I never expected an AI app to replace my keyboard, but I was wrong

I haven’t typed a single word in this story — all of it came to life through my voice. I simply spoke to my computer what I wanted written, and the words appeared on the screen with punctuation corrected and my verbal fallacies ignored, without me lifting a finger.

As a writer-journalist, I take pride in my ability to pound the keys fast enough to blur the motion without even once peeking at the keyboard. But there often comes a point when my fingers can’t keep up with the pace of my rushing thoughts. And when your profession routinely demands meeting tight deadlines, speed becomes the most important factor. And somewhere along the way, typing itself became a holdback.

I recently discovered a tool that pushed my keyboard into a secondary role and elevated dictation to the center of my writing workflow. Wispr Flow lets me put words together without touching the keyboard at all. And honestly, I think I’m addicted.

Percussion to verbal shift

If you imagine a writer, you probably picture someone slouched over an old-school typewriter, hammering away at a QWERTY keyboard. On my desk, the image isn’t very different — except I have a compact Logitech keyboard instead of a bulky typewriter. I’ve always believed typing is an art form in itself, separate from writing, much like playing a percussion instrument. You build a relationship with the instrument — in this case, the keyboard — and switching modes can feel unsettling at first.

You could call them teething problems, because my first dictation-heavy writing session was less than elegant, having stepped out of my comfort zone. But once I found my footing, things clicked (no pun intended), and I picked up pace with ease. Voice typing turned out much faster than typing, and that changed how my raw thoughts translated onto the page. My writing now feels closer to the spoken language; it’s more conversational and less constrained by the mechanics of written words. That’s quite an abstract advantage, but there are some tangible ones as well.

Don’t tell my editors, but I struggle with certain spellings and can’t get them right even after multiple attempts. “Simultaneously,” for instance, is something that always requires rewrites. With voice typing, it lands correctly on the very first try. For someone who can’t ignore that red wiggly line under misspelled words and breaks the flow to fix it, this alone frees my mind up.

Traditional voice typing tools can do some of this, one would argue, but Wispr Flow works differently. It understands when I correct myself mid-sentence (something I do constantly while speaking) and smartly displays only the corrected version. Without that layer of AI keeping up with my ever-changing thoughts, dictation would become slower than typing because you’d be stuck fixing what you just said after every other sentence. That’s friction, not effortlessness.

The vocal pickle

That said, Wispr Flow isn’t foolproof. While dictating this very story earlier, it randomly dropped a word, and that happens often. At other times, it ends a sentence where a comma should be, or inserts a hyphen when I specifically asked for an em-dash. Despite having context and access to language models to understand intent, it still makes mistakes.

There have also been moments when it swapped in words that don’t belong at all. In one story, it replaced “NASes” with “nurses” when the entire story was about network storage — a gaffe that would have been deeply embarrassing (and funny) if it had slipped through. Because of this, I dictate in short bursts, usually a couple of sentences at most. The text only appears on the screen once I release the Fn key on my Mac; it doesn’t show up word by word in real time. That means I have to actively watch what it is listening to and typing.

My voice writing is limited to my work needs, so I use Wispr Flow exclusively on a Mac. But I can’t even use it on Android even if I wanted to, because it’s not openly available there yet and requires joining a waitlist even for an early access. It’s available on iPhones but requires switching the keyboard to Wispr Flow — if there’s a similar requirement on Android, you’d be better of sticking to Gboard’s own excellent voice typing.

Despite all that, I can affirm that it gets things right the vast majority of the time, easily hitting 90–95% accuracy. For the kind of writing I do and the volume I’ve put through it over the past few weeks, that’s an impressive hit rate. I started with the free trial, but I trust it enough now that I’ve paid the $144 for the full year. That’s how quickly it’s become part of my work routine.

Keyboard meeting handwriting’s fate?

There’s always a debate about preserving tools from bygone eras when newer ones take over. We saw it when typewriters replaced handwriting, only to be replaced by keyboards eventually. Handwriting itself has become so rare that many of us struggle to do more than sign our names with a pen.

Using Wispr Flow for an extended stretch made me wonder if voice typing could do to keyboards what keyboards once did to handwriting.

My output may be higher, but I’m typing far less than I used to. While critics worry about AI dulling cognition, my concern is rather simpler: muscle memory. Typing is a percussive craft, and like any craft, it fades if you stop practicing it. So, for now, I’ve only delegated part of my process. I use voice typing to get the vomit draft out. The keyboard stays for editing and all the small corrections that dictation doesn’t handle well — and it’s still my reliable tool for writing when travelling because I just can’t be seen talking to a computer at the airport.

But for now, productivity has won. The keyboard has taken a back seat to my voice, and honestly, I have no regrets. It’s made my writing life easier in ways I didn’t expect — enough that I find myself wanting others to try it too. Whether this lasts or I get back to my keyboard is a question for another day.