How I Built a 24/7 AI Content Machine Using OpenClaw
For the past month, I've been running an experiment: can AI agents work 24/7 to generate, publish, and maintain content on my websites? The answer is yes — and the setup is simpler than you'd think.
The Architecture
OpenClaw runs as an agent on my VPS. It connects to Azure AI Foundry for content generation (DeepSeek) and image creation (FLUX). Python scripts handle everything — from topic selection to HTML generation to publishing.
Why Server-Side?
API keys stay on the server, never in the browser. A PHP proxy handles all Azure calls. Nginx basic auth secures the admin panel. This means I can manage everything from my phone without exposing credentials.
Automated Publishing
A systemd timer fires every 4 hours. The script picks a random Nepal-related topic, generates a full article with SEO meta tags, optionally creates a featured image, and publishes to both English and Nepali sites.
Systemd Timer → Python Script → Azure AI → HTML → Nginx
The Results
In 30 days, the bot published 12 articles across two sites. Traffic increased 3x. The key insight: AI-generated content works when it's focused on factual, news-style topics rather than opinion pieces.
Key Takeaways
- Server-side AI proxy keeps API keys secure
- Systemd timers are more reliable than cron for long-running tasks
- Focus AI on factual content — news, guides, comparisons
- Always monitor output quality with a review dashboard
- Images from FLUS provide unique visuals, not stock photos