-off in Slack, and document prose in Notion. You do not configure this or pick a mode. Rubil adjusts automatically every time.

Chrome extension vs Mac app: the Chrome extension runs inside the browser tab where Gmail lives and drops text inline. The Mac app works when you have Gmail open in any browser by detecting the focused field, and it also works across desktop apps like Apple Mail, Slack, and WhatsApp. Both share the same Glossary and formatting engine.

Why not the alternatives: Google's built-in voice typing (Ctrl+Shift+S) only works in Google Docs, not Gmail, and adds no formatting. Voice In transcribes but applies zero formatting intelligence, producing the same raw text everywhere. Voicy offers a floating mic button in Gmail but its output is identical regardless of platform. Rubil skips the manual reformatting step entirely.

Quick tips: set up your Glossary early with the names, terms, and acronyms you use most. Rubil works in reply threads, new compose windows, and forwarded messages. You never need to say "period" or "comma" because punctuation and structure are handled automatically. Audio is processed transiently and never stored, and no transcripts are kept. Your Glossary is stored in our cloud database (Supabase), isolated to your account with row-level security and encrypted in transit and at rest, so it syncs across your devices. 50+ languages are supported. Glossary expansions let you save blocks of text like your address, signature, or booking link and trigger them with a short phrase while dictating.

### Voice dictation for Slack: how to get every name, @mention, and acronym right

URL: https://www.rubil.io/blog/voice-dictation-slack

Slack has no built-in voice dictation, and generic transcription tools get your vocabulary wrong: names get the common spelling, acronyms turn into syllables, and project names lose their casing. Fixing a few words per message cancels out the time dictation was supposed to save. Rubil adds voice dictation to Slack with chat-appropriate formatting and a Glossary that teaches it your team's names, acronyms, and terminology so every word lands right the first time. It is available as a Chrome extension for Slack in the browser and a Mac desktop app for the native Slack client, both sharing the same Glossary and formatting engine.

On Chrome, Rubil's mic button appears in the Slack message input; click it, talk, and formatted text lands ready to send. On Mac, Rubil detects the Slack desktop client's focused text field and injects formatted text directly. Output is Slack-specific: concise, punctuated, with chat-appropriate line breaks and no greeting or sign-off. The same spoken words produce a structured email in Gmail or document prose in Notion, because Rubil detects the app and formats accordingly.

The Glossary matters most in Slack. It corrects names, acronyms, and product terms and converts @mentions to the right handles. A message like "hey Kevin can you check the a pack numbers for project helix and loop in Christina from the okay are team" becomes "Hey @kevin.jones, can you check the APAC numbers for Project Helix and loop in @christina.lowes from the OKR team?" The Glossary also handles platform-aware behavior: an acronym like OKR stays as OKR in a Slack message but can expand to "Objectives and Key Results" in a client-facing email. Rubil auto-suggests entries when it notices you correcting the same word repeatedly. Expansions trigger entire blocks of text from a short phrase, which fits Slack's many templated messages: standup updates, sprint goals, a calendar link, or a QA checklist.

Why not the alternatives: Voice In is Chrome-only and has no personalization, so it cannot help in the Slack desktop app and guesses every name. Voicy transcribes accurately but has no Glossary and no platform-specific formatting, so output is identical in Slack and Gmail. Speechify offers basic transcription with punctuation cleanup but no vocabulary learning and no Slack-specific formatting. Quick tips: start with 10-15 Glossary entries for names, acronyms, and project names. Audio is processed transiently and never stored, and no transcripts are kept. Your Glossary is stored in our cloud database (Supabase), isolated to your account with row-level security and encrypted in transit and at rest, so it syncs across your devices. 50+ languages are supported.

### Voice dictation for Notion: how to dictate structured pages that don't need editing

URL: https://www.rubil.io/blog/voice-dictation-notion

https://www.rubil.io/voice-dictation-notion — Dictating into Notion with generic voice tools dumps flat prose into a structured app, so you spend more time adding headers, bullets, and sections by hand than the dictation saved. This guide covers why Notion dictation is harder than Slack or Gmail, and how Rubil detects Notion and shapes speech into real blocks, headers when you intend them, lists when you describe items, clean prose when you talk in full thoughts, without rewriting your words. Includes concrete examples for meeting notes, PRDs, and journal entries, plus the Glossary's role for team names and terms and coverage across the Chrome extension and Mac app.

### Voice Dictation in Google Docs: 2026 Guide

URL: https://www.rubil.io/blog/voice-dictation-google-docs

Unlike Gmail, Slack, and Notion, Google Docs ships a free built-in Voice typing tool, so this guide is honest about it and wins by being the better answer rather than pretending the native option does not exist. To use the built-in tool, open a Doc in Chrome, click Tools then Voice typing (or press Cmd+Shift+S on Mac, Ctrl+Shift+S on Windows), click the microphone, and talk. You say punctuation out loud ("period," "comma," "new paragraph"), and it supports basic formatting and navigation voice commands that work inconsistently by language and browser version. It supports 100+ languages, needs an internet connection because audio is processed on Google's servers, and is free with any Google account.

When built-in Voice typing is enough: short documents, notes for yourself, brainstorming, occasional use, and zero budget. If that is all you need, the built-in tool is right for you.

Where it stops: it is Docs-only and disappears the moment you switch to Gmail, Slack, Notion, or any other app; it produces raw text with no formatting intelligence and no awareness of document type; it has no personal vocabulary, so names, acronyms, and jargon come out wrong ("Marcos" becomes "Marcus," "APAC" becomes "a pack"); you have to speak your punctuation; and browser support is limited (Chrome is the most reliable, Google lists Edge and Safari too but performance varies, and Firefox is not supported).

Rubil in Google Docs produces clean document prose with correct punctuation and casing from natural speech, no spoken punctuation required. The same voice input formats differently per app (structured email in Gmail, concise message in Slack, document prose in Notion) because Rubil detects the app you are in. The Glossary teaches Rubil your colleagues' names and @handles, context-aware acronyms, project names with strict casing, and expansion triggers, all syncing across Chrome and Mac. Rubil formats your words and never rewrites or generates them. Price is $5/month or $39/year on the founding annual plan, with a free tier of 7 days unlimited and then 1,000 words a week. Alternatives: Voice In is a Chrome extension with basic transcription but no platform-specific formatting or personal vocabulary; Aqua Voice is a Mac and Windows desktop app with strong technical vocabulary that uses a floating-box paste model. On privacy, Google's Voice typing processes audio on Google's servers per their policies, while Rubil's transient architecture sends audio to Groq (discarded immediately) and transcripts to OpenAI (not stored), with the Glossary in Supabase encrypted in transit and at rest and every processor named on the privacy page.

---

## Contact

- Website: https://www.rubil.io
- Support: support@rubil.io
- Blog: https://www.rubil.io/blog