What we’re learning about voice to structure

When we launched Sway here, the idea was simple: To speak out your thoughts and get structure.

Since then, a lot has happened. We shipped quite a few features. But what’s been more interesting to me is how people actually use it. A few things surprised us.

First: People don’t speak like they type. Voice is messy, circular, emotional and sometimes contradictory.

And when it gets structured, it often reveals something the person didn’t consciously articulate before.

That part still feels a bit magical I think. 😁

Second: Our newly released Feature “Continue your thoughts” became way bigger than we expected.

Originally it was just a small idea inspired by a comment here.

But users kept wanting to go one layer deeper. They wanted Momentum.

Third: Our new “Second perspective” feature doesn’t feel like advice. It feels like distance.

Like stepping half a meter outside your own narrative. That shift alone can change the tone of thinking.

We also added text input recently. And interestingly, people use it very differently from voice.

Voice feels reflective. Text feels like dumping and organizing.

Same engine, different psychology.

And one more thing we didn’t expect:

Output language matters a lot. Some users think in one language and prefer structure in another.

We’re still figuring this out. Voice to structure feels like a new behavior category, not just a feature.

Curious if you’ve used tools like this:

What feels natural? What feels forced? Where does it break?

Appreciate this community more than you know. A lot of this came directly from your comments.


Christian 🫶