Karpathy said we should build LLM knowledge bases, and 48 hours later someone made Graphify: one command, full semantic knowledge graph.
We applied that idea to incident management.
Most tools tell you what just happened. But during an incident, you actually want:
(1) What happened last time this broke?
(2) What fixed it?
(3) What's likely to break next?
Incident data, alerts and teams are fed into Graphify.
Now incidents aren’t logs, they’re nodes within a semantic graph:
What this brings:
(1) Instant incident memory
(2) services that fail together naturally cluster
(3) Alert noise vs signal becomes obvious
(4) Team load patterns become visible
(5) Query the graph for deeper insights, context and relations
If you’re using Rootly, here is a small plugin to explore your incident data. Check it out: github.com/Rootly-AI-Labs/rootly-graphify-importer
Karpathy said we should build LLM knowledge bases, and 48 hours later someone made Graphify: one command, full semantic knowledge graph.
We applied that idea to incident management.
Most tools tell you what just happened. But during an incident, you actually want: (1) What happened last time this broke? (2) What fixed it? (3) What's likely to break next?
Incident data, alerts and teams are fed into Graphify. Now incidents aren’t logs, they’re nodes within a semantic graph:
(1) nodes = services, incidents, alerts, responders (2) edges = how everything connects
What this brings: (1) Instant incident memory (2) services that fail together naturally cluster (3) Alert noise vs signal becomes obvious (4) Team load patterns become visible (5) Query the graph for deeper insights, context and relations
If you’re using Rootly, here is a small plugin to explore your incident data. Check it out: github.com/Rootly-AI-Labs/rootly-graphify-importer