Collect the facts
Pull read-only context from Docker, Coolify, Airflow, host metrics, backup jobs, and existing runbooks.
AI operations copilot for small teams
JM Labs is building an evidence-first operations copilot for teams running self-hosted infrastructure. It connects system state, logs, and runbooks, then uses Claude to produce grounded diagnoses and safe remediation plans.
Storage pressure likely interrupted the nightly backup and triggered the application restart loop.
No changes are run automatically.
The product
Small teams already have monitoring. What they lack is the time to correlate every signal, check every runbook, and decide on the safest next action.
Pull read-only context from Docker, Coolify, Airflow, host metrics, backup jobs, and existing runbooks.
Use Claude to connect structured state with unstructured logs and operational knowledge—not just summarize alerts.
Cite the evidence, expose uncertainty, and propose reversible steps before any operational change.
How it works
Build status
The first prototype is intentionally narrow: one operator, one stack, and one high-quality diagnostic loop. No invented traction. No autonomous production changes.
Coolify and Docker health, logs, storage, and backup freshness.
Evidence citations, confidence, and validated remediation checklists.
Private deployments, shared runbooks, and human-approved actions.
Origin
JM Labs is an independent, pre-incorporation project focused on practical operations tooling for small technical teams.
It starts from direct experience operating self-hosted infrastructure and data systems without a dedicated SRE team.
Currently building
Product validation is being prepared.