Every brand creator eventually reaches the same moment.
They have been using AI tools for a few months. They know their way around Midjourney. They have produced some images they are proud of. But every session still feels like starting from scratch. There is no system. No templates. No documented process. Just a blank prompt field and the hope that today's session goes better than yesterday's.
A brand AI toolkit solves this. It is the complete set of tools, configurations, templates, and documented processes that turns ad-hoc AI experimentation into a repeatable professional content production system.
This guide builds yours from scratch. By the end of this weekend you will have a toolkit that produces consistent on-brand content efficiently every time you sit down to create.
A complete brand AI toolkit has seven components. Each one is a specific document, file, or configured tool that you build once and use indefinitely.
- Component 1 - Brand visual foundation document
- Component 2 - Configured AI tool accounts
- Component 3 - Prompt template library
- Component 4 - Asset folder system
- Component 5 - Custom brand GPT
- Component 6 - Reference image library
- Component 7 - Quality checklist
Build all seven and you have a toolkit. Build three or four and you have a collection of useful things that does not quite work as a system. The components depend on each other and the toolkit works best when all seven are present.
Build three or four and you have a collection of useful things that does not quite work as a system.
Component 1 - Brand visual foundation document
This is the most important component. Everything else in the toolkit draws from it. Without it your templates produce inconsistent results and your tool configuration has no direction.
Your brand visual foundation document is a single reference file - one or two pages - that defines your brand's visual language in AI prompt-ready terms.
It contains five sections:
Color vocabulary: 4 to 6 specific descriptive color phrases. Not hex codes. Phrases like 'deep burgundy with subtle violet undertones, rich and saturated, matte velvet finish' that AI generators actually understand and reproduce accurately.
4 to 6 specific descriptive color phrases. Not hex codes.
Lighting signature: one complete lighting phrase that defines your brand's characteristic light. Example: 'soft diffused natural window light from the left, warm golden undertones, gentle graduated shadow falling to the right, single soft catchlight'. This phrase goes into every prompt.
Material vocabulary: 4 to 6 surface and prop words that consistently appear across your brand imagery. The physical world your brand inhabits.
Composition rules: 3 simple rules about framing, subject position, and negative space.
Mood keywords: 5 to 8 single words capturing your brand's emotional quality.
How to build it fast: use the AI Visual Style Finder to answer 12 questions and get your named visual style, color direction, lighting signature, material vocabulary, and prompt vocabulary all in one output. Then use the Brand Color Palette Translator to convert your hex codes into descriptive color phrases. And use the Brand Vocabulary Builder to expand your word bank.
Document everything in a Google Doc or Notion page. Title it Brand Visual Foundation. This is your toolkit's source of truth.
Your AI tools need to be set up for your brand before they are useful at a professional level. Configuration done once saves significant time on every future session.
Midjourney configuration checklist:
Personalization profile: complete at minimum 200 rankings, ideally 500 or more, with your brand aesthetic as the ranking criteria. Activate with --p on all brand content prompts. This is the highest-leverage configuration step available in Midjourney.
Default settings: in Midjourney settings set your default model version to V8.1, your default stylize value to your preferred range (100 to 200 for most brand photography), and enable remix mode for flexible iteration.
Brand moodboard: create a Midjourney project titled with your brand name. Save your 20 best generated images to this project. Use images from this board as style references (--sref) to anchor new generations.
ChatGPT configuration checklist:
Custom GPT: build a custom GPT (covered in Component 5 below) with your brand visual foundation uploaded as context. This becomes your primary prompt generation tool.
Higgsfield configuration checklist:
Preferred presets: identify and note the 3 to 5 camera presets that best match your brand motion aesthetic. Write these into your toolkit document so you use them consistently rather than experimenting from scratch each session.
Reference image upload: upload your product reference images (PNG with transparent background) to Higgsfield so they are ready for Nano Banana placement without uploading each time.
Component 3 - Prompt template library
Your prompt template library is a document containing one reusable prompt template for each content type you produce regularly. These templates have your brand visual foundation built in and only require you to fill in the specific subject for each use.
Standard templates to build:
Product hero template: for your main product showcase images. Centered composition, strong lighting, premium feel. Your highest-use template.
Product lifestyle template: for product in context and use. Looser composition, environmental depth, aspirational mood.
Flat lay template: for top-down styled arrangements. Overhead camera specification, props and surface description, brand palette.
Brand mood template: for atmospheric and abstract brand imagery. Higher stylize value, more artistic language, mood-focused.
Texture and detail template: for close-up material and craftsmanship shots. Macro lens specification, shallow depth of field, material-focused language.
Video base template: for generating still images to be animated. Motion-friendly language, clear atmospheric elements, composition suited to camera movement.
Template format: write each template with the brand system elements fixed and a clear [VARIABLE] placeholder for the specific subject content. Store all templates in a single Google Doc titled Prompt Template Library. Keep the most recently refined version of each template at the top of its section.
Component 4 - Asset folder system
A folder system that makes any asset findable in under 30 seconds.
Recommended structure:
- Brand Name
- Brand Foundation
- brand-visual-foundation.pdf
- color-vocabulary.txt
- prompt-template-library.txt
- quality-checklist.txt
- Reference Images
- product-references (PNG files with transparent backgrounds)
- scene-references (Midjourney scenes for Nano Banana)
- style-references (moodboard images)
- logo-files (PNG files for Nano Banana logo placement)
- Generated Content
- 2026
- July
- product-hero
- product-lifestyle
- flat-lay
- brand-mood
- video-base-images
- published
- archive
Filing discipline: generated images go into their content type folder immediately. Selected images move to published after posting. Rejected but potentially useful images go to archive. Never leave images unsorted in downloads.
