Contextus: The AI that finds what you didn't know to look for • Arango
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
Contextus is a graph-guided AI discovery agent built on ArangoDB that surfaces latent connections in graph data that traditional search, RAG, or dashboards cannot find. It works in four stages: ingesting a graph and training a GNN on its structure, running BFS traversal to map structural neighborhoods, using cosine similarity between GNN embeddings to predict links between structurally similar but topologically distant nodes, and finally using a LangGraph ReAct agent to explain discoveries in domain-specific language. Unlike RAG systems that retrieve text or search that finds named entities, Contextus answers 'what didn't I know to ask?' by finding indirect pathways and latent relationships. It runs natively on ArangoDB's multi-model platform, storing graph, document, and vector data together without a separate vector database or ETL pipeline. Use cases span biomedical knowledge graphs, financial networks, HR org graphs, software repositories, and supply chains. A demo on a 27,583-node RISC-V chip design graph found architectural similarities between memory management units across separate processor repositories.
