All Projects
SAAS

İçerik Madeni — Multi-Project AI Content Production & Distribution Platform

A fully autonomous AI content factory serving multiple projects from a single hub. 5 providers, 100+ categories, 8+ languages, pull-based REST API.

vevona·İçerikmadeni.app
CLIENT Vevona · İçerik Madeni
CATEGORY SaaS
YEAR 2025
PROJECT DETAILS

Projects about

İçerik Madeni ("Content Mine") is a fully autonomous AI content factory serving multiple projects from a single hub. A pull-based REST API lets external platforms fetch publish-ready content on demand with a Bearer token — production and consumption are completely decoupled.

20,000+Total content capacity
5Integrated AI providers
100+Active categories
8+Supported languages

Business problems solved

  • Manual content cost: Instead of hiring writers/editors per project, category-based auto-generation — the categories × target_count formula feeds the queue automatically.
  • Multi-AI provider chaos: Five providers (Claude, Gemini, GPT-4, Groq, OpenRouter) under one panel. Model/provider hot-swap with no code change.
  • Transient error recovery: Dynamic retry that parses wait time directly from the 429 response. Stuck/failed records are auto-requeued every 30 minutes.
  • Multi-tenant isolation: Each project has its own token, content schema, and generation rules — projects never see each other.

Modules

🧠 AI Engine Layer

5 provider integrations, per-provider API key + rate limit + delay. Hot-swap provider/model from the admin panel.

⚙️ Automation Layer

Laravel Scheduler with hourly/daily/weekly triggers. Category-based auto-generation, Excel bulk topic import.

🔌 Pull REST API

Bearer-token GET /content, GET /content/{id}, POST /content/{id}/pulled. Production ↔ consumption fully decoupled.

🔔 Webhook Notifications

External platforms get pinged the moment content is ready — push works alongside pull.

🛡️ Smart Retry & Recovery

Wait-time parsed from 429, 3 retries. Transient vs permanent error split. Stale-record auto-cleanup.

📋 Audit & Prompt Templates

Every item logs duration, tokens, provider, model. Per-project temperature, max tokens, system prompt.

Technology

Stack
PHP 8.2Laravel 12Filament 3LivewireMySQLREST APIArtisan Scheduler
AI Providers
Anthropic ClaudeGoogle GeminiOpenAI GPT-4GroqOpenRouter (300+ models)
Deploy
cPanel / LinuxGitHub → SSH (single command)

Outcome

Average 15–45 seconds per item. Doubling topics, projects, or languages requires no code change — the system scales itself. Adding a new project = new token + new template; everything else is automatic.

The next project yours could be yours.

Get in touch to discuss a similar project.

Get in Touch