Skip to main content

Brand Voice

Teach UniPulse's AI your brand's unique writing style so every generated caption, hashtag set, and content calendar entry sounds authentically you. Brand Voice is the backbone of consistent AI output across all generation features.

The BrandVoice Model​

Each Brand Voice is a structured profile that the AI references during every generation call:

BrandVoice
├── name              → Display name (e.g., "Acme Corp Voice", "Casual Fitness")
├── description       → Brief description of the brand's identity
├── toneAttributes    → JSON: formality, humor, energy, authority levels
├── vocabulary        → JSON: preferred terms, industry jargon, words to avoid
├── writingStyle      → JSON: sentence structure, paragraph length, formatting
├── samplePosts       → JSON: array of real posts used as style references
├── isDefault         → Boolean: auto-apply this voice to all AI generations
└── workspaceId       → Linked to a specific workspace
info

Brand Voice is workspace-scoped. Each workspace can have multiple Brand Voices, but only one can be set as isDefault: true. The default voice is automatically applied to all AI generations unless you explicitly select a different one.

Setting Up a Brand Voice​

Navigate to Settings > Brand Voice and choose your setup method:

Method 1 — Sample Posts (Recommended)​

The most effective way to train the AI. Provide 5–10 real posts that represent your brand at its best.

  1. Click Create Brand Voice
  2. Select Paste Examples
  3. Add your best-performing posts — these are stored in the samplePosts JSON array
  4. The AI analyzes patterns across your samples to extract tone, vocabulary, and style
  5. Review the auto-detected attributes and adjust as needed
tip

Choose posts that got high engagement — they represent what resonates with your audience. Include a mix of content types (promotional, educational, entertaining) for a well-rounded voice profile.

Method 2 — URL Import​

Import content directly from your existing social profiles:

  1. Select URL Import
  2. Provide links to your social profiles or specific posts
  3. UniPulse fetches and analyzes the content
  4. Review the extracted Brand Voice attributes

Method 3 — Manual Configuration​

Define every attribute by hand for maximum control:

  1. Select Manual Config
  2. Configure each attribute section (see below)
  3. Optionally add sample posts as references

What the AI Learns​

Tone Attributes (toneAttributes JSON)​

Defines the emotional and stylistic character of your brand's communication:

AttributeSpectrumExample
FormalityCasual ↔ Formal"Hey there!" vs. "Dear valued customers"
HumorSerious ↔ PlayfulStraightforward facts vs. witty observations
EnergyCalm ↔ EnthusiasticMeasured statements vs. exclamation-heavy copy
AuthorityPeer-level ↔ ExpertConversational advice vs. authoritative guidance

Vocabulary (vocabulary JSON)​

Controls word choice across all AI output:

CategoryPurposeExample
Preferred termsWords and phrases to use frequently"community" instead of "users"
Industry jargonDomain-specific terminology"ROI", "conversion funnel", "reps"
Branded phrasesYour signature expressions"Built different.", "The Acme way"
Words to avoidTerms that don't fit your brand"cheap", "basically", "synergy"

Writing Style (writingStyle JSON)​

Governs the structural patterns of generated text:

PatternWhat It Controls
Sentence lengthShort punchy lines vs. flowing long-form sentences
Paragraph structureSingle-line paragraphs vs. dense blocks
Emoji usageYour typical emoji patterns and frequency
Hashtag styleBranded hashtags, count preferences, placement
CTA patternsHow you typically drive action (direct, subtle, question-based)
FormattingUse of line breaks, bullet points, capitalization

Sample Posts (samplePosts JSON)​

An array of real posts that serve as concrete style references. The AI uses these for pattern matching — they're the most influential part of the Brand Voice configuration.

Using Brand Voice​

Once configured, Brand Voice integration is automatic across all AI features:

FeatureHow Brand Voice Is Used
AI Caption GenerationCaptions match your tone, vocabulary, and style
Caption RewritingRewrites preserve your brand identity
Hashtag GenerationIncludes your branded hashtags and preferred count
CTA GenerationCTAs follow your action-driving patterns
Content CalendarAll calendar entries use your voice via brandVoiceId
Content RepurposingTransformed content maintains brand consistency

Look for the Brand Voice: Active indicator in the Composer to confirm your default voice is being applied.

Multiple Brand Voices​

Workspaces can create multiple Brand Voice profiles for different use cases:

Use CaseExample
Product linesDifferent voice for each product (e.g., premium vs. budget line)
Content typesFormal voice for announcements, casual for behind-the-scenes
PlatformsSlightly different tone for LinkedIn vs. Instagram
Agency clientsEach client workspace gets its own Brand Voice

Setting the Default​

Only one Brand Voice per workspace can have isDefault: true. To change the default:

  1. Go to Settings > Brand Voice
  2. Click the star icon next to the voice you want as default
  3. The previous default is automatically unset

When generating content, you can also manually select a non-default voice from the dropdown in the Composer.

Tips for a Strong Brand Voice 💡​

  1. Quality over quantity in samples — 5 excellent sample posts beat 20 mediocre ones
  2. Include variety — Mix promotional, educational, and entertaining posts in your samples
  3. Be specific with vocabulary — The "words to avoid" list is just as important as preferred terms
  4. Review periodically — As your brand evolves, update your Brand Voice to match
  5. Test with generation — After setup, generate a few test captions to verify the AI captures your style; refine attributes based on the output

Behind the Scenes​

When any AI endpoint is called (caption generation, rewriting, hashtag generation, calendar generation, etc.), the system:

  1. Checks if a brandVoiceId is specified in the request (or falls back to the workspace default where isDefault: true)
  2. Loads the full BrandVoice record including all JSON fields
  3. Injects the tone attributes, vocabulary, writing style, and sample posts into the Gemini 2.5 Flash prompt
  4. The AI uses these as style constraints alongside your generation parameters

This means Brand Voice doesn't just influence the output — it fundamentally shapes the prompt that drives generation.