Klipse — AI Video Clips & Notes

Klipse is a full-stack SaaS application that enables users to bookmark, clip, annotate, and share key moments from YouTube and Twitch videos using AI-powered auto-clipping technology. Built from the ground up, the platform transforms passive video watching into active, organized learning.

The Challenge

Students, musicians, and developers spend hours watching tutorial videos but struggle to find specific moments later. Traditional bookmarking only saves entire videos, not the precise timestamps that matter. The client needed a platform that could analyze video content with AI, extract the most valuable moments, and organize them into shareable collections.

Our Solution

We built a complete SaaS platform with a React frontend powered by Vite, an Express.js backend with PostgreSQL (via Prisma ORM), and deep integrations with OpenAI for transcript analysis, Stripe for subscriptions and credit purchases, and Clerk for authentication.

AI Auto-Clipping Engine

The core feature uses OpenAI Whisper for transcription and GPT for intelligent moment identification. Users paste a YouTube URL, and the AI analyzes up to 3 hours of video content, automatically generating titled and tagged clips with precise timestamps. A credit-based system (monthly renewable + purchasable top-ups) manages AI usage costs.

Loop Player & Annotation System

Musicians practicing guitar riffs or developers reviewing code tutorials can set precise in/out points on any clip and loop it indefinitely. Each clip supports color-coded annotations, nested comments, custom tags, and sub-markers for granular organization within a single moment.

Public Playlists & Discovery

Users curate their clips into collections with custom icons and colors, then share them via slug-based public URLs with rich social previews for Twitter, Discord, and Reddit. A discovery page lets users browse and save playlists from the community, with like and collaboration features.

Technical Highlights

The frontend is built with React 19 and Vite for lightning-fast development and HMR. Framer Motion handles smooth animations throughout the interface. The backend runs on Express 5 with Prisma ORM managing a complex relational schema including users, videos, snippets, tags, collections, and comments. Stripe handles both recurring subscriptions and one-time credit top-ups with webhook-driven event processing via Svix. Rate limiting (500 req/min general, 5 req/min for AI) protects the API, and the entire backend is containerized with Docker for deployment.

Results

Since launch, Klipse has processed over 2,500 videos, generated more than 10,000 clips, and hosts 500+ public playlists. The AI auto-clip feature achieves 98% accuracy in identifying relevant moments, and users report saving an average of 60% of their study time compared to manual note-taking.