Short answer: Dictation lets you produce project documentation by speaking instead of typing, capturing READMEs, design docs, and runbooks at speaking speed of 130 to 150 words per minute. Voice Keyboard Pro types your words at the cursor in any app, so you draft fast and shape the structure afterward.
Nearly every team has a documentation problem, and nearly every team knows it. The README is three months out of date. The architecture lives in one engineer's head. The runbook for the 3 a.m. incident is a Slack thread someone half-remembers. It is not that people do not care. It is that documentation is the work that always loses to other work, and the reason it loses is almost always the same: writing it is slow, the blank page is intimidating, and the payoff is deferred.
Dictation attacks the two biggest causes of documentation debt directly. It makes writing roughly twice as fast, and it lowers the activation energy that keeps you staring at an empty file. When the cost of getting the first draft down drops far enough, "I'll document it later" stops being the default. This guide covers why documentation gets dreaded, how speaking changes the equation, which documents lend themselves to voice, and a practical drafting workflow you can adopt on a Mac today.
Why Documentation Gets Dreaded and Deferred
If you want to fix a habit, it helps to understand why it is hard in the first place. Documentation suffers from a few specific frictions.
The blank page. Starting is the hardest part. A typed first sentence is a deliberate, slow act, and when you are not sure how to begin, the cursor blinks at you while you stall. Speaking sidesteps this entirely, because talking through what something does is something you already know how to do without preparing.
Typing speed caps your output. A good typist runs 80 to 100 words per minute. People speak at 130 to 150 words per minute comfortably. A design doc that takes forty minutes to type might take twenty to talk through. That difference is not a luxury when documentation is competing for time against shipping features.
Documentation is mostly prose, which is exactly what typing is worst at. Code is dense with symbols where the keyboard wins. Documentation is paragraphs of explanation, the natural-language format where voice has its biggest advantage. You are using your slowest input method for the task most suited to your fastest one.
The payoff is delayed and invisible. Nobody thanks you for the runbook until the night it saves the on-call engineer. Because the reward is deferred, anything that raises the upfront cost makes the task easy to skip. Lower the cost and the calculus shifts.
How Speaking Changes the Equation
Voice does not just make documentation faster. It changes how it feels to produce, and that change is what actually moves the needle on whether the docs get written.
The first shift is momentum. When you speak, you do not stop to perfect each sentence, so you reach the end of a thought before your inner critic can stall you. You get a complete, messy draft, and a messy draft is infinitely easier to improve than a blank page. The second shift is tone. Documentation written by voice tends to sound like a person explaining something to a colleague, which is exactly the register good docs should have. Typed docs drift toward stiff, over-formal prose; spoken docs stay conversational and clear because that is how explanation naturally comes out of your mouth.
You do not have a documentation problem. You have a "starting the document" problem. Speaking solves it.
The third shift is volume. Because each paragraph costs less, you write the parts you would otherwise cut. The caveat about the flaky dependency, the reason you chose this database, the gotcha that will bite the next person, all the context that makes documentation worth reading rather than just present.
What You Can Dictate
Almost every kind of project documentation is mostly prose, which means almost all of it benefits from voice. A few of the highest-value targets:
- READMEs. What the project is, how to run it, how to contribute. Talk through it the way you would for a new hire.
- Architecture and design docs. The reasoning behind the system, the trade-offs you weighed, the options you rejected and why. This is the documentation with the highest long-term value and it is almost entirely explanation.
- Runbooks and incident playbooks. Step-by-step recovery procedures. Speak them while the memory of the last incident is fresh, when typing them out feels like too much.
- Architecture Decision Records. Short, structured notes capturing why a decision was made. Dictate the context and consequences; the decision itself is usually one line.
- Onboarding guides. The tribal knowledge that lives in people's heads. Voice is ideal because you are essentially narrating what you would tell someone over their shoulder.
- Release notes and changelogs. A real description of what changed and why it matters to users, instead of a terse list of commit subjects.
The same speak-the-prose principle that powers dictating code comments applies here, just at a larger scale. The comment explains one function; the design doc explains the whole system. Both are explanation, and explanation is what voice does best.
Setting Up on Mac
Voice Keyboard Pro is a native macOS menu bar app that types wherever your cursor sits. It works in your text editor, in a Markdown file, in a Confluence or Notion page in the browser, in a Google Doc, in a pull request description. The flow never changes:
- Open the document and place your cursor where you want to write.
- Hold your dictation hotkey.
- Talk through the section as if explaining it to a teammate.
- Release. The text appears at your cursor in under a second.
- Move to the next section, or switch to the keyboard to add structure.
Because it is system-wide, your documentation tooling does not have to change. Whatever app your team already writes docs in, the same hotkey works there. And the privacy posture is straightforward: Voice Keyboard Pro's transcription engine converts your speech to text, and the company stores only operational pings, never your audio or the content of what you dictate. Internal design details and unreleased plans stay between you and your document.
