I switched to voice-first for 30 days. Here's what changed.
For one month I tried to use my voice instead of my keyboard for as much as possible. Morning planning, client emails, journal entries, blog drafts, meeting notes, even some code. Here's the honest version of what worked and what didn't.
Why I tried it
Two reasons. First: I think faster than I type. Always have. The friction between "a thought I want to capture" and "the words on screen" is where most of my best ideas get diluted. By the time I've typed the third sentence, the original thought has been edited beyond recognition.
Second: I started getting wrist pain. Not severe, but noticeable. The kind of low-grade ache that you start to dread. I figured if voice could replace even a third of my keyboard time, it'd be worth experimenting with.
Week 1: awkward
The first three days were rough. Speaking your thoughts out loud feels weird, especially in a coworking space full of other people. I switched to working from home for the rest of the week.
The bigger problem was that my voice notes were terrible. Disorganized, full of false starts, and reading them back was painful. I'd open a transcript and immediately want to delete it. I came close to giving up on day 4.
What saved the experiment was realizing the raw transcript isn't the output — it's the input. The AI rewrite layer (turning "um, so I think we should, like, definitely make sure that, you know, the meeting tomorrow doesn't go too long" into "Keep tomorrow's meeting short") is what makes the whole thing work. Once I started running every voice note through a rewrite step, the quality jumped enormously.
Week 2: clicking
By the second week I was using voice for:
- Morning planning. 90 seconds of talking through the day, AI rewrites it as a structured list.
- Replies to long client emails. I'd talk through what I wanted to say, AI would polish it into something professional.
- Capturing ideas during walks. This was the big one. I do my best thinking walking around the neighborhood and previously I'd lose 80% of those thoughts by the time I got home. Now I'd open the app, talk for 30 seconds, and have a usable note waiting.
- Long-form blog drafts (like this one). Talking through an outline, then talking through each section, then editing on the keyboard. Way faster than typing from a blank page.
What didn't work
- Code. Voice for code is a gimmick. The dictation gets variable names wrong constantly. Don't bother.
- Anything technical with lots of numbers, URLs, or jargon. Hard for the model, exhausting for me to dictate.
- Quick replies in a chat. Slack messages, iMessage, Discord. Faster to type. Voice is for thoughts longer than a sentence.
- Anywhere I couldn't talk privately. Cafés, coworking, busy streets. Voice notes for personal stuff in public is awkward.
Week 3-4: the unexpected change
The thing I didn't expect: my writing quality went up. Not because the AI was making it better (though that helped), but because I was thinking through arguments more clearly when I had to articulate them out loud. Typing rewards meandering. Talking rewards getting to the point.
I noticed it most in client emails. The voice-first replies were 30-40% shorter, more direct, and got better responses. Several clients commented that I was "more decisive" — which I read as "you used to write three paragraphs and now you write one."
The wrist pain went away around day 18.
Tools I used
I tried four apps over the month. Apple's built-in dictation is fine for short snippets but doesn't do AI rewriting. AudioPen and Bear were both clean experiences but limited. I ended up settling on Waver because it had the writing styles I needed (different ones for emails vs blog drafts) and the meeting recording I needed for client calls.
I'm not religious about any specific app — they're all converging on the same feature set. The thing that mattered was committing to the workflow for long enough that the awkwardness wore off.
Would I keep doing it?
Yes, but not as a 100% replacement. Voice for thinking, drafting, and capture. Keyboard for editing, code, and short replies. A 60/40 split feels right.
If I had to summarize the takeaway: voice is great for getting from zero to a draft, terrible for the last 10% of polish. Use it accordingly.
If you want to try a 30-day version of this yourself, Waver is free for the first 10 notes — try it.