Uramaki is an AI-powered app and PWA that turns a single prompt into ready-to-publish social content. It generates storytelling and outputs it as posts, carousels, or reels for Instagram, Facebook, and TikTok.
I owned the entire product from concept to release: product strategy and UX/UI design, full-stack application development, database design and integration, and infrastructure setup and deployment.
While producing social content for walletf, I needed to speed up the workflow. Canva templates helped with layout but not with the full pipeline: I still had to write the copy, source images, and adapt everything for each format and platform.
Key challenges:
Uramaki was created to bring ideation, copy, and visuals into one guided flow and reduce the time and friction of social content production.
I defined the product concept and designed the interface and user flows for the app and PWA: from prompt input to generated storytelling, refinement controls, and export to post, carousel, and reel formats for each platform.
The UI keeps the focus on the creation flow—clear hierarchy, minimal steps, and immediate preview of the AI output so users can iterate quickly and adjust copy or image style without leaving the tool.
The application and PWA, from one prompt, generate the narrative and present it in publish-ready formats: single posts, carousels, and reels for Instagram, Facebook, and TikTok.
Key features include:
The result is a streamlined, output-focused experience: from a text prompt to social-ready content in one continuous flow.
I designed and implemented the backend, database, and infrastructure to support the AI pipeline, user data, and generated content: database schema and integration for projects and outputs, API integration with AI services for copy and image generation, and deployment infrastructure for the PWA and server-side components. The stack is built for reliability and scalability.
I developed the full application: frontend for the PWA and app experience, backend logic, database integration, and infrastructure setup and deployment. The product is live and usable end-to-end—from concept and design through to a deployed, working application.
The biggest challenge was training and refining the AI instructions until the output was not only visually polished, but also optimized for both vertical storytelling and horizontal carousel narratives. The system had to understand pacing, slide hierarchy, visual continuity, and how a story changes when it moves from a vertical flow to a swipeable sequence.
