Dev.to
7/29/2026

Building a stateless AI observability platform on $0/month infrastructure
Original: I built an AI observability platform with $0 – zero dependencies, zero ops, stateless
Short summary
A solo developer built Cognilumin, a stateless AI observability platform running on $0/month infrastructure using free-tier AI providers (Gemini, Groq, OpenRouter) with vanilla PHP and JavaScript. The platform includes an explanation tool, a topology visualizer, and an execution log tracer for AI agents. The author shares lessons on prompt engineering as infrastructure and the benefits of stateless architecture for small teams.
- •Cognilumin: stateless AI observability platform on $0/month with zero dependencies
- •Three tools: Illuminator, Topology Visualizer, and Execution Logs for AI debugging
- •Key lesson: prompt engineering across multiple LLM providers required more iteration than UI work
Generated with AI, which can make mistakes.
Is this a good recommendation for you?



