On video calls, Maria is sharp and commanding. She's a senior account executive with eight years selling to US enterprise. Then she moves to Slack, and something shifts: the same thought that took her one sentence to explain on the call becomes three rewritten drafts before she hits send.
The Call vs. the Draft
Maria speaks English fluently. Her performance reviews say so: "Excellent exec presence on calls." But they also note she could be more concise in writing. She knows the gap isn't real. It's the feeling of exposure that comes with async text.
On a call, her accent carries confidence. In Slack, she reads her own words back and hears them as careful, smaller, less like her. This is the hidden tax for senior knowledge workers in Northern and Central Europe writing in English all day. Fluency on video doesn't transfer to written Slack.
The rewrite loop is real: Maria spends 30 minutes on a single Slack message to her team that would take two minutes to say. Same thing happens in Gmail drafts to clients and Linear comments where she's documenting a decision. The pattern is identical. Voice gives her permission. Text makes her second-guess.
Why Existing Tools Don't Solve This
The obvious answer is voice input. And it makes sense. You speak naturally on calls, so why not speak into your Slack drafts?
But existing voice tools fall into two categories. There's dictation software like Dragon NaturallySpeaking, which captures every word you say, including the filler sounds and restarts. There's transcription, which works only if you record first and transcribe after. Neither one gives you back the time or the voice you want.
And there's Apple Dictation, built into your Mac, which is free and convenient until you realize it only works in macOS and caps the utility. If you're on Windows, or if you need seamless input across Slack, email, and browsers, it falls apart.
What they all share: they're designed for beginners or as afterthought features, not for people who are fluent but careful. None of them let you speak naturally and end up with text that sounds like you wrote it, not like you got transcribed.
The Exposure Trap
What's actually happening in those rewrites is identity protection. On calls, your accent is an asset. In async text, you experience it as liability. Every message reads back to you with an internal translator's filter: "Would a native speaker write this? How do I sound?"
This happens to everyone learning a language. It's supposed to be temporary. But for senior professionals years into English fluency, the filter becomes invisible infrastructure. You stop noticing you're running every Slack through it.
You catch yourself rewriting "I think we should consider" to "I believe we should consider" because the first one sounds too tentative. You change "This is an interesting approach" to "This approach has merit" because tentative sounds safer than eager. Three rewrites later, you've deleted the voice that survives your accent.
The real problem isn't vocabulary. It's that existing tools either leave you with raw transcription or with marketing-smoothed text that sounds like neither the call version nor the written version of yourself.
What Changes When Input Matches Intent
Voice input works when you can speak naturally and get output that's actually usable. That means:
- Local processing on your device, no cloud upload of your drafts to somewhere you can't control
- Works everywhere you already type: Slack, Gmail, Linear, browsers, even terminal if you're technical
- No word limit and no metering, so you're never thinking about how much you can afford to dictate
- Output that sounds like you typed it, not like you spoke and then someone cleaned it up
When input matches intent, the tool disappears. You speak. It captures. You edit if you want. You send.
For Maria, that means the 30-minute Slack rewrite becomes a two-minute voice draft plus ten seconds of reading it back. The version of herself that shows up in async text starts to match the version on calls.
The Person You Are on Calls Belongs in Your Writing
Your performance reviews probably say something similar to Maria's: sharp on calls, great in meetings, executive presence clear. That person is the real you. The careful, rewritten version in Slack isn't a weaker version of your English. It's a weaker version of you because the medium changed how you express yourself.
The gap isn't in your skill. It's in the tool you're using to input your thought. When the input method trusts that you already know English, trusts that you already sound like yourself on video, and doesn't make you pay a rewrite tax for permission to exist in writing, something shifts.
You stop translating. You start writing the way you actually think.