Blog

Real-world stories from running AI agents in production.

How We Built a 24/7 Claude Code Telegram Bot on a Free VM
Anthropic's official Telegram integration doesn't work on headless Linux — an unfixed bug since launch. We built a workaround using ccgram + hooks on a free Oracle ARM VM. Here's the full architecture.
How we automated code review with 5 AI agents
We replaced our manual PR review process with a swarm of specialized agents — a security scanner, a logic checker, a style enforcer, a test coverage analyzer, and an orchestrator. Here's what we learned.