---
title: Voice as the interface: 24k words dictated with Wispr Flow, a note by Arthur Galvão
url: https://arthurgalvao.work/notes/voice-as-the-interface
lang: en
---

# Voice as the interface: 24k words dictated with Wispr Flow, a note by Arthur Galvão

> Jul 28, 2026. Posted on [X](https://x.com/arthurgalvaorl/status/2082131261083001256), [LinkedIn](https://www.linkedin.com/feed/update/urn:li:activity:7487896810930356225/)

Markdown version of https://arthurgalvao.work/notes/voice-as-the-interface, the same content as the page, for AI agents and language models.

Most of my communication with AI these days starts with my voice.

I've been using Wispr Flow for almost 2 months, and the counter just passed 24,000 dictated words. Prompts, briefs, strategic planning...

The speed difference is real. We speak at 150+ words per minute, while most people type at 40 to 80. That gap changes how you work with AI: talking hands the model more context, and more context means better output.

It helps beyond AI too, messages and docs included, but prompting is where I feel it most. Natural language is becoming the main interface for building things, and speaking is the most natural way to produce it.

There are free dictation options out there. Wispr Flow is the one that stuck with me, the most polished I've tried.

[All notes](https://arthurgalvao.work/notes.md)