Voice DNA is SalesSheet's AI-powered writing style analysis feature. It scans your sent emails from Gmail and messages from Slack to build a personalized communication profile - capturing your tone, vocabulary, formality level, and signature phrases. Once your profile is built, every AI-drafted email and message automatically matches your natural writing style.
Instead of generic, robotic-sounding AI drafts, Voice DNA ensures that messages composed by the AI assistant feel authentically yours. Your contacts will never know the difference between an email you wrote yourself and one the AI drafted for you. This article explains how Voice DNA works under the hood, what data it uses, how your privacy is protected, and how plans differ.
How Voice DNA Works
Voice DNA collects writing samples from your connected accounts - Gmail sent emails and Slack messages you've posted. These samples are analyzed by AI to extract patterns in how you communicate: the words you favor, your level of formality, your typical greeting and sign-off styles, and recurring phrases that make your writing distinctive.
The analysis produces a style profile that includes a tone summary (e.g., "Professional yet approachable"), a formality score on a Casual-to-Formal scale, individual traits scored from 0 to 100 (such as Directness, Warmth, Technical depth, and Persuasion), and a list of your signature phrases.
Internally, the process has three stages. First, SalesSheet's server-side integration fetches your sent messages through the Gmail API or Slack API. Second, the collected text is fed into a large language model that specializes in stylistic analysis. The model identifies recurring vocabulary, sentence structures, punctuation habits, greeting formats, and closing patterns. Third, the extracted patterns are distilled into a compact style profile JSON object that the AI assistant references every time it drafts content on your behalf.
Because the profile is a structured data object rather than a copy of your emails, it is lightweight and fast for the AI to apply. The model does not memorize your messages word-for-word; it captures abstract patterns like average sentence length, preference for contractions, and the ratio of questions to statements in your writing.
What Data Is Used
Samples are individual pieces of your writing - sent emails from Gmail or messages from Slack channels. Voice DNA reads only messages you authored, never incoming messages from other people. The more samples analyzed, the more accurate your profile becomes.
For Gmail, SalesSheet fetches messages from your Sent Mail folder using the Gmail API read-only scope. Only the body text of emails you composed is extracted - attachments, images, and recipient-authored quoted text are stripped out before analysis. For Slack, the integration reads messages you posted in channels where the SalesSheet Slack app is installed, excluding messages from other team members, bot messages, and system notifications.
The quality of your profile depends on the volume and variety of your writing samples. An account with 50 diverse sent emails spanning introductions, follow-ups, proposals, and casual check-ins will produce a more nuanced profile than an account with 50 nearly identical templated messages. Voice DNA benefits most from natural, varied communication samples.
Plan Differences
Voice DNA is available on the Pro and Ultra plans. The two plans differ in the number of sources you can connect and the volume of samples analyzed:
- Pro Plan: Gmail analysis only. Up to 50 sent email threads are collected and analyzed. This is enough for a solid baseline profile that captures your primary communication style. Slack analysis is not available on Pro.
- Ultra Plan: Gmail and Slack analysis. Up to 500 Gmail threads and 250 Slack messages are collected, producing a significantly richer and more accurate profile. The larger sample size allows Voice DNA to distinguish between your formal email tone and your conversational Slack tone, and the AI can blend both appropriately depending on the context of the draft.
On either plan, you can re-run the analysis at any time to refresh your profile with newer writing samples. If you upgrade from Pro to Ultra, simply navigate to the Voice DNA page and run the Slack analysis to enhance your existing Gmail-based profile.
Your Style Profile
After analysis, your Voice DNA profile contains several components:
- Tone Summary: A natural-language description of your overall communication style, like "Direct and confident with occasional warmth."
- Formality Score: A slider from Casual to Formal showing where your writing typically falls.
- Traits: Individual dimensions scored 0–100, including Directness, Warmth, Technical depth, and Persuasion.
