We are manufacturers and sellers of men's clothing. We are developing an internal AI tool to automate the creation of advertising photo and video creatives for Facebook/Instagram. The project is being created for internal use, not for sale.
What needs to be done
Develop a user-friendly Dashboard where the user creates a project, fills out the product card (description, photo, price, target audience, etc.), after which a chain of AI agents is automatically launched, resulting in ready-made advertising photos and videos.
Under the hood
AI Agent 1 — Creative Strategist
AI Agent 2 — Script Generator
AI Agent 3 — Creative Director
Generation of prompts for AI photos.
Generation of prompts for AI videos.
Creation of prompts for each scene.
Automatic sending of prompts to AI for obtaining ready-made images and videos.
All agents must automatically pass results to each other.
Interface
Dashboard.
Project creation.
Product card.
Uploading product photos.
Generation status.
Gallery of ready-made photos.
Gallery of ready-made videos.
Option for re-generation.
Working with colors
It is necessary to implement convenient work with variants of a single product.
For example:
Linen suit
Black
White
Mocha
Graphite
Blue
Functionality:
Uploading one color.
Uploading several colors of one product at once.
Ability to choose which colors participate in generation.
Choice of generation mode:
For example:
The user selects 100 creatives.
If 5 colors are uploaded, the system automatically distributes the generation:
Black — 20
White — 20
Mocha — 20
Graphite — 20
Blue — 20
Or it can create a creative where different colors of the product are automatically shown in different scenes.
Generation settings
The user must be able to choose:
Section "Scripts"
For each ready creative, it is necessary to save:
This is necessary so that the editor can manually use these prompts to generate individual frames, images, or video clips in any AI services.
Technical part
It is necessary to:
connect the OpenAI / Claude / Gemini API;
connect the image generation API;
connect the video generation API;
implement backend with automatic data transfer between AI agents;
make the architecture scalable so that in the future it will be easy to add new AI agents and new AI models without a complete redesign of the project.
When responding, please send:
Important: if you have experience in developing multi-agent AI systems or AI services with integration of multiple APIs, please indicate this in your response.