Creative Technology
/
AI Systems
/
Content Production
CREATIVE IDEATION ENGINE — Capture, Generate, Instantly
(
1
)
/ About the project
THE PROJECT
Built a multi-input creative ideation engine for Studio Bubull, a creative studio that needed a faster way to capture raw inspiration and transform it into structured creative direction. The system accepts three input types: a photo, a voice note, or typed text, sent directly from a phone via Telegram. From a single moment of inspiration, the pipeline generates a complete creative package including a creative direction brief, a brand style guide, and three AI-generated mood board images, all delivered to Google Drive with a Slack notification to the team.
THE SCOPE
Owned the full architecture and build. Designed the n8n workflow around a smart routing model that detects the input type on arrival and dispatches it down the correct processing path. Photos are downloaded via the Telegram API, converted to base64, and passed directly to the Claude API for rich visual analysis. Voice notes are downloaded and transcribed in real time using the OpenAI Whisper API. Text input is extracted and normalized.
All three paths converge at a single merge node before hitting the Claude API, which generates a creative direction brief, brand style guide, and three distinct image generation prompts from the input. Those prompts are dispatched to the Gemini image generation API, producing three mood board images covering landscape, portrait, and editorial formats. All outputs are uploaded to Google Drive and the team is notified via the Slack API with a direct link to the folder.
THE SOLVE
Studio Bubull needed creative ideas to move at the speed of inspiration, not the speed of a production schedule. A designer spots something on the street, hears a sound, or has an idea in the shower. By the time they are back at a desk, the feeling is gone. This system closes that gap entirely. A photo, a voice memo, or a quick text sent to Telegram is all it takes. Within seconds, a fully structured creative direction package is waiting in Google Drive: brief, brand guide, and mood board images ready to present, share, or build from. Inspiration captured, interpreted, and packaged before the moment passes.
HOW IT WORKS
Step 1 — Multi-Input Capture
A user sends text, image, or voice via Telegram, triggering the workflow.
Step 2 — Input Routing
n8n detects the input type and routes it to the correct processing path.
Step 3 — Input Normalization
All inputs are converted into text:
Text passes through
Images are described by Claude
Voice is transcribed via Whisper
Step 4 — Creative Direction Generation
Claude generates a creative brief, brand style guide, and 3 image prompts.
Step 5 — Image Generation
Prompts are sent to Gemini to generate 3 cohesive images.
Step 6 — Output Assembly
All assets are merged into a single structured package.
Step 7 — Document Packaging
Creative direction and brand guide are formatted as documents.
Step 8 — Google Drive Delivery
All files are uploaded to Google Drive automatically.
Step 9 — Slack Notification
Slack sends a notification when everything is ready.
Step 10 — Done
End-to-end creative direction delivered in seconds.
OTHER
PROJECTS
Get in touch