Xennvon: Turn Long Videos into Shorts with AI

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass.

What I Built

Meet Xennvon, an AI-powered video repurposing platform designed to make short-form content creation easier for creators.

Creating short videos from long podcasts, videos, and other content can take hours of manual editing, clipping, captioning, and preparing metadata. Xennvon aims to simplify this workflow by bringing these tasks together in one platform.

With Xennvon, creators can automate parts of their content workflow, including:

• Turning long-form videos into short clips.
• Processing videos using FFmpeg and yt-dlp.
• Generating subtitles and captions.
• Creating AI-assisted titles, descriptions, and hashtags.
• Streamlining the process of preparing content for short-form platforms.

The goal is simple: reduce repetitive editing work so creators can spend less time on production and more time creating, connecting with their communities, and doing things beyond their screens.

Demo

Live application: https://xennova-ishu.vercel.app/

The frontend is deployed on Vercel. The backend is being deployed on Microsoft Azure.

Code

The source code repository is currently private to protect the original ideas and intellectual property behind Xennvon. You can explore my other projects and contributions on
GitHub: https://github.com/itsishant.

The project is built around a full-stack architecture, combining a web interface, backend processing, AI-assisted workflows, and video-processing tools.

How I Built It

Xennvon combines full-stack development with AI-assisted content automation.

Core technologies and tools:

• Frontend: React, Vite, and TypeScript.
• Backend: Node.js and TypeScript.
• Video processing: FFmpeg and yt-dlp.
• Transcription: faster-whisper.
• AI inference: Ollama, which allows compatible open-weight language models to run locally or on a server.
• Data and infrastructure: MongoDB, Redis, and Microsoft Azure for backend deployment.
• Frontend hosting: Vercel.

The workflow combines video processing with AI-generated metadata to reduce the manual effort required to prepare short-form content.

Ollama is particularly interesting because it makes local model inference possible without requiring every AI task to depend on a proprietary hosted language-model API. The actual model and configuration determine the capabilities and resource requirements.

Why Does Open Innovation Matter?

Open innovation gives developers more control over how AI-powered applications are built and deployed.

For Xennvon, Ollama provides a way to experiment with compatible open-weight models, switch models, and run inference on infrastructure I control. This creates flexibility in how AI features are developed and deployed.

Open models can also provide greater control over inference costs and where processing happens, depending on the model, hardware, and deployment setup.

I believe AI-powered tools should make development and creative work more accessible without forcing developers into a single provider or model ecosystem.

Xennvon is my attempt to explore that flexibility through a practical, full-stack application.

My Agent Session

Optional: I'll share my development session here if available.

Prize Categories

• General Hacktoberfest Open-Source AI Challenge.