Every message I send to Claude these days, I say out loud. I use Wispr Flow, though plenty of dictation tools do the same job: hold a key, talk, let go, and the words appear wherever my cursor is. Claude, ChatGPT, email, Slack, it doesn’t matter. I genuinely can’t remember the last time I typed a proper prompt.
I’m not trying to sell anyone an app. I just think dictation is one of the most underrated ways to get more out of AI, and almost nobody I talk to does it.
How I actually work now
I talk to AI the way I’d brief a colleague. I explain what I want, why I want it, what I’ve already tried and what I’m worried about. Then I stop, read what comes back and talk again. Most of my messages are a few minutes of me thinking out loud, ums and all, and the dictation tool tidies them up into something readable.
That’s the bit people underestimate. When I typed, I wrote short, clipped prompts, because typing a long brief is a chore. When I talk, I give the full picture without thinking about it.
Why talking works better with AI
AI tools do better work when they know more about what you’re after. That isn’t just my opinion. Anthropic’s own guidance for Claude says that giving context, including the reason behind an instruction, helps it give more targeted answers. It suggests thinking of the model as a brilliant new colleague who doesn’t yet know how you work.
You would never brief a new starter in 12 typed words. But that’s what most of us do with AI, because typing is slow and we’re busy. Talking removes that friction. The context comes for free, and the answers get noticeably better.
It also changes how I think. Saying a problem out loud forces me to work out what I actually mean, a bit like explaining something to a colleague and spotting the answer halfway through the sentence.
What it saves me
I’m nosy about my own habits, so I keep a rough log of how much I use Claude. Over the last couple of weeks I’ve said tens of thousands of words to it every week. If I’d typed all of that, I’d have needed hours I simply don’t have.
The research backs up what I feel. A study from Stanford, Baidu and the University of Washington found that speaking into a phone was about three times faster than typing on it, with fewer errors too. By my rough maths, that’s most of a working day every week I’m not spending at a keyboard. My productivity has gone through the roof, and I don’t think I’m exaggerating.
faster to speak than to type on a phone, with fewer errors, in a study by Stanford, Baidu and the University of Washington.
Stanford News, 2016Where typing still wins
It isn’t for every moment. The researchers behind that study made the same point: there are plenty of situations where typing still makes more sense.
Open-plan offices are the obvious one. Nobody wants to hear me briefing an AI on a client proposal, and I wouldn’t want to say anything confidential where others can hear. Short edits are often quicker to type. And dictation still gets names and jargon wrong sometimes, so I read everything before I send it, especially anything going to a client.
So I wouldn’t tell anyone to stop typing altogether. I’d just say the default has flipped for me. I talk first and type when I need to.
If you want to try it
Start with whatever you’ve already got. Mac and Windows both have dictation built in, and the AI apps have voice modes of their own. If it clicks, a dedicated tool like Wispr Flow is worth a look, because it works everywhere and cleans up the text as you go.
Then give it a proper week. The first day feels odd. By the end of the week, going back to typing long prompts feels like writing a letter by hand.
And if you’re thinking about how AI tools talk about your brand, rather than how you talk to them, our free AI visibility check asks ChatGPT 10 real customer questions and shows you where you stand straight away.




