You're a senior PM or engineer, which means your spoken English is excellent. On calls with US stakeholders, you're sharp. You articulate the problem, you push back cleanly on timelines, you explain the tradeoff. Minutes after that call ends, you open Slack to write the follow-up, and something shifts. The message you write is half the person you just were on the call.
The call-to-Slack gap
Spoken English and written English are not the same skill. For non-native speakers, especially those raised in German or Swedish contexts, the gap is much larger than people assume. You've trained your ear and your mouth in real conversations over years. You've done hundreds of sales calls, technical standups, stakeholder discussions. Spoken English has rhythm, intonation, and audience feedback. If someone doesn't understand you, they interrupt and you clarify in real time.
Slack has none of that. Slack feels permanent. Slack feels recorded. Slack feels judged by someone who isn't present in the room to hear the tone of your voice.
The written version of you is always more careful. You're not second-guessing your vocabulary; you're second-guessing whether your intended meaning will survive the lack of context. A tone that reads as confident in person reads as curt in text. Directness becomes bluntness. You compensate by softening, adding hedges and qualifiers you'd never use on a call: "perhaps", "maybe we could consider", "if I'm not mistaken". By the time you hit send, you've rewritten the message five times, and what emerges is someone smaller than who you actually are.
Why Slack drafts become a rewrite loop
Pick a 4-5 sentence follow-up message after a call. Time yourself. First draft takes three minutes. You read it. Something feels off. The word order is too formal. You rewrite. Five minutes. Read again. Better, but now it sounds defensive. You dial it back, adding softer language. Ten minutes total. One more pass to make sure nothing sounds too direct. Thirteen minutes. You send it.
You've just spent 13 minutes writing what you explained in 30 seconds on the call. But the rewrite wasn't about clarity. You were fluent in what you said. The rewrite was about protection. It was about trying to translate the confidence you felt in speech into a written register that doesn't punish you for being direct.
This is not universal to non-native speakers. This is specific to senior people who are fluent in spoken English but experience written English as a formal register where mistakes feel more exposed. Your spoken fluency gives you the confidence to try; your written medium takes that confidence away.
Why existing tools make it worse
Grammarly exists to catch mistakes. It's good at that. It highlights your missing articles, suggests tense corrections, flags informal phrasing. But Grammarly doesn't solve your actual problem. Your actual problem isn't grammar. Your grammar is fine. Your actual problem is that you've spent 30 minutes rewriting a message you explained perfectly in 90 seconds on a call, and Grammarly can't give you that time back.
Slack's native voice typing works the same way. You record something, and what comes back is a transcript. That transcript is often grammatically correct but tonally flat. It sounds like the literal version of what you said, not the refined version. So you rewrite it anyway, because the transcription is just a starting point, not a finished message.
The tools assume the problem is "how do I write," when the actual problem is "how do I write and still sound like myself."
What changes when you skip the rewrite cycle
Maria is an account executive at a European B2B SaaS company. Her performance reviews are consistent: "Exec presence on calls" is always noted as a strength. "Could be more concise in writing" is the gap. But she knows the gap isn't real. It's not that she can't write concisely; it's that she rewrites every message four times, and the final version is half-concise and half-cautious because she's protecting against being misunderstood.
When she could speak her message and get back a clean, polished version without the rewrite cycle, something shifted. Not because her language improved. Because she stopped translating herself. The messages she sent sounded like her, not like a watered-down interpretation of her.
The friction drops by about 80% when you remove the rewrite cycle. Instead of a 30-minute Slack message, you're looking at a three-minute dictation followed by a 15-second read-through. Instead of five versions, you're reviewing one. The psychological effect is harder to measure but more important. Your team hears your actual voice in async communication. Your follow-ups after calls maintain the momentum of the call instead of resetting it to polite caution.
The technical piece that makes it work
What's different is that the transcription happens locally on your device, on Whisper's large model, which achieves 96.3% word accuracy on LibriSpeech. Instead of giving you a transcript, the system polishes it in place, applying context from where you are in the app: Slack, email, Google Docs, Salesforce, Notion. The result isn't a transcript. It's a draft that already sounds like you, not like a robot reading you aloud.
You don't rewrite. You review. Those are different tasks. Rewriting is protection. Review is refinement.
Who this is for
This only works if you're already fluent. If you're still learning written English, you need the practice and the feedback that Grammarly gives you. If you're a native English speaker, this solves a problem you don't have. This is specifically for people who are excellent in their second language but feel smaller in writing than they do in speech. If that's you, the cost is zero. The time you get back is entirely gain.
The hard part isn't the tool. It's believing that the problem was the medium, not you.