Component 5 - Custom brand GPT
A custom GPT is a version of ChatGPT configured with your specific brand context. Once built it becomes your most efficient prompt generation tool - you describe what you need in plain language and it generates a full on-brand prompt using your exact vocabulary, lighting signature, and material preferences.
How to build it:
- Go to chat.openai.com and click Create a GPT in the left sidebar. Requires ChatGPT Plus.
- In the configuration panel set the name to your brand name plus Prompt Writer. Example: Aesop Prompt Writer or Minimalist Skincare Prompt Writer.
- In the instructions field paste your complete brand visual foundation document. Then add these specific instructions:
You are a brand prompt writer for [brand name]. When the user describes an image they want to create you generate a complete Midjourney prompt using the brand visual foundation above. Always include the brand color vocabulary, lighting signature, material vocabulary, and mood keywords in every prompt. Always follow the 6-layer framework: content type, subject, colors and materials, composition, lighting, camera and lens. Always end with --ar [your default ratio] --v 8.1 --stylize [your default value] --p. Return only the prompt text with no explanation.
Upload your brand visual foundation document as a file reference.
Test by typing: 'I need a product hero shot of our new face oil in a dark glass bottle'. The GPT should return a complete on-brand Midjourney prompt without you having to specify any brand elements.
Once working this GPT reduces prompt writing time from 5 to 10 minutes to under 60 seconds for any content type.
This GPT reduces prompt writing time from 5 to 10 minutes to under 60 seconds for any content type.
Component 6 - Reference image library
A curated set of images that define your brand aesthetic and serve as visual references for AI generation.
Four types of reference images to collect:
Product references: clean PNG files of every product with transparent backgrounds. One reference per product SKU. Used for Nano Banana product placement and Midjourney Omni Reference.
Scene references: your best Midjourney-generated empty scenes organized by type. Marble scenes, wood scenes, lifestyle interiors, outdoor settings. These are your reusable scene library for Nano Banana placement workflows.
Style references: 10 to 20 images that best represent your brand aesthetic. Can be your own generated images, editorial tear sheets, competitor references, or any images that capture your visual direction. Used as --sref references to anchor new generations.
Logo files: high-resolution PNG files of every brand logo, wordmark, and icon. Used with Nano Banana for accurate logo placement in product imagery.
Organize all reference images in the Reference Images folder in your asset system. Name files descriptively so any reference is findable quickly.
Component 7 - Quality checklist
A written checklist of criteria that every image must meet before entering your content library. Having this written down removes subjectivity from the selection process and ensures consistent quality standards even when you are working quickly.
Standard quality checklist:
Brand alignment: does this image feel like our brand? Would someone who knows our brand recognize this as ours without seeing our name?
Color accuracy: does the image use our color vocabulary? Are the tones consistent with our defined palette?
Lighting consistency: does the lighting match our lighting signature? Is the light direction, temperature, and quality consistent with our brand standard?
Technical quality: is the image sharp where it should be sharp? Are there any AI artifacts, distorted elements, or blurry patches?
Subject accuracy: if the image features a product is it accurate in shape, color, and proportion? Are there any obvious AI errors in the main subject?
Composition integrity: does the composition follow our composition rules? Is the framing intentional?
Platform fit: is the aspect ratio correct for the intended platform? Does it work at display size?
Any image that fails any criterion does not enter the content library. It goes to archive and the prompt is refined before the next generation attempt.
Save this checklist as a plain text file in your Brand Foundation folder. Print it and keep it next to your computer if that helps you apply it consistently.
Saturday morning (2 hours): build your brand visual foundation. Use the AI Visual Style Finder, Brand Color Palette Translator, and Brand Vocabulary Builder. Document everything in your Brand Visual Foundation document.
Saturday afternoon (2 hours): configure your tools. Complete 200 Midjourney rankings. Set up default settings. Identify Higgsfield presets. Prepare product reference images with backgrounds removed.
Saturday evening (1 hour): set up your asset folder system. Create all folders. Move any existing assets into the right locations.
Sunday morning (2 hours): build your prompt template library. Write and test one template per content type. Refine until each produces consistent on-brand results across 3 to 5 test generations.
Sunday afternoon (2 hours): build your custom brand GPT. Test it with 5 different content descriptions. Refine the instructions until outputs are consistently on-brand. Collect and organize your reference image library.
Sunday evening (30 minutes): write your quality checklist. Save it to your Brand Foundation folder.
Total time investment: approximately 9 to 10 hours across one weekend. After that your toolkit runs indefinitely with only occasional maintenance - updating templates when you learn something new, adding to your personalization profile rankings, saving strong new outputs to your moodboard and reference library.
Start building your visual foundation today
The brand visual foundation document is the most important component of your toolkit and the right place to start. Everything else builds on top of it.
Our AI Visual Style Finder gives you the foundation in 12 questions and under 5 minutes. You get your named visual style, complete color direction, lighting signature, material vocabulary, composition tendency, and prompt vocabulary bank - everything you need to write your first brand visual foundation document.
CropUp Ai