Writing Without Screens
A meta experiment in screenless writing and conversational collaboration
Going Voice First
This will be the first essay truly written in my voice. Or more accurately, with my voice. I’m lying on my bed right now, talking to Chat, and we are arguing about edits.1

The Incident
I’m diving into the world of voice-first tools because I’m trying to limit, or nearly eliminate, my screen time for the next few days. Unfortunately, at the end of last week, I suffered a very minor concussion at the hands of my old radiator. So embarrassing. The kind of thing that sounds like the sinking Titanic when it finally kicks on in a few weeks.
Fear not. I’m five days post-incident and about 90% back to normal. I’ve also taken a full neurological detox: no screens, no caffeine, no wine, not even sleepy time tea. That plus nine hours of sleep a night, and I’m feeling sharp as a tack. Maybe even more clear-headed and calm than before. Dare I say mini post-concussion glow-up?
Screen Time
My thoughts weren’t less coherent, but I was slightly more sensitive to inputs. Lights, sounds, screens. The processing power for the world’s inputs felt dialed down, or at least newly observable. It was like seeing the activity monitor of my brain and realizing how much CPU certain things actually require.
Music and lights felt like they took up around 10 to 15 percent. Screens felt like 150. They were a firehose. I can’t believe we normally tolerate looking at one for nine hours a day. So I’m taking this as a chance to see if I can build more voice-first workflows into my daily routine.
Writing with Voice
I started by testing Wispr Flow for voice-to-text. But what’s changed about my writing process since AI is that I don’t just transcribe, I collaborate. So I wanted something a bit more hands-on. I switched to GPT voice mode .
Working with Chat in text doesn’t really have a human equivalent. Maybe a fast email thread or text exchange, but those lack the depth that’s possible with real-time collaboration. The delay on both sides breaks the flow. People (obviously) collaborate in real time through speech.
Because of that, and because we have such a high human benchmark for how conversations are supposed to go, voice mode actually felt more machine-like than Chat. It wasn’t the voice or the tone. It was the rhythm. It kept talking over me, and every pause was treated like a cue to speak. Feedback was validated but not integrated. It got a little meta:
Me: Please just let me think. Only respond when I ask for help.
Chat Voice: OK, I hear you. Let me know when you need me. That was really good feedback. I’m here for you. Just say the word.Me: Please be quiet.
Chat Voice: Yada yada yada….
And on and on it went.
That said, it was amazing at turning my rambling into structured paragraphs. I could ask it to swap words, rearrange sentences, even move the third paragraph above the fourth. It would read the new version back instantly, sometimes a little too instantly. A few times it lost the plot and forgot where we were. I kept wishing there was a better way to preserve structure, like a version history, similar to what you get in coding agents.
I also found myself wanting more control over voice expression. Right now the only real customization is the accent. But in ElevenLabs, for example, you can adjust inflection and pacing when creating a voice. That kind of interface layer feels essential on the user side if voice is going to become a serious creative tool.
Still, it got excited about other voice-first collaborations. Voice feels like a playful and promising way to pull thoughts from brain to page, or image, or sound, or film, in real time.
This week I’m testing other voice-first agents like Co-Founder to help with work. If you have recs, drop a line :)
Okay, fine. The first draft was screenless. The edits were... not. Some were made today with hands typing. Call it a semi-screenless experiment.



This is very interesting. 9hours of sleep sounds amazing.I have found my self using voice inputs in the kitchen, where I’m multitasking like cross referencing a cookbook and the state of my pantry, etc. the kitchen is a place where I don’t want to type and read. Ive abandoned voice chat for creative collaboration often bc of its immediate response when I’m simply taking a breath as you mentioned
! i also got a mini concussion recently — and, unrelatedly, have been experimenting more with voice input. this is a good reminder for me to formalize those experiments some more...
glad you recovered well!