We built On-Call Health to help teams detect signs of overload in on-call incident responders. Burnout is too common for SREs and other on-call engineers and that’s who we serve at Rootly. We hope to put a dent in this problem with this tool.
Here is the project GitHub repo https://github.com/Rootly-AI-Labs/On-Call-Health, and here is the hosted version https://oncallhealth.ai. The easiest way to try the tool is to log into the hosted version, which has mock data.
The tool uses two types of inputs: - Observed signals from tools like Rootly, PagerDuty, GitHub, Linear, and Jira (incident volume and severity, after-hours activity, task load…) - Self-reported check-ins, where responders periodically share how they're feeling
We provide a "risk level", a compound score derived from objective data. The self-reported check-in feature is taking inspiration from the Ecological Momentary Assessment (EMA), a research methodology also used by Apple Health's State of Mind feature.
We provide trend data for all metrics, for both teams and individuals, to help managers spot anomalies that may require investigation. Our tool doesn't provide a diagnosis, nor is it a medical tool; it simply highlights signals.
It can help spot two types of potential issues:
1. Existing high load: when setting up the tool, teams and individuals with a high risk level should be looked at. A high score doesn't always mean there's a problem – for example, some people thrive on high-severity incidents – but it can be a sign that something is already wrong. 2. Growing risk: over time, if risk levels are steeply climbing above a team or individual baseline.
Users can consume the findings via our dashboard, AI-generated summaries, our API, or our MCP server.
The project is fully open source and self-hostable, and the hosted version is available at no cost. We have a ton of ideas to improve the tool to make on-call suck les,s and we are happily accepting PR and welcome feedback on our GitHub repo. You can reach out directly to me at [email protected]