Statuspage costs $399/mo and doesn't monitor anything. We're building the $8/mo version on Cloudflare Workers — with LLM and MCP server checks nobody else does.
No spam. One email when it's ready to try, nothing else.
The incumbents are either expensive and blind, or cheap and ugly. Nobody's built the version for a developer who also runs an AI endpoint.
Plain uptime pings miss the failure modes that actually matter for an AI-era stack.
Status code, latency, keyword match. The table-stakes check, done right.
A tiny fixed prompt against your OpenAI-compatible endpoint — checks valid output, time-to-first-token, tokens/sec. Catches creeping latency plain checks never see.
Confirms the initialize handshake succeeds and the tool list actually returns — not just that the port is open.
The market's already moving. UptimeRobot recently hit legacy customers with roughly 425% price increases on their old plans — a real, current wave of people looking for an alternative.
Nobody owns AI-endpoint monitoring yet. OpenRouter and SigNoz solve this only for their own platforms. If you're running a self-hosted vLLM box or your own MCP server, there's no equivalent tool watching it.
Sources: microgaps.com uptime-monitoring market analysis; web-alert.io on LLM API monitoring practices. Figures are directional, not guaranteed — we're building this in public and will show real numbers as they exist.