For developers and AI agents Private preview
A graph database your AI agent can use in a minute.
- Create a database
- Connect Claude to its MCP server
- Ask a question
What you get
Managed Ladybug
The graph database formerly known as Kùzu, hosted for you, with daily backups and export at any time. Snapshots keep a copy apart from the database, to start a new one from.
Already on another graph database?
Connect your own Neo4j or Aura database and use it like a hosted one: the same endpoints, MCP server and chat. Connect your own Amazon Neptune cluster the same way. Or let MicroGraphs run a Neo4j Aura instance for you, on one bill. Or an Amazon Neptune cluster, with read replicas that scale with your reads.
An ontology, questions included
Design the ontology once: the node labels, relationship types and properties, and the business questions the graph answers. Version it, publish it, apply it to a database; then code and agents call a question by name.
Tune, monitor and scale
One database tested past 30,000 read queries a second. Watch each database's load, and let autoscaling add memory and instances as it grows and take them away as it eases. A load test runs your own queries at your own traffic on a copy, and says which configuration holds them. A slow query log keeps every query over your threshold, with the load of that moment.
Your graph on your website
Publish the graph behind a query or a business question on your own site. Style it in a designer, refresh it on a schedule, or answer each visitor live, in 2D or 3D.
See the whole graph
Every database draws a query's result as a graph, and lets you walk it node by node. A graph of millions of nodes is laid out once, near your database, then panned, zoomed and searched as one picture. Turn it around in 3D, or put nodes on a globe at their latitude, longitude and altitude. Plot nodes on their own values along axes. Lay them over the streets of OpenStreetMap.
See the pictures →