🏗️System Architecture

@divz.mysterious · Eco Resin · Crochet · Handmade Decor · Slow Crafting

Goal: 100K+ followers

Frontend

Next.js 14 (App Router) with React Server Components

Recharts for data visualization

TailwindCSS + custom design system

Framer Motion for animations

OAuth redirect handlers per platform

Real-time data subscriptions via Supabase

Backend (API Layer)

Next.js API Routes (serverless functions)

Vercel Cron Jobs for scheduled data fetching

Rate limiter middleware per platform

Token refresh service (background)

LLM orchestration layer (OpenAI / Claude API)

Webhook receivers for real-time platform updates

Data Layer

Supabase (PostgreSQL) with Row-Level Security

Real-time subscriptions for live dashboard updates

Edge Functions for lightweight compute

Supabase Storage for media cache / thumbnails

Automated daily snapshots of analytics data

Full-text search on trending topics

Data Flow Pipeline

OAuth Login→

Token Store→

Cron Fetch→

Raw → DB→

Analytics Engine→

AI/LLM Layer→

Dashboard UI

User authenticates with each platform via OAuth 2.0 flow

Encrypted tokens saved to Supabase with expiry tracking

Scheduled jobs pull analytics from each platform API

Normalized data stored as snapshots and post records

Compute signals, patterns, trends from historical data

Feed processed data to LLM for recommendations

Real-time rendered via Supabase subscriptions

Database Schema

users— Creator accounts

id (uuid PK), email, name, avatar_url, created_at, updated_at

platform_connections— OAuth tokens per platform

id, user_id (FK), platform (enum), access_token (encrypted), refresh_token (encrypted), scopes, token_expires_at, platform_user_id, connected_at

analytics_snapshots— Daily analytics snapshots

id, user_id (FK), platform, followers, reach, engagement_rate, saves, shares, watch_time_hrs, snapshot_date

content_posts— Individual post performance

id, user_id (FK), platform, platform_post_id, title, type (reel/short/pin/video), views, saves, shares, comments, completion_rate, posted_at, fetched_at

trending_topics— Trending topic signals

id, topic, platform, category, growth_rate, heat_score, search_volume, detected_at, expires_at

ai_recommendations— AI-generated content ideas

id, user_id (FK), concept, hook, video_length, shot_list (jsonb), editing_style, music, caption, hashtags (text[]), virality_score, generated_at

viral_patterns— Detected viral patterns

id, user_id (FK), insight, impact, category, data_points, detected_at

creator_watchlist— Tracked trending creators

id, handle, platform, followers, growth_rate, niche, format_description, last_checked

API Endpoints

Method	Endpoint	Description	Auth
GET	/api/auth/[platform]	Initiate OAuth redirect for given platform	Public
GET	/api/auth/callback	Handle OAuth callback, exchange code for tokens	Public
POST	/api/auth/refresh	Refresh expired platform tokens	Cron
GET	/api/analytics/overview	Cross-platform summary metrics	Auth
GET	/api/analytics/growth	Historical follower growth data	Auth
GET	/api/analytics/posts	Post-level performance with sorting/filters	Auth
GET	/api/analytics/signals	Algorithm signal scores and benchmarks	Auth
GET	/api/trends/topics	Current trending topics with heat scores	Auth
GET	/api/trends/keywords	Rising keywords with search volume	Auth
POST	/api/ai/recommend	Generate AI content recommendations	Auth
POST	/api/ai/patterns	Run viral pattern detection analysis	Auth
GET	/api/creators/trending	List trending creators in niche	Auth
GET	/api/opportunities	Content gap analysis results	Auth
POST	/api/cron/fetch-analytics	Scheduled analytics data fetch	Cron
POST	/api/cron/detect-trends	Scheduled trend detection pipeline	Cron

AI Analysis Pipeline

Trend Detection Engine

Aggregate search data from Pinterest Trends, YouTube Search Suggest, IG Explore

Calculate velocity scores: growth rate over 7/14/30 day windows

Cross-reference with creator's niche keywords for relevance scoring

Rank by heat score (velocity × relevance × recency)

Content Recommendation Engine

Input: top trends + creator analytics + algorithm signals + top posts

LLM prompt: generate content concepts optimized for creator's strengths

Score each concept: virality probability based on historical pattern match

Output: ranked list with full production briefs (hook, shots, caption, tags)

Viral Pattern Detector

Analyze top 50 posts: extract duration, hook timing, camera angle, audio type

Statistical correlation: which attributes correlate with high save/share rates

Generate natural language insights with confidence intervals

Refresh weekly as new content is published

Opportunity Gap Analyzer

Demand signal: keyword search volume + engagement on related content

Supply signal: count of active creators + content frequency in niche

Gap score: demand / supply ratio normalized to 0-100

Suggest specific content ideas to fill each gap