What is Arango?

The Arango Contextual Data Platform delivers a governed, continuously maintained context graph that AI agents and applications query for the entities, relationships, and sources behind an answer

What is the Arango Contextual Data Platform

The Arango Contextual Data Platform delivers a contextual data layer (also called a context layer): a governed, continuously maintained context graph that AI agents and applications query for the entities, relationships, and sources behind an answer.

The platform consists of three modules: the Agentic AI Suite, the Platform Suite, and ArangoDB. Every service in the platform reads and writes through ArangoDB, and agents and applications access the context layer through the ArangoDB Query Language (AQL), the HTTP API, or the Model Context Protocol (MCP).

Agentic AI Suite

The Agentic AI Suite builds the context graph, keeps it current, and answers questions from it.

AutoGraph Studio turns enterprise data into a context graph and retrieves from it:

Build: Ingests documents as-is (PDF, Office, text, and Markdown), clusters them into knowledge domains, and uses a large language model (LLM) to derive the entity types for each domain. It then extracts entities and relationships into the context graph. Each domain gets the processing depth it needs: full GraphRAG (graph-based retrieval-augmented generation) for relationship-rich content, or a lighter VectorRAG path where similarity search is enough.
Maintain: Updates are incremental. Inserting, deleting, or replacing a document reprocesses only what changed, and new documents join the nearest existing domain. Domains that have drifted since they were clustered are flagged; reclustering runs when you start it.
Retrieve: Selects a retrieval strategy for each question (vector search, graph traversal, or a combination) and answers against the context graph.
Trace: Documents, chunks, and entities carry provenance metadata, so answers can be traced back to their source files.

The suite also includes:

AQLizer translates natural-language questions into AQL, so users can query across domains in the context graph without writing AQL first.
Ada is an AI assistant for natural-language interaction and development.
Reasoner analyzes and optimizes AQL queries with AI-assisted reasoning.
GraphML applies machine learning to graphs for link prediction, classification, and embeddings.
Graph Analytics runs graph algorithms such as PageRank on dedicated compute.

Platform Suite

The Platform Suite runs, secures, and manages the platform on Kubernetes.

Kubernetes Operator: deployment, scaling, upgrades, backups, and license management.
Role-based access control (RBAC): controls which users and agents can read or act on which data.
Web interface: Graph Visualizer and Query Editor for exploring and querying the context graph.
Bring Your Own Code: run custom services alongside the platform.

ArangoDB

ArangoDB is the multi-model database at the foundation of the platform. Graph, document, key-value, full-text, and vector data share one storage engine, one index layer, and one transaction system. A document, its graph edges, its full-text index entries, and its vector embedding are stored and updated together, so there is no second store to synchronize and no cross-system join by ID.

A single AQL query can start from a full-text match, walk the graph, filter on document attributes, and rank by vector similarity:

FOR article IN articlesView
  SEARCH ANALYZER(article.body IN TOKENS(@topic, "text_en"), "text_en")
  FOR related, edge IN 1..2 OUTBOUND article cites
    FILTER related.published >= @since
    SORT APPROX_NEAR_COSINE(related.embedding, @queryVector) DESC
    LIMIT 10
    RETURN { title: related.title, via: edge._from }

ArangoDB is also available on its own for applications that store and query connected data.

What is a contextual data layer?

To answer a question reliably, an AI agent needs more than retrieved text passages. It needs to know which entities are involved, how they connect, what they mean, what is true now, where each fact came from, and whether the caller is allowed to see it. In most architectures, that context is split across a vector store, a graph database, a document store, and the pipelines that sync them.

A contextual data layer holds this context in one place as a context graph. Each layer below adds an answer the one before it cannot give:

Layer	What it adds	What it answers
Graph	Entities and relationships that can be traversed at scale	How is it connected?
Ontology	Shared entity types, relationships, and constraints	What does it mean?
Knowledge graph	Meaning applied to real entities across source systems	What is this, and how is it connected?
Context graph	Current state, sources, and access policies	What is true now, and can I trace why?

What the contextual data layer delivers

Capability	How it works	Result
Relational depth	Multi-hop graph traversal in the same query as filters, text search, and vector ranking	Answers that span systems and relationships, not isolated passages
Traceability	Provenance metadata links graph content to source files; graph paths show how an answer was reached	Explainable answers that can be verified
Freshness	Incremental updates; only changed documents are reprocessed	Context stays current without full rebuilds
Governed access	RBAC applied at the data layer for users and agents	Agents retrieve only what they are permitted to see
Fewer moving parts	One engine for graph, document, key-value, full-text, and vector data	No separate stores to sync, secure, and operate
Model choice	Access through AQL, the HTTP API, and MCP	Use any LLM, agent framework, or application

Ways to run it

Arango Contextual Data Platform

Self-managed: Runs on your own Kubernetes cluster, in the cloud, on-premises, or in air-gapped environments. Available in two editions:

Developer Edition: free under a non-commercial license.
Enterprise Edition: licensed for production and commercial use.

Both editions require ArangoDB Enterprise Edition 3.12.12 or later. Licenses activate and renew automatically through the Kubernetes Operator. See License Management.

Managed: The Arango Managed Platform (AMP) runs the Contextual Data Platform for you on Google Cloud or AWS, with monitoring, upgrades, and backups handled.

ArangoDB

Self-managed: ArangoDB Community Edition is free to use under the ArangoDB Community License. Enterprise Edition adds clustering features, including SmartGraphs for sharding connected data without cross-node traversal, and security features. See Features for the per-edition breakdown.

Managed: AMP runs ArangoDB for you on Google Cloud or AWS, with monitoring, upgrades, and backups handled.

Which product do I need?

Two offers are available to try for free:

ArangoDB with the Platform Suite: the multi-model database plus the services to run, secure, and explore it on Kubernetes.
Arango Contextual Data Platform: everything above, plus the Agentic AI Suite, which builds and queries a context graph for AI agents.

The deciding question is simple. Will you model and query your data yourself, or do you want the platform to build AI-ready context for you?

If you are building…	Choose
An application that stores and traverses connected data, such as recommendations, fraud detection, network and IT asset graphs, or identity and access	ArangoDB with the Platform Suite
An application that combines graph, document, key-value, full-text, and vector data in a single query	ArangoDB with the Platform Suite
Your own retrieval or RAG (retrieval-augmented generation) pipeline, where you design the schema and write the AQL queries	ArangoDB with the Platform Suite
AI agents, copilots, or assistants that need trusted context from enterprise documents	Arango Contextual Data Platform
GraphRAG without hand-building a schema, ontology, or extraction pipeline	Arango Contextual Data Platform
Answers to natural-language questions across domains, traced back to their source	Arango Contextual Data Platform
Shared memory and context for multiple agents, kept current as data changes	Arango Contextual Data Platform

Not sure? If your project involves AI agents or LLMs answering questions over documents, start with the Arango Contextual Data Platform. It includes ArangoDB and the Platform Suite, so you won’t outgrow it.

Next steps

Pick an installation path: