🎨 Stuck? Tired? Out of ideas?
Meet Gimai — your creative co-pilot powered by AI.
Instead of waiting for you to craft the perfect prompt, Gimai takes the lead
🧠 Built on the concept of reverse interaction, Gimai flips the script:
You don’t prompt the AI — the AI prompts you.
Zero inspiration? Doesn’t matter.
No creative energy? We’ve got you.
⚡️ Let Gimai extract the magic from your brain and turn it into visuals — fast.
🎯 MVP now live on Yandex.Cloud:
👉 https://tinyurl.com/gimai-demo
🖼️ Gimai is an AI-powered image generation service built around the idea of a prompt framework with reversed human-computer interaction.
Unlike traditional tools where a user provides a prompt and the machine responds, Gimai reverses the flow: the system first builds a UI and asks the user contextual questions to understand intent before generating an image.
Powered by the advanced FLUX.1 neural network, the service runs as a lightweight containerized API that can be deployed in seconds.
🧠 How It Works
- You provide a topic or creative intent (e.g., “cyberpunk portrait”).
- Gimai builds an adaptive questionnaire — asking about style, mood, composition, palette, etc.
- You respond interactively, helping the system build a meaningful and refined prompt.
- The final prompt is constructed based on your responses.
- The prompt is sent to the FLUX.1 model using the TogetherAI API (https://together.ai).
- An image is generated and returned.
Gimai is not just a generator — it's a semantic interface that helps you express ideas, even before you fully define them.
🎯 MVP
The MVP is currently deployed on Yandex Cloud and can be accessed at: https://tinyurl.com/gimai-demo
🚀 Quick Start
You can run Gimai locally using Docker in just a few steps.
docker run -d \
--name image-generator \
-p 5000:5000 \
-e API_KEY=YOUR_SECRET_API_KEY \
climbmunchkin/gimai:latest
Replace YOUR_SECRET_API_KEY with your actual Together AI API key.
Once started, the service will be available at:
👉 http://localhost:5000
Top comments (0)