Brand Voice
How Borker encodes your brand identity and injects it into every AI generation prompt.
Brand voice is the configuration that controls how your content sounds. It's separate from topics (what you write about): brand voice is pure style and identity.
Think of it as writing a detailed brief for a new team member on your first day. The more specific you are, the better they'll represent you.
What brand voice is (and isn't)
Brand voice controls:
- Your company identity (who you are, what you do, why you exist)
- Personality traits (e.g., "confident but not arrogant")
- Writing principles (do this, don't do that)
- Terminology (words to use, words to avoid)
Brand voice does NOT control:
- What topics to write about → that's your Topic Pools
- How often to post → that's Scheduling
- The intensity of tone → that's Voice Attributes
Brand voice fields
Company basics
| Field | What to put here |
|---|---|
company_name | Your actual company name |
tagline | Your one-line pitch |
mission | Why you exist: what you're trying to change |
what_we_are | A description of what you do and for whom |
Keep these tight. The AI uses them as context in every prompt.
Personality traits
This is the most important section. Each trait has four parts:
- Trait: a short label ("Direct", "Curious", "Technically Grounded")
- Application: what this trait looks like in practice
- Example Good: a sample sentence that embodies the trait correctly
- Example Bad: a sample sentence that violates the trait
The good/bad examples are critical. They give the AI concrete reference points, not just abstract adjectives.
Example: "Confident, not arrogant"
| Field | Value |
|---|---|
| Trait | Confident, not arrogant |
| Application | Make strong claims when you have evidence. Don't hedge on obvious truths. Don't boast about superiority. |
| Good | "This approach works better because X, Y, Z." |
| Bad | "We're the best in the industry at this." |
Example: "Technically precise"
| Field | Value |
|---|---|
| Trait | Technically precise |
| Application | Use correct terminology. Don't simplify to the point of inaccuracy. |
| Good | "The request is authenticated with a JWT signed using RS256." |
| Bad | "We use a special security token to keep you safe." |
Aim for 3–5 traits. More than that dilutes the signal.
Writing principles
Dos and don'ts with examples. This is where you encode your house style.
Examples:
- Do: Open with the most important point, not context
- Don't: Use passive voice ("It was decided" → "We decided")
- Do: Include one specific, concrete detail per post
- Don't: Use buzzwords: "revolutionize," "game-changing," "disruptive"
Each principle should have an example to make it unambiguous.
Terminology
Use words: Words that feel right for your brand. Industry terms your audience expects. Your product's actual name.
Avoid words: Words that feel off-brand, vague, or that you're tired of seeing AI overuse ("leverage," "delve," "unlock").
Anti-fabrication rule: Brand voice does not grant the AI license to invent facts. Borker's prompts explicitly instruct Claude never to fabricate statistics, quotes, or specific claims. If you want specific data in posts, include it in the topic angle or the post generation prompt.
Industry relationships
Describe your relationship to adjacent companies, categories, and competitors. This helps the AI understand how to position you without making claims you wouldn't make yourself.
Example: "We compete with traditional agencies on price and speed, but we're not trying to replace human creativity. We augment it."
Topic focus areas
High-level content themes that matter to your brand. These inform the AI's framing even when a specific topic isn't one of your listed topics. Keep it to 3–6 areas.
Example focus areas for a B2B SaaS company: "AI in operations," "startup growth tactics," "founder mental health," "product-led growth."
How brand voice is injected
When you trigger a content generation, Borker assembles a system prompt using your brand config:
graph LR
A[Brand Voice Config] --> C[System Prompt Builder]
B[Voice Attributes] --> C
C --> D[Claude API]
E[Topic + Angle] --> D
F[Content Type Format] --> D
D --> G[Generated Content]Your brand voice becomes the AI's standing instructions. The topic and content type are the task. Every generation call follows this structure.
The more specific your brand config, the less the AI has to guess, and the more consistent your output will be.
Tips for effective brand voice setup
- Write personality traits based on actual mistakes. If you've seen AI write something that sounds off for your brand, that's your "bad example."
- Be specific about your audience. "Technical founders" is more useful than "professionals."
- Don't try to be everything. Pick 3–5 traits and commit to them.
- Review the output of a few posts before finalizing. The brand voice config is a working document, not a one-time setup.
- Keep mission statements short. 1–2 sentences. Longer doesn't mean better.