The call went great. You were sharp, present, made the client laugh twice, closed the next step. Now you're writing the follow-up Slack to your team, and you've drafted it four times already.
The first version sounded too confident. The second sounded too careful. The third you read aloud to check how it lands. The fourth, this one might be it. Actually, re-read it one more time. It's been 30 minutes.
This is not a vocabulary problem. You know the words. The friction is something else: the gap between how you sound on a call and how you sound in writing, and the identity tax of closing that gap alone.
The presence that doesn't translate
When Maria, a senior account executive with eight years selling to US enterprise, gets on a video call with clients, her presence is unmissable. She's sharp. She catches nuance. She thinks in quick, connected sentences. Her accent is not a liability; it's distinctive. But when she moves to Slack, internal notes to her team, follow-ups to prospects, deal summaries in Salesforce, something collapses. The writing sounds smaller than her speaking. More careful. Like she's translating herself instead of expressing herself.
She isn't alone in this. Senior knowledge workers in Stockholm, Berlin, Amsterdam, Madrid, people who've spent years fluent in English corporate environments, hit the same wall. On a call, you're fluent. In Slack, you're cautious. It's not ability; it's the medium. Speech has momentum, accent, pacing, personality. Writing in a second language asks you to slow down and perform precision you didn't know you needed to maintain.
The cost isn't obvious until you see it named. The performance review Maria got last cycle flagged "exec presence on calls" as a strength. "Could be more concise in writing" was marked as a gap. She knows that gap isn't real. It's the medium. The medium is the problem.
Why Grammarly made this worse
You'd think a tool like Grammarly would solve this. It doesn't. Grammarly catches grammar, yes. But it's built on an assumption: that non-native writers need help getting the words right. You don't. You need help getting from "rough spoken thought" to "clear written thought" without erasing the voice that survives your accent.
Grammarly doesn't do that. It rewrites your voice into something safer, more formal, more cautious. It's designed for native speakers polishing their own writing. For someone writing in a second language, it often makes the written version sound even more distant from how you actually think.
The other class of tools, transcription apps like Otter.ai, assume you want to convert speech to text. That's different. You don't want a transcript. You want to speak naturally and have the tool understand what you meant, not just what you said, then give you back a clean written thought that sounds like you.
The rewrite tax is real
For Maria, the process looks like this: a five-minute conversation becomes a 30-minute Slack message. She drafts it, re-reads it three or four times, revises it, and then sends. The message itself is maybe 150 words. The time spent is 30 minutes of cognitive load that didn't happen on the call.
But it's not unique to Maria. Non-native writers across enterprise teams report the same pattern: conversations that flow naturally become written messages that feel like they need multiple passes before sending. The tax isn't on getting the words right. It's on feeling safe sending them.
That rewrite cycle kills momentum. It breaks the connection between thought and expression. It forces you to perform a version of yourself that's smaller and more guarded than the person in the meeting.
What you actually need
You don't need grammar rules. You don't need a translation layer. You need something that understands the gap between your spoken thought and written expression, and closes it without erasing your voice.
That means:
- Capture what you actually said (rough, natural, imperfect)
- Polish it into clear written language (without making it sound formal or overcautious)
- Keep the voice that survives your accent
- Do this fast enough that it doesn't interrupt your workflow
Existing tools solve one or two of these. Recitey is built to solve all of them at once.
How it works in practice
You speak a message into Slack, rough, natural, the way you'd explain it on a call. Recitey runs Whisper (the model OpenAI trained, with 96.3% accuracy on standard English speech) locally on your device, captures what you said, and polishes it into a clean sentence in under two seconds. No cloud dependency. No variable cost per word. No usage caps on free tier. It works in Slack, email, Google Docs, Salesforce, your browser, everywhere you type.
The difference: it's not trying to teach you English. It's trying to preserve what you're actually saying while removing the friction of writing it. It listens to your accent, understands your pacing, and gives you back a sentence that sounds like you, not like a careful version of you.
For Maria, that means a five-minute thought becomes a five-minute Slack message instead of a 30-minute one. The message sounds like her, sharp, present, connected, not like a watered-down version she's performed for safety. It sounds like the person who was on the call.
What changes after
Once you stop paying the rewrite tax, async communication stops feeling like a performance. You're not rewriting to sound smarter or safer. You're expressing the thought you already have, and the tool is removing the friction of format, not voice.
Your presence doesn't disappear in writing. It was never gone. It just got trapped behind the medium. Remove the friction, and it comes through.