I wrote this on a walk

I wrote most of this on a walk this morning. I didn’t type a word of the first draft. I spoke it.

I’ve been a chatty girl my whole life. I’m a verbal processor, which is a fancy way of saying I figure things out by talking. For most of my career that meant I needed a person. After a rough client meeting I’d call someone on my team and we’d talk it through. What happened, what we do about it, what makes sense, what doesn’t. By the end I’d have an answer, but more than that I’d have it out of my head. As someone who naturally overthinks and loves to spiral, talking is how I stop.

Even better for me, the voice tools that showed up in AI over the last year changed how much of that I can do on my own.

Let’s talk about four ways I use voice every day now.

Sometimes the person I need to talk it out with is just… me.

After a hard meeting or a tangled decision, I’ll record myself talking it through. Sometimes it’s a real follow-up call. Sometimes I’m alone, narrating a problem to nobody (or my bestie, Claude), which used to feel a little ridiculous and is now genuinely the most productive ten minutes of my day. Either way the thoughts get out of my head and land somewhere I can actually look at them. That alone settles the spiral.

This newsletter is a perfect example. I had something I wanted to say, I went for a walk, and I talked the whole idea through from start to finish. No outline, no structure, just the thought as it showed up.

Then I hand all of it (loose, rambling, goes sideways more than once) to Claude. It cleans it up, gives it a shape, hands me back a few bullets and some opening lines, and I write from there.

The great thing is I’ve basically already written it on the walk. The talking is the draft. By the time I sit down, the blank page is gone and all that’s left is polishing.

I’ll also throw research requests in mid-thought. I’ll just say something like, “find me the studies on verbal processing and how much faster people talk than type, pull that in here.” It drops in as a note to myself, and when I send the whole thing off, the research comes back done. (Fun fact, from exactly that request: most people talk around three times faster than they type. Which is the entire reason this has skyrocketed my productivity.)

One more win: my stream of consciousness never comes out in order. I’ll say something that belonged three thoughts ago, then circle back, then jump ahead. I ask Claude to put it in an order that flows, and it does.

I love using my voice to delegate and instruct, both to my IRL team and to my AI tools.

One of my favorite things about having a team is getting to say “here’s how I want to handle this” out loud instead of typing the whole thing out. But that doesn’t work for everything. Not everyone’s in the room, and nobody wants a 75-minute voice memo sitting in their inbox. So I use voice-to-text through Wispr Flow and just talk. It automatically processes what I’m saying to make it clear and succinct.

When it comes to directing my AI tools, I’ve found I give better instructions out loud. When I type, I tend not to provide enough detail to get the job done correctly. Something about talking takes the pressure off, so I keep going, add the side context, the asides, the where-I-want-this-to-land. That extra context is exactly what makes the output good.

This might be my favorite.

When I’m reviewing something, like this newsletter, I keep dictation on the whole time. As I read, I talk through my edits out loud. What’s working, what’s off, what feels like me and what doesn’t, what I’d cut.

That does two things. It makes this piece better, obviously. But it also makes every future piece better, because I’m not just fixing the draft in front of me, I’m telling my AI why it needs to be fixed. I’m giving it feedback instead of silent corrections.

That distinction is partly how I’ve made using AI a superpower in my work, and it’s where most people go wrong. They treat their AI like a dumping ground. They fix the output, move on, and never explain what was off or what “right” would have looked like. So the tool never actually learns them.

You have to treat AI like a team member, not a vending machine. Build the feedback loop. Tell it what you want more of and less of.

It’s the same thing that trips up new managers with their first junior hire. (That one’s a whole newsletter of its own, so I’ll save that for next week!)

All four come back to the same place. I think better out loud than on a keyboard, and for most of my life that meant I needed another person in the room to do my best thinking. Now I don’t always. The keyboard was the bottleneck the whole time.

If you’ve never tried it, do one thing this week. Next time you’re stuck on a blank doc, get up, go outside, and just talk it. Then let your AI clean it up. I’d bet you write it twice as fast and sound twice as much like yourself.

Cheers,

Audrey