I used to spend about 45 minutes every morning on emails. Writing, rewriting, second-guessing tone. It was the most cognitively expensive part of my day and none of it felt like real work.
Then I started talking instead of typing.
Not dictation in the old sense. Not just speech-to-text where you spend the next 10 minutes correcting errors. I mean a voice-to-action system that takes what you say and turns it into the finished thing. Draft the email. Write the reply. Enter the data. Not a transcript of your rambling. The actual finished output.
I built Genie 007 (genie007.com) to solve this for myself first. You say a rough thought, the system figures out what you actually meant, and delivers the action ready to use. Works everywhere — Windows, Mac, Android, iOS, Chrome extension. Any app, any website, no integrations needed.
Here's what actually changed after 4 months:
Email: I go from zero to sent in about 3 minutes instead of 15. I say what I want, it drafts, I review and send. My inbox is no longer the thing I dread at 8am.
Posts and replies: I do most of my community engagement while walking. I talk through a comment, it lands ready to post. Felt strange for a week. Now it feels obvious.
Second-language writing: Three users told me that using voice to write in English removed the inhibition they had when typing. You think in your first language, say roughly what you mean, get native-level output. One user said it gave her her voice back. That one stayed with me.
The thing nobody tells you about a voice-first workflow is that it changes WHAT you write, not just how fast. When you remove the friction of typing, you write more. You reply more. Ideas that used to die between brain and keyboard actually make it out.
4 months in: inbox time down from 45 minutes to about 10. I post 3-4x more often. Reply rate to DMs significantly better.
The main pushback: doesn't it make mistakes? Yes, occasionally. But the edit takes 20 seconds. The draft from scratch took 10 minutes. The asymmetry is obvious once you see it.
If you're a solo founder doing your own writing and outreach, the math is brutal. Every hour on formatting prose is an hour not on customers, product, or real thinking. The writing should be the fast part.
What part of your day eats the most time on writing that isn't really writing? For me it was email. For others it's Slack. For others it's documentation.
Turning rough speech into the intended action, rather than simply transcribing it, is what makes this interesting. As an ML engineer, I’m very curious about how Genie 007 understands context and preserves the user’s voice. I’d love the opportunity to take part in a project like this.
Thanks, Dexter. Most transcription tools stop at accuracy - the voice preservation problem is harder because it's about learning how someone writes, not just what they say. The context layer builds from structured prompts plus a style profile that develops over time. Still actively working on this side of it. Happy to chat more on the technical side if useful.