Onyx Database
AI-Controlled Database Cloud

Structured data, made flexible and intelligent.

Onyx Cloud is an API-driven database management system and AI agentic control plane with built-in RAG (Retrieval Augmented Generation) capabilities for querying data across structured tables (using rest api and client sdk builder pattern constructed queries) and unstructured documents (inverted index searching). It automatically generates optimized queries for structured data and enables natural-language access across your entire data model. Onyx can be used to atomically save data, with extremely fast read and write performance, as well as resolve flexible relationships to other graphs.

How Onyx WorksView Architecture
workflow.ts
1. Save & Query
const widget = { 
  name: "M2-55", 
  category: "Electronics",
  description: "A powerful gadget..."
}

await db.save(tables.Product, widget);

const results = await db.from(tables.Product)
  .search("tech AND powerful")
  .and("category", "Electronics")
  .list();
2. Natural Language AI
const answer = await db.chat("what is a M2-55?");

What Is Onyx?

A modern database for connected data

Unlike traditional SQL, NoSQL, or standalone graph databases, Onyx combines a flexible graph-native core with an AI-driven control plane.

Built for Evolving Teams

Onyx is designed for systems where data is:

  • Highly connected
  • Constantly evolving
  • Analyzed continuously

Why Traditional Databases Break

The Old Way

  • • Fixed schemas and rigid relationships
  • • Manually written brittle queries
  • • Analysis performed outside the DB

The Failure Point

Approaches fail when schemas evolve or AI needs structured access. The result is brittle analytics and duplicated logic.

How Onyx Is Different

Onyx models data as entities in collections, but relationships are not stored as rigid edges.

  • Relationships defined at schema level
  • Resolver queries describe connections
  • Graphs attach at the model level
  • Resolution happens inside the DB

Backed by indexed reference assignment and optimized traversal structures to avoid N+1 query patterns.

AI Control Plane

Rather than writing one-off SQL, users work with workflow trees that express intent and generate queries.

A flash-model AI plans workflows, but every step is visible and reviewable.

First-Class Outputs

  • statistical summaries
  • derived datasets
  • narrative explanations
  • visualization-ready results

How Teams Use Onyx

  • Graph-shaped domain models
  • Analytics over evolving schemas
  • AI and RAG backends
  • Authorization and policy systems
  • Internal data platforms

Why It Matters

Tradeoffs are a thing of the past

SQL Structure
NoSQL Flexibility
Graph Relations
BI Analytics

Onyx separates data modeling, relationship definition, and data workflow orchestration, and optimizes each independently. This is what makes AI-directed, trustworthy data workflows possible inside a real database system.

Rethink how database work gets done

Learn how Onyx works, explore the architecture, or dive into the technical model behind resolver-defined relationships and AI-controlled workflows.

Get Started