> ## Documentation Index
> Fetch the complete documentation index at: https://docs.nomiq.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Nomiq AI Best Practices for Exceptional Brand Results

> Master the most effective techniques for writing briefs, iterating strategically, refining visuals, and integrating Nomiq via API for exceptional results.

Generating a basic brand in Nomiq takes five minutes. Generating an *exceptional* brand — one that is differentiated, accessible, and production-ready — requires a deliberate approach at every stage of the workflow. This guide compiles the most effective techniques used by power users and agencies, organized into four areas: writing the brief, iterating effectively, refining visuals, and integrating via API.

## Pre-Generation Checklist

Before you click **Generate**, confirm the following. Skipping any item frequently results in generic outputs that require multiple additional rounds of iteration.

* [ ] Your prompt includes a specific **industry or business type** (not just a vague category like "tech")
* [ ] Your prompt includes a **target audience** description with demographic and psychographic detail
* [ ] Your prompt includes **three or more emotional adjectives** that reinforce each other
* [ ] You have added at least **one negative constraint** to eliminate the most predictable industry cliché
* [ ] Your prompt does **not** contain metaphors or pop-culture references that the engine may interpret literally
* [ ] You have selected the correct **Logo Style** (logotype, symbol + text, icon only) before generating
* [ ] You know which brand name or concept you want to **lock immediately** after the first generation

***

## Writing the Brief

### Do: Be Specific About Your Audience

The Strategy Model uses audience description to calibrate brand voice, and the visual sub-models use it to set complexity, energy level, and color temperature. Generic audience labels produce predictable defaults. Specific descriptions produce differentiated outputs.

| ❌ Don't        | ✅ Do                                                                              |
| :------------- | :-------------------------------------------------------------------------------- |
| "Young people" | "Urban women aged 22–32 who follow independent fashion and value sustainability"  |
| "Businesses"   | "Operations directors at mid-market logistics companies with 500–2,000 employees" |
| "Gamers"       | "Competitive FPS Twitch streamers aged 18–26 who follow pro esports circuits"     |

### Do: Use Emotional Adjectives to Drive Color and Font Selection

The Color and Typography models map emotional adjective clusters to visual spaces. Three to five well-chosen adjectives consistently outperform lengthy descriptive paragraphs. Choose adjectives that form a coherent emotional cluster — avoid contradictory combinations like "aggressive and delicate."

```text theme={null}
Strong adjective clusters:

Clinical, authoritative, precise
→ Cool grays + indigo, geometric sans-serif, tight tracking

Warm, approachable, playful
→ Warm amber + cream, rounded sans-serif, generous leading

Gritty, rebellious, anti-establishment
→ High-contrast black/white + muted accent, condensed display font

Luxurious, refined, exclusive
→ Monochrome + one warm accent, high-contrast serif logotype
```

### Don't: Use Vague Language or Metaphors

The Semantic Extraction step processes your prompt literally. Metaphors, cultural references, and abstract comparisons produce unpredictable results. Translate every abstract concept into concrete design language before submitting.

| ❌ Vague / Metaphorical              | ✅ Concrete Design Language                                                               |
| :---------------------------------- | :--------------------------------------------------------------------------------------- |
| "Make it feel like a warm hug"      | "Use warm amber tones and a friendly, rounded sans-serif"                                |
| "It's the Apple of its industry"    | "Use a minimal monochrome palette with one accent and clean geometric typography"        |
| "Neon and vintage at the same time" | Choose one: "High-saturation neon palette on black" or "Desaturated vintage muted tones" |
| "Cool startup vibes"                | "Saturated single-color palette on white, bold geometric sans-serif"                     |

<Tip>
  Add negative constraints to eliminate the most predictable visual clichés for your industry. Healthcare brands default to blue crosses, tech brands default to gradient blues, eco brands default to green leaves. Name and explicitly ban the clichés you want to avoid.
</Tip>

***

## Iterating Effectively

### Do: Lock Strategy Before Regenerating Visuals

This is the single most impactful practice in the entire Nomiq workflow. The Strategy Model output — brand name, tagline, voice matrix — is the semantic seed for every visual sub-model. If you attempt to fix your logo or palette while strategy is unlocked or incorrect, the visual models will continue producing outputs biased toward the wrong tone.

**Always follow this sequence:**

1. Generate once.
2. Review the strategy output first.
3. If strategy is correct, lock it immediately.
4. Only then evaluate and iterate on visuals.

### Do: Use Element-Level Regenerate Buttons

Every brand element in Brand Studio has its own **Regenerate** button. Use these instead of the global **Regenerate All** control whenever you have at least one element worth keeping.

| Scenario                           | ❌ Wrong        | ✅ Right                                      |
| :--------------------------------- | :------------- | :------------------------------------------- |
| Like the palette, dislike the logo | Regenerate All | Lock palette → Regenerate logo only          |
| Like the logo, dislike typography  | Regenerate All | Lock logo → Regenerate typography only       |
| Like the name, dislike the tagline | Regenerate All | Lock name → Iteration Prompt for new tagline |

### Don't: Over-Iterate in a Single Prompt

Trying to fix colors, typography, and logo symbol in one iteration prompt produces incoherent results. The engine attempts to satisfy all constraints simultaneously and often satisfies none of them well. Fix one element at a time, lock it, then move to the next.

