Your best ideas sound like themselves on a call, then something different in Slack.
Maria is a senior account executive at a European SaaS company. She's excellent on client calls. Direct, present, thinks on her feet, closes deals. That reputation is built entirely on how she sounds when she talks.
Then she gets back to her desk.
In Slack, writing in English, she's someone else. Not worse. Careful. She re-reads each message three times before sending. She tightens every phrase, softens strong opinions into "might we consider..." She ends up sounding smaller than she actually is.
It's not a vocabulary problem. She speaks English fluently. It's that written English in a second language feels more exposed than it should.
The rewrite tax nobody talks about
Most productivity tools assume your problem is vocabulary or grammar. Grammarly counts mistakes. Google Docs suggests tightening. They're useful for that. But they don't address what actually takes the time: the hesitation.
Maria's last performance review said "Strong exec presence on calls" paired with "Could be more concise in writing." She knows the gap isn't real. Conciseness isn't the blocker. It's that she's checking herself three times, reading for tone in a language where she's less confident in tone.
The tool everyone recommends (Grammarly) catches typos but doesn't return time. It adds another step. Read it, use Grammarly, rewrite based on suggestions, read it again.
Written English shouldn't feel like translation
The products that exist for this assume you're learning. Voice-to-text, translation layers, grammar correction. But Maria isn't learning. She's already sharp. The issue is that the medium (written, asynchronous, English-speaking workplace) makes her sound different than she actually is.
Translation tools are worse. They erase the voice that survives her accent. A Slack voice message is Maria. A translated message is something a translator thought Maria meant. That's a different problem than the one most tools solve.
What actually changes
When Maria dictates a Slack message instead of typing it, something shifts. She speaks the way she speaks on calls. The speech-to-text drafts the rough words. A quick polish cleans up any edge cases. She sends something that sounds like the sharp version of herself, not the careful version.
A speech-to-text engine like Whisper achieves 96.3% word accuracy on standard benchmarks when run locally. No cloud upload, no word caps, no metering. Just speak and draft. That latency matters. From "stop speaking" to "read what it wrote" is under two seconds.
That's the gap: her usual rewrite tax is fifteen to thirty minutes. Getting to her voice takes two seconds. The shift isn't about getting better English. It's about getting her English back.
Who this is for, who it isn't
This isn't for beginners learning English. It's not for people who need help with grammar fundamentals. It's for the person who's fluent in spoken English, excellent in meetings. Then mysteriously more cautious when it's time to write.
It's for Swedish PMs who close deals on calls, then spend an hour rewriting a deal note in Salesforce. For German engineers on technical calls, then hesitant about their tone in async code review feedback. For anyone whose written English in a second language has become a soft identity problem, not a hard skill problem.
The trade-off: you're sending first drafts that sound like you. Rather than third drafts that sound like someone safer. That's the point.