TL;DR
- People do not speak like writers. ‘Um, tell Sarah the second section needs a pass but keep it positive’ is the input. The email is the output.
- Transcription products treat messy speech as a defect. Lazur treats it as the spec.
- The demo is always the same: you say the messy version, Lazur pastes the send-ready version, you hit send.
Messy speech to finished writing is the product demo that matters. People do not naturally speak like writers. They say:
“Okay so, um, tell John that I looked at the thing and I think we should probably push it to next week because the numbers aren't ready yet, but don't make it sound like we're delaying it because of a problem.”
A transcription app hears a defect. Voice writing hears a spec: tell John, slip a week, numbers not ready, do not sound like a fire drill.
The demo (copy this)
You say
“uh can you tell Sarah that I looked at the deck and the second section needs another pass but overall it's looking good, keep it positive”
Lazur pastes
Hey Sarah, I reviewed the deck. The second section could use another pass, but overall it’s looking really good.
Then you send. Almost no keyboard. That is the “zero editing” promise: say it once; don’t edit it. Not a vague “hours saved per week.”
Another one, Slack:
You say
“hey Mike, can you move our meeting to Thursday afternoon, same agenda”
Paste
Hey Mike, could we move our meeting to Thursday afternoon? Same agenda works for me.
Gmail: “reply and say I'll review it tomorrow, keep it warm.” Cursor: “refactor this to use zustand and add error handling for failed logins.” Same loop. Intent Mode is when the speech is an instruction about this thread, not a draft of the reply.
Why “sounding spoken” is not a user failure
Filler, asides, and “the thing” instead of the project name are how humans think out loud. Forcing people to speak like a prompt box is the old enemy: translating thoughts into computer instructions. The ambient writing bet is the opposite.
Aqua and Wispr will keep getting better at pretty transcripts. If the paste is still the ramble with commas, you have not left transcription. See Aqua Voice vs Lazur.
How to practice this week
- Do not clean up your speech on purpose.
- Put the cursor in the real app, emails by voice.
- Include tone in the brief (“keep it positive,” “casual,” “not a problem”).
- Score whether you still rewrite the paste.
Download Lazur and run the Sarah-deck sentence on a real recipient you would have emailed anyway.
FAQ
- Can dictation handle messy speech?
- Basic dictation will type the mess, maybe with punctuation. Voice writing and Intent Mode extract task, tone, and facts, then write the message you would have typed. That is how Lazur is built.
- How do I dictate without sounding like a transcript?
- Stop dictating the letter. Speak the brief: who, ask, tone, constraint. Let rewrite or Intent Mode produce the prose. See how to write emails by voice for the same pattern.
- What is an example of messy speech vs finished writing?
- You say: ‘uh can you tell Sarah that I looked at the deck and the second section needs another pass but overall it's looking good, keep it positive.’ Lazur pastes: ‘Hey Sarah, I reviewed the deck. The second section could use another pass, but overall it’s looking really good.’