Your incidents
deserve a better
An AI SRE that autonomously investigates production incidents, remembers every past investigation, and maps your entire service topology.
See OpenSRE in action
Watch a full incident investigation in 60 seconds
Kubernetes
Prometheus
Grafana
Datadog
Elastic
Splunk
Jaeger
New Relic
Sentry
PagerDuty
Slack
GitHub
ConfluenceKubernetesPrometheusGrafanaDatadogElasticSplunkJaegerNew RelicSentryPagerDutySlackGitHubConfluenceSRE teams are stuck in a loop of manual, repetitive investigation.
Alert fatigue
Drowning in alerts at 3 AM, manually checking dashboards one by one. Your team spends 45 minutes triaging before investigation even begins.
Context switching
Jumping between Grafana, kubectl, Splunk, PagerDuty. Every tool is a tab, every tab is a context switch, every switch is time lost.
Tribal knowledge
When your senior SRE leaves, their investigation playbooks walk out the door. Each incident feels like starting from scratch.
Meet OpenSRE
OpenSRE investigates incidents, surfaces root causes, and learns from every investigation.
From alert to root cause. In minutes, not hours.
Alert fires
PagerDuty, Slack, or any webhook triggers the investigation. No human needed to start.
Agent activates
The planner formulates hypotheses and dispatches specialized sub-agents in parallel.
Deep investigation
Kubernetes, metrics, logs, and traces are queried simultaneously. Memory recalls similar past incidents.
Root cause report
Findings are correlated, root cause identified, and a structured report is delivered.
Built for real incidents.
Not a chatbot. A production-grade investigation engine with memory, knowledge, and autonomous reasoning.
Episodic memory
Every investigation is stored with multi-factor similarity matching. The agent recalls context from past incidents before investigating new ones — building institutional knowledge that never leaves.
Knowledge graph
Neo4j-powered service topology maps your entire infrastructure. Dependency traversal and blast radius analysis give the agent a systemic understanding — not just symptom checking.
46 investigation skills
From Kubernetes debugging to Prometheus queries to runbook execution. Skills load progressively — the agent only loads what it needs, keeping context focused.
Multi-agent architecture
A planner agent coordinates specialized sub-agents that investigate in parallel. Like a senior SRE delegating to a team, but faster.
100%opensource.
Novendorlock-in.
Self-host on your infrastructure. Extend with custom skills. Your data never leaves your network. Built by SREs, for SREs.