Want a morning pages app? Here is how to keep Julia Cameron's three-page practice by speaking it, plus what to look for and how privacy works.
Julia Cameron's morning pages ask for three handwritten pages of stream-of-consciousness writing, first thing, before your brain can edit. The practice works. The friction is real: your hand cramps by page two, your wrist slows long before your thoughts do, and on the mornings you most need to clear your head, you skip it because writing by hand feels like a chore. Most people who quit morning pages do not quit the idea. They quit the pen.
Speaking the pages instead removes the physical bottleneck. You talk at roughly 130 words a minute and handwrite at maybe 20, so a three-page dump that took 30 minutes on paper takes a fraction of that out loud. This guide covers what a morning pages app needs to do and where the privacy trade-offs sit for something this personal.
Cameron's rules are deliberately loose: three pages, longhand, every morning, no stopping, no censoring, nobody reads them but you. The point is not good writing. It is to move the noise out of your head onto a surface so the rest of your day starts clearer.
Two of those rules fight each other once you try to keep the habit. "No stopping" wants speed. "Longhand" caps your speed at the rate your hand can move. When your thinking outruns your writing, you either lose thoughts or slow your thinking to match the pen, which defeats the exercise. That gap is why so many people trail off after a couple of weeks.
A digital approach keeps the spirit (unfiltered, private, daily) while cutting the physical limit. Typing is faster than handwriting but still pulls you toward editing: you see typos, you backspace, you tidy. Speaking is the closest match to "no stopping, no censoring," because you can keep talking while the words appear on their own.
Not every note-taking or journaling tool suits this. The practice is specific, so the tool should be too.
Contextli is a context-aware speech-to-text desktop app. It sits at the system level and types what you say directly into whatever window is focused. For morning pages that means you open a blank notes file or a plain document, start dictation, and talk. The words appear as you speak, in the place you want them, with no copy step from a separate transcription window.
The relevant Mode is General Dictation. Email Mode and Messaging Mode reshape your words to fit a channel, which is exactly what you do not want at 6am. General Dictation stays out of the way and gives you a straight transcript, matching the "no censoring" rule.
You can also feed a Mode three or four examples of how you actually write, or specific instructions, and every dictation matches that. Most people will not bother for a private dump, and that is fine. If you keep a lightweight structure (you always start with the date and a one-line weather note), you could set that up once so it appears without thinking. The customization is there if you want it and invisible if you do not.
One honest limit: Contextli types into the focused app. It has no built-in journal, no calendar of entries, no streak counter. It is the input layer, not the journaling product. You bring the destination (a notes file, a document, a folder), and it fills that destination with your voice.
The pen was never the point of morning pages. Getting the noise out of your head before the day starts was the point. Voice gets it out faster.
Morning pages only work if you are honest, and you are only honest if you trust that nobody else will ever read them. So it is worth being precise about where the words go. Contextli has a three-rung privacy stack you can turn on independently:
Stack all three and the practice is fully offline and fully private: nothing leaves the laptop. For a ritual built on saying things you would not say out loud to anyone, that matters.
The choices for keeping this practice come down to how fast you can capture and how private the words stay. Here is a plain comparison.
| Approach | Capture speed | Editing pull | Privacy control | Fits "no stopping" |
|---|---|---|---|---|
| Handwriting (pen and paper) | ~20 wpm | Low | Total (paper) | Poor: hand tires |
| Typing into a notes app | ~40-60 wpm | High (you see and fix typos) | Depends on the app | Fair |
| Native OS dictation | ~130 wpm | Medium | Varies by OS | Fair |
| Contextli General Dictation | ~130 wpm | Low (keep talking) | Local models, BYOK, sync off | Good |
Native macOS Dictation and Windows Voice Typing are free and fine for a quick note. Where they fall short for a daily private ritual is control: you get less say over where the audio is processed. Wispr Flow is faster if all you need is raw transcription, but morning pages do not need speed records; they need a low-friction, private surface you return to every morning.
Maya, a consultant, used to keep paper morning pages but stopped when her hand cramped past the second page. Now she opens a blank local notes file, starts General Dictation, and talks for about eight minutes while her coffee brews. What used to take 30 minutes of writing (and often got skipped) now fits before her first call. She keeps local models on with the internet off, so the pages never touch a server.
David runs a small agency and wanted the practice but never had the wrist patience for it. Using the customize-by-example flow, he set General Dictation to always drop the date and a one-line weather note at the top, then speaks the rest. Handwriting a full three-page dump would take him past 20 minutes; speaking it takes about seven. He runs it with cloud sync off, so the entries live as plain files in a folder only he opens.
Cameron's rules ask for three pages of unedited stream-of-consciousness, first thing, seen by nobody. The medium is a means to that end. Speaking keeps the "no stopping, no censoring, private" spirit and often serves it better, since you are less tempted to tidy as you go.
No. Contextli's free tier gives you 100 credits per month, no credit card required, which is enough to run a morning routine for a while and decide whether voice suits your practice before you commit.
Wherever you point them. Contextli types into the focused window, so if you open a local notes file or a plain document, the words land there. It does not keep its own journal, so you control the destination and the folder.
Yes. You can run transcription on your own machine with the internet off (local models), supply your own API key so data goes straight to your provider (BYOK), and disable cloud sync so Contextli stores nothing. Stack all three and nothing leaves your laptop.
You can feed a Mode a few examples or specific instructions once, and it will apply them to every dictation. A date line or a fixed opener appears without you saying it each time. If you prefer a pure blank dump, skip this entirely.
General Dictation. Email Mode and Messaging Mode reshape your words for a channel, which you do not want for a raw morning dump. General Dictation gives you a straight transcript that matches the practice, and it runs offline on your own machine if your laptop is modern enough for the local models.
If morning pages are part of a wider capture habit:
For the wider set of daily-life contexts this fits into, see the Contextli contexts library.
If handwriting has been the thing stopping you, keep the practice and drop the pen. Open a blank local file tomorrow morning, start General Dictation, and talk for three pages' worth without stopping. The free tier gives you 100 credits per month, no credit card required, enough to see whether speaking your pages makes the habit stick. Download Contextli and try it on your next morning.
Junaid Khalid
Founder & CEO
Founder and solopreneur writing about how modern businesses run leaner and faster with AI. I build software that turns everyday work, from capturing thoughts to writing and staying organized, into something effortless, and I share what I learn along the way.
A recovery journal with daily prompts for people in recovery. What to write, how to keep it private, and how to capture entries by voice in under a minute.
Junaid Khalid
Build a recipe experiment log that captures what worked, what failed, and what to try next. Voice-first notes so you write while the pan is still hot.
Junaid KhalidCapture real estate showing notes by voice the moment you leave each property, so five homes in one afternoon never blur into one.
Junaid Khalid