A Documentation-First Workflow
The technique that makes voice documentation work is to separate drafting from structuring. Do not try to dictate a perfectly formatted document in one pass. Do it in two.
Pass one: talk it out
Open the file, put your cursor under each heading in turn, and just explain. Do not worry about formatting, transitions, or polish. If you are documenting a service, talk through what it does, what it talks to, how to run it locally, what fails first, and what you would tell someone debugging it at midnight. Let it be a wall of text. The goal of pass one is to externalize everything you know while it is in your head, at the speed you can speak it.
Pass two: shape it
Now switch to the keyboard and edit. Add headings and subheadings, break the wall of text into sections, drop in fenced code blocks and links, cut the repetition, tighten the wording. This is fast because editing existing text is far easier than producing it. You are sculpting, not quarrying. Most people find a document that would have taken an hour to type straight through is done in half the time when drafted by voice and edited by hand.
This draft-then-shape rhythm is the same one that makes voice effective for meeting notes and other long-form writing: capture fast and loose, then refine. The structure is cheap to add once the substance exists.
Capturing Decisions as They Happen
Some of the best documentation is written during meetings, not after them. Voice Keyboard Pro includes a Meeting Mode with speaker detection and AI notes, plus calendar meeting detection so it knows when you are in a discussion. The practical value for documentation is that the decisions, the reasoning, and the action items get captured while they are being made, rather than reconstructed days later from memory. A design review that produces a written ADR before everyone leaves the room is documentation that actually reflects reality. For teams that live in meetings, this turns the meeting itself into a documentation source rather than a thing you have to document afterward.
Keeping Technical Terms Accurate
Documentation is dense with names that general transcription has never encountered: your service names, framework names, internal acronyms, the database you chose. Spoken aloud, these can come back wrong. Voice Keyboard Pro's Smart Vocabulary is a personal dictionary with replacement rules that fixes this. You add the terms your team uses, including how you pronounce them, and they render correctly every time. Spend ten minutes seeding it with the proper nouns from your stack and the accuracy problem largely disappears. This is the step that separates people who try voice once and quit from people who keep using it: invest in the vocabulary early and the friction goes away.
Won't It Read Like a Transcript?
The most common worry about dictated documentation is that it will sound rambling, like a transcript of someone thinking out loud rather than a finished document. It is a fair concern, and the answer is the two-pass workflow above. Pass one is allowed to ramble, because nobody but you ever sees it. Pass two is where it becomes a document. The editing pass is short precisely because the substance is already there: you are cutting filler, merging repeated points, and adding structure, not composing from scratch.
In practice, dictated-then-edited documentation often reads better than typed documentation, not worse. Typed prose tends to be over-engineered, with writers polishing each sentence into something stiff and formal. Spoken prose carries the natural cadence of explanation, the way a senior engineer talks a junior through a system. After a light edit, that voice is what survives, and it is exactly the clear, direct tone that makes documentation pleasant to read. The trick is simply to never ship pass one. Treat the dictated draft as raw material, give it five minutes of shaping, and the transcript worry disappears.
It also helps to dictate in logical chunks rather than one unbroken monologue. Speak a section, pause, glance at it, then speak the next. This keeps each block coherent and makes the editing pass faster, because you are correcting small units instead of untangling one long stream. Over a few documents, you develop a rhythm: a sentence or two of thought, a brief look, continue. That cadence produces drafts that need very little cleanup.
Helping the Whole Team Adopt It
Documentation debt is usually a team problem, not an individual one. If you want voice documentation to stick across a team, a few things help. Make the easy wins visible first, like dictating PR descriptions and release notes, where the time saved is obvious and the stakes are low. Share a starter Smart Vocabulary list of the team's common terms so nobody has to build it from scratch. And set the expectation that a fast, imperfect draft beats no document at all. The biggest lever on documentation quality is simply that more of it exists, and that is exactly what lower writing cost produces. Voice fits naturally alongside the other tools in a well-run setup, which we cover in our roundup of the best Mac productivity apps for 2026.
The Bottom Line
Documentation does not get written because writing it is slow and starting it is hard, not because engineers do not value it. Dictation removes most of both obstacles. You speak the explanation at the speed you think it, you skip the blank-page paralysis because talking is something you already do effortlessly, and you end up with a complete draft you can shape in a fraction of the time it would take to type from nothing.
Voice Keyboard Pro is a native macOS menu bar app that types wherever your cursor is, with Smart Vocabulary for your team's terms and a Meeting Mode that captures decisions as they happen. There is a free tier with daily limits, and Pro is $4.99 a month or $34.99 a year. Open that README you have been avoiding, hold the hotkey, and just explain it. The first draft will be done before you would have finished typing the first paragraph.