Oxeoxe

YouTube-trends pop-culture publication

01The Challenge

Pop-culture moves at the speed of YouTube's trending tab, but turning a stream of raw video stats into readable, publishable editorial every day is more than a small team can write by hand.

02The Solution

Oxeoxe ingests YouTube trending data into Supabase, then a Sanity Studio workflow uses Google Gemini to generate category summaries and full articles, stored as Portable Text in Sanity. The Next.js frontend renders the content with Algolia search, a live YouTube widget cached in Upstash Redis, social embeds, and Disqus comments.

03Features

Pulls the daily top-10 YouTube videos per category (views, likes, rank changes) from Supabase by region and date.

04System Architecture

Source

YouTube API

Data

Supabase

Gemini

CMS

Sanity

Algolia

Frontend

Next.js

Source

YouTube API

Data

Supabase

Gemini

CMS

Sanity

Algolia

Frontend

Next.js

05Technology Stack

Next.js 15 · React 19 · Tailwind · Portable Text · Zustand

06Key Learnings

01Generating editorial as Portable Text (not raw HTML) kept AI output clean, structured, and safe to render in Sanity.
02Heavy prompt engineering, banned-phrase lists and a target reading level, was what made Gemini's copy feel human.
03Caching the live YouTube widget in Upstash Redis avoided hammering the Data API while keeping the homepage current.
04Indexing posts into Algolia made the growing archive searchable without a custom backend.

Visit oxeoxe.com ↗

The editor: pick a trending video and give it an angle

Home page with the lead story playing

Oxeoxe

YouTube-trends pop-culture publication

01The Challenge

Pop-culture moves at the speed of YouTube's trending tab, but turning a stream of raw video stats into readable, publishable editorial every day is more than a small team can write by hand.

02The Solution

03Features

Pulls the daily top-10 YouTube videos per category (views, likes, rank changes) from Supabase by region and date.

04System Architecture

Source

YouTube API

Data

Supabase

Gemini

CMS

Sanity

Algolia

Frontend

Next.js

Source

YouTube API

Data

Supabase

Gemini

CMS

Sanity

Algolia

Frontend

Next.js

05Technology Stack

Next.js 15 · React 19 · Tailwind · Portable Text · Zustand

06Key Learnings

01Generating editorial as Portable Text (not raw HTML) kept AI output clean, structured, and safe to render in Sanity.
02Heavy prompt engineering, banned-phrase lists and a target reading level, was what made Gemini's copy feel human.
03Caching the live YouTube widget in Upstash Redis avoided hammering the Data API while keeping the homepage current.
04Indexing posts into Algolia made the growing archive searchable without a custom backend.

Visit oxeoxe.com ↗

Home page with the lead story playing