PixiqueAI

01 — Overview

Project overview

PixiqueAI needed a product that could take AI image processing — historically a technical, script-driven workflow — and turn it into something anyone could use directly from the browser, without sacrificing the performance and reliability a paying user expects from a SaaS product.

02 — Challenge

The challenge

AI image transformation is compute-heavy and asynchronous by nature. The core challenge was designing an interface and backend architecture that could handle uploads, queued AI processing, credit consumption and billing, while still feeling instant and predictable to the end user.

03 — Approach

The approach

The build started with the data model: how images, jobs, credits and users relate to each other, and how state moves from 'uploaded' to 'processing' to 'ready' without leaving the user guessing. From there, the interface was designed around that state machine rather than the other way around — every screen reflects exactly where a job stands.

04 — Solution

The solution

The result is a Next.js and TypeScript application with Supabase handling authentication, storage and the relational data layer, Replicate powering the underlying AI image models, and Stripe managing subscriptions and credit top-ups. Cloudflare sits in front of media delivery for fast, cached image loading worldwide, and the entire application is deployed on Vercel for fast, reliable global delivery.

05 — Features

Key features

• →Drag-and-drop image uploads with client-side validation and progress feedback
• →Asynchronous AI processing queue with real-time job status
• →Credit-based usage system tied to Stripe subscriptions and one-off top-ups
• →Secure authentication and per-user storage via Supabase
• →Batch processing for multiple images in a single job
• →Responsive, modern SaaS interface built for both desktop and mobile use
• →CDN-backed image delivery for fast load times globally

06 — Outcome

Outcome

PixiqueAI shipped as a working, production SaaS product with a complete end-to-end flow — from sign-up, to upload, to AI processing, to billing — running live in production today.

07 — Takeaways

Key takeaways

• Designing around asynchronous state early prevents a fragile interface later.
• A credits-based billing model needs to be enforced consistently on both the client and the server to stay trustworthy.
• Offloading heavy compute to a specialised AI infrastructure provider (Replicate) kept the core application lean and maintainable.