---
title: "Review names, numbers and negation first"
description: "Prioritize meaning-changing details when reviewing a spoken draft: who, how much, when, which object and whether an action is allowed."
slug: "names-numbers-negation"
category: "voice-intent"
language: en
editorial_status: published
capability_status: guide
reviewed_at: 2026-10-13
keywords: ["speech to text errors","AI prompt accuracy","review dictated numbers"]
---

# Review names, numbers and negation first

Prioritize meaning-changing details when reviewing a spoken draft: who, how much, when, which object and whether an action is allowed. Correct these before polishing style.

## A fast review pass
Look for personal or product names, quantities, dates, units, filenames and “not” or “only.” These details can change the result even when the rest of the sentence is excellent. Compare them with your actual intended request, not merely with plausible grammar.

For “Use two samples; do not change the live version,” losing “not” is more significant than an extra filler. For “Replace the 15 mm dimension,” a changed unit can invalidate the request. Paste the exact identifier when spelling it aloud repeatedly fails.

## Keep uncertainty visible
If the draft contains an unverified fact, keep the qualifier or ask the agent to verify it. Transcription and refinement should not promote a hypothesis into a confirmed specification.

CueSpec’s current workflow separates draft creation from explicit sending so you can perform this check. It does not remove the need to confirm the selected conversation. A correct sentence in the wrong task can still trigger the wrong work.

## Related reading

- [Local transcription and optional text refinement](/transcription-vs-refinement/)
- [Mix speech with typed technical identifiers](/dictate-technical-identifiers/)
- [Send your first reviewed reply](/first-reviewed-reply/)
- [CueSpec product facts at a glance](/product-facts/)