Skip to content
    New featureAI10 September 2026

    Unhealthy AI model alerts — an e-mail when AI stops working

    Monokaido now keeps watch over your company’s AI and e-mails you when it stops working. Every night it tests your own AI model (OpenAI, Anthropic or Gemini) with your key, using the kinds of requests your AI features sent in the last 7 days; when the model works again, you get an e-mail saying so. On top of that, every 15 minutes it checks your company’s AI requests and, without waiting for the nightly test, alerts you when the provider refuses the key, the credit has run out, the model cannot handle a request or a fallback model is answering instead of the one you chose, and also when at least 20% of at least 5 requests to a model failed in the last hour. The “Unhealthy AI model” e-mail is on by default for every user of the company, and each user can switch it off in Settings → Notifications, in the AI section. The AI part of the daily summary e-mail now also shows how many AI requests failed, with the most frequent error, and what AI cost in euros.

    Let's try it on your own catalogue

    We'll walk through your feeds, categories and parameters live.