Leo: Creator Agent
Practical recipes for generating creatives using the Vinci Creator agent. Each recipe includes a Purpose, Prompt example, and Result summary, with insights and improvement ideas for future iterations.
🧠 Vinci Creator Agent Cookbook
Practical recipes for generating creatives using the Vinci Creator agent.
Each recipe includes a Purpose, Prompt example, and Result summary, with insights and improvement ideas for future iterations.
🧩 1. Retrieve Brand Data and Generate Visuals
Purpose:
Fetch brand color palette from the vector database, then use it to generate a poster or video aligned with brand identity.
Prompt Example:
"What colors are used in my brand guidelines?"
"Let's create an ad for apparel using these guidelines."
Results:
Brand colors were fetched and used to generate an image: Generated Poster
Remarks:
Workflow successfully combined data retrieval + creative generation. It can be replicated using a simple image generation tool.
🧱 2. Upload Brand Guidelines
Purpose:
Store brand assets and documents in the brand knowledge base.
Attachments:
Results:
File uploaded and saved to brand guidelines memory collection.
🧍♂️ 3. Save User Information to Personal Memory
Purpose:
Enable user personalization through persistent memory.
Prompt Example:
"My name is Eric, commit this to memory."
Results:
User information stored in personal memory collection.
Ideas:
Link personal memory to campaign preferences (tone, industry, style).
🔍 4. Fetch Stored User Information
Purpose:
Retrieve saved user or brand data from memory.
Prompt Example:
"What is my name?"
Results:
System fetches stored data ("Eric").
🧾 5. Update Brand Guidelines
Purpose:
Modify stored brand information dynamically.
Prompt Example:
"We are shifting our motto to xyz, save this to brand guidelines."
Results:
Brand memory updated with new motto.
🎨 6. Generate Marketing Visuals
Purpose:
Create image-based marketing assets aligned with campaign ideas.
Prompt Example:
"Make a Pinterest-style visual showing '5 ways to boost video engagement.'"
Results:
Remarks:
Visual colors followed brand theme. Minor spelling errors detected.
📝 7. Correct Spelling in Generated Images
Purpose:
Edit existing image text to fix spelling mistakes.
Prompt Example:
"Fix spelling mistake for 'seconds' in this image."
Attachments:
Original: Input Image
Results:
Corrected: Output Image
Remarks:
Simple correction workflow. Works best for text-focused visuals.
📺 8. Generate a YouTube Thumbnail
Purpose:
Create attention-grabbing YouTube thumbnails using creative composition.
Prompt Example:
"Generate a YouTube thumbnail — half face of a surprised creator, half interface of our product."
Results:
Image generated combining human and UI elements.
🧑🎨 9. Change Image Backgrounds
Purpose:
Perform background color replacement in generated assets.
Prompt Example:
"Change the background from red to blue."
Results:
Original: Red Background
Modified: Blue Background
🏭 10. Studio Workflow — Apparel Ad Campaign (REVISED)
Purpose:
Automate full-scale ad campaign generation using user-provided visuals and produce a consistent set of social images, short videos, and banners.
Prompt Example:
“Generate a full ad campaign for our new apparel launch, including social media posts, a short video, and banner ads.”
Attachments (user-provided):
- Apparel image B: https://storage.googleapis.com/vinci-dev/user_2vif6iZcm907SpANz9idNlneL7P/general_uploads/20251016_163328_6610-ladies-cvc-t-shirt-1660594041.jpg All Generated Assets (explicit list):
Social Posts / Images
- Image 1: https://storage.googleapis.com/vinci-dev/generated_images/66220690-8d7b-450a-8f6d-00c210979c94.pngSource / Notes: Derived from Apparel image A — hero lifestyle shot; caption idea included.