<Tip>
  Treat each iteration like a commit in version control — make one meaningful change, verify the result, lock it, then proceed. Nomiq saves a snapshot after every iteration, so you can always revert to a previous state using **Version History** (the clock icon in the top-right of Brand Studio).
</Tip>

### Don't: Rewrite Your Full Prompt After a Partial Success

If you like your logo symbol but dislike your color palette, do not discard the prompt and start over. Rewriting the prompt resets the semantic seed and all four sub-models regenerate from scratch, including the logo you liked. Use the Iteration Prompt bar instead and lock the elements you want to keep.

***

## Visual Refinement

### Do: Test Your Logo at Small Sizes

A logo that looks impressive at 400px often becomes illegible at 32px — the size it appears in browser tabs, app icons, and notification badges. After locking your logo symbol, use the **Preview Panel** in Brand Studio to test it at 16px, 32px, and 64px. If the symbol loses its clarity at 32px, use the Iteration Prompt to request a simpler, more geometric form.

```text theme={null}
Example iteration prompts for logo simplification:

"Simplify the logo symbol to use no more than three geometric shapes."
"Make the logo symbol work as a monochrome silhouette at 32x32 pixels."
"Remove any fine lines or detail from the symbol — it must read clearly at small sizes."
```

### Do: Verify Color Accessibility Before Exporting

Nomiq's Validation Layer enforces WCAG AA contrast on AI-generated palettes, but manual color overrides bypass this check until you save. After any manual hex code edit, confirm that all foreground/background combinations in the palette pass the contrast check. Red warning icons in the color panel indicate failing pairs.

| WCAG Standard   | Minimum Contrast Ratio | Applies To                         |
| :-------------- | :--------------------- | :--------------------------------- |
| AA Normal Text  | 4.5:1                  | Body copy, UI labels               |
| AA Large Text   | 3:1                    | Headings ≥ 18pt, bold ≥ 14pt       |
| AAA Normal Text | 7:1                    | Enhanced accessibility requirement |

### Don't: Add Text Inside the Logo Symbol

The Vector Model structures all logos as `[SVG Symbol] + [Brand Name Text]`. It cannot reliably render legible, correctly spelled text inside the vector graphic. If you need a wordmark or lettermark style, select the **Logotype** layout option before generating — this instructs the Vector Model to build the brand name itself as the primary visual asset instead of constructing a separate symbol.

***

## API Integration

### Do: Always Pass the Seed for Iterative API Calls

When building an integration that refines brands across multiple API calls, include the `seed` value from the initial generation response in every subsequent request. The seed preserves the Context Window across calls. Without it, each request is treated as a new project.

```json theme={null}
{
  "prompt": "Make the palette significantly darker and reduce color variety to two tones.",
  "seed": "a3f92c1d-7b44-4e8a-b1c0-9f3d2e7a0c55",
  "locked_elements": ["strategy", "typography", "logo"]
}
```

### Do: Append the Brand Voice Matrix to Copy Generation Prompts

When using the Nomiq API alongside a general-purpose LLM to generate marketing copy, always include the `brand_voice_matrix` object from the Nomiq brand kit in your LLM prompt payload. This prevents the LLM from defaulting to generic corporate language and keeps copy tonally consistent with the generated brand identity.

```json theme={null}
{
  "messages": [
    {
      "role": "system",
      "content": "You are a copywriter. Adhere strictly to the following brand voice: {{ brand_voice_matrix }}"
    },
    {
      "role": "user",
      "content": "Write a 60-word homepage hero headline and subheading."
    }
  ]
}
```

### Don't: Use Over-Specific API Constraints

Passing exact hex codes, exact font names, or pixel-level vector coordinates as hard constraints in API prompts frequently causes the Validation Layer to reject the output and force regeneration loops. Provide directional constraints instead and let the engine calculate exact values within that direction.

| ❌ Over-Specific            | ✅ Directional                                                         |
| :------------------------- | :-------------------------------------------------------------------- |
| `"color": "#1A1A2E"`       | `"palette_direction": "deep navy, near-black backgrounds"`            |
| `"font": "Helvetica Neue"` | `"typography_direction": "clean geometric sans-serif, medium weight"` |
| `"logo_stroke_width": 1.5` | `"logo_style": "thin-line geometric symbol"`                          |

<CardGroup cols={2}>
  <Card title="Prompt Engineering" icon="terminal" href="/ai-engine/prompt-engineering">
    Deep-dive into prompt anatomy with real-world examples across multiple industries.
  </Card>

  <Card title="Iterations & Editing" icon="rotate-right" href="/ai-engine/iterations-and-editing">
    Learn the full iteration workflow, locking system, and manual editor in detail.
  </Card>
</CardGroup>