- Signature Phrases: Recurring expressions you use frequently, displayed as teal pill badges (e.g., "Let's circle back," "Happy to help").
- Vocabulary Patterns: Jargon, transitions, and word choices that define your style.
The completeness bar on the Voice DNA page reflects how much data your profile is built from. A profile at 30% completeness has enough data to produce reasonable drafts, but a profile above 70% will capture subtler aspects of your style such as how you handle objections, deliver bad news, or adjust formality based on context.
Privacy and Data Handling
Your writing data stays in your SalesSheet account and is never shared with other users or used to train AI models. Voice DNA analysis runs through secure Supabase Edge Functions, and the resulting style profile is stored only in your account. You can re-analyze at any time to update your profile, and the raw samples are not permanently stored after analysis - only the extracted style patterns are retained.
Here is a detailed breakdown of the privacy safeguards in place:
- No cross-user sharing: Your Voice DNA profile is private to your account. Team administrators cannot view or export another user's style profile.
- No model training: Your writing samples and the resulting style profile are never used as training data for any AI model. The analysis is inference-only.
- Transient sample storage: Raw email and Slack text is held in memory during the analysis request and discarded once the style profile is generated. No raw message content is persisted to the database.
- Encrypted at rest: The style profile JSON stored in your account is encrypted using the same AES-256 encryption that protects all SalesSheet user data.
- OAuth scopes: Gmail integration uses the read-only scope, meaning SalesSheet can read your sent emails but cannot modify, delete, or send emails on your behalf through the Voice DNA feature. Slack integration uses a similarly limited read scope.
If you ever want to remove your Voice DNA data, you can delete your style profile from the Voice DNA page. This immediately removes the profile from your account and the AI assistant will revert to generic drafting until you run a new analysis.
Consejo Pro
For the most accurate Voice DNA profile, make sure your Gmail account has at least 30–50 sent emails covering different types of communication - introductions, follow-ups, proposals, and casual check-ins. The wider the variety, the better Voice DNA captures your full range of expression. If you recently changed your writing style (for example, adopting a more casual tone for a new role), re-run the analysis to update your profile so AI drafts reflect your current voice.
Importante
Voice DNA requires an active Gmail or Slack OAuth connection. If your Gmail token expires or you revoke access in your Google account settings, your Voice DNA profile will remain intact but you will not be able to re-analyze until you reconnect. Navigate to Settings > Integrations > Gmail to re-authorize if needed.
Solución de Problemas
Why does my Voice DNA profile say "Setup Required"?
This means you haven't run your first Voice DNA analysis yet. Navigate to the Voice DNA page from the sidebar, connect Gmail if you haven't already, and click "Analyze" on the Gmail source card. The analysis typically takes 30–60 seconds depending on how many emails are collected.
Can I use Voice DNA with just Slack (no Gmail)?
Slack analysis is available on the Ultra plan. You can analyze Slack messages independently or in combination with Gmail. However, for the richest profile, we recommend connecting both sources - Gmail captures your longer-form writing while Slack captures your conversational style.
My Voice DNA profile seems inaccurate or doesn't match my style.
Profile accuracy improves with more and more diverse writing samples. If your profile feels off, check your completeness percentage on the Voice DNA page. Profiles below 50% completeness may miss subtleties. Try re-running the analysis after sending more emails, or upgrade to the Ultra plan and connect Slack for additional samples. Also ensure that the Gmail account you connected is your primary work account - if you connected a personal account with very different writing patterns, the profile may not reflect your professional tone.
The analysis failed with an error message.
Analysis errors are typically caused by an expired OAuth token or a temporary network issue. First, check that your Gmail connection is still active by visiting Settings > Integrations > Gmail. If the connection shows as disconnected, re-authorize and try again. If the connection is active, wait a minute and retry - transient server-side errors usually resolve on their own. If the problem persists after multiple attempts, contact support at andres@salessheets.ai with the error message text.