Building and running an AI powered product
summary
Storic is a web app that builds personalized storybooks from eight user photos, fine-tunes image models per order and prints and ships physical books; the team chooses Stable Diffusion/Flux checkpoints, fine-tuned with LoRA, generates watercolor-style art via diffusion, and runs inference on demand using Replicate for deployment while managing queuing and cost. Image generation takes about 6–10 seconds per image, training is iterative, and beta tests show personalization works but input quality and latency remain the main pain points.
AI-written summary · about
chapters (12)
- 0:00🚀 Introduction to AI-Powered Product Development
- 3:00Challenges in AI Model Selection and Fine-Tuning
- 5:00The Evolution of Storic: From Concept to Product
- 8:00Key Technologies: Stable Diffusion, Flux, and Hugging Face
- 11:00Text-to-Image Generation: Creating Personalized Illustrations
- 15:00Fine-Tuning with Laura and Textual Inversion
- 18:00The Role of User Data and Feedback in Fine-Tuning
- 22:00Using Replicate for AI Model Deployment and Scaling
- 26:00Handling Latency and Improving User Experience
- 30:00AI Agent System: A Solution for Optimizing the User Journey
- 33:00Key Takeaways: Navigating the AI Development Landscape
- 35:00Final Thoughts on Building User-Centered AI Products
related talks
36mDesigning the Future: Stable Diffusion & the Rise of AI-Driven SaaS
01h 03mBuild & deploy AI-powered apps
34mBuilding Owly: An AI Comic Video Generator for My Son
25mHow Mixbook leveraged genAI to offer personalized experiences
33mAgentic AI – What's Next?
29mIntegrating Stable Diffusion Generative AI in Artist Workflows
29mTools and Roadmap for AI-Native Developers
44mBringing stories to life with AI, data streaming and generative agents