Ultrahost/ docs
All primitives

Data / relationships and context

Graph Database

Discover connections hidden in your data.

The idea

Model the relationships, not just the records.

Imagine an application for organizing professional contacts. You know Alice, Alice knows Bob, and Bob works at a company you'd like to approach.

A graph database makes relationships like these easy to represent and explore.

Instead of focusing only on individual records, it focuses on how things are connected.

Application blip

My Network

5 sample entities · 5 relationships
People
Companies
Projects
RelationshipsKnowsWorks atParticipates in
React Flow

Select a person to highlight their direct relationships, or trace a connection to Acme Corp.

Read relationships as a list
  • You — Knows — Alice
  • Alice — Knows — Bob
  • Bob — Works at — Acme Corp
  • Alice — Participates in — Atlas Project
  • Bob — Participates in — Atlas Project

Behind the scenes

Turn connections into questions you can ask.

Nodes
Things in your domain: people, companies, and projects.
Edges
Relationships between things: knows, works at, participates in.
Graph queries
Questions that explore how those relationships connect.

A graph query can follow several edges to find a path—for example, moving from a person through shared contacts to a company.

What could you build?

  • Social networks
  • Fraud detection
  • Recommendation engines
  • Knowledge graphs
  • Supply-chain analysis

Implementation notes

  • Model people, companies, and projects as nodes.
  • Define edges such as Knows, Works at, and Participates in.
  • Use graph queries to discover multi-step paths.

Implementing it

With a coding agent

Ask your agent to map your connections

Agent setup (mock link)

Copy this prompt, fill in the bracketed details, and paste it into your coding agent. It describes the goal, the relationships, and the guardrails.

Bracketed text marks details only you can fill in. 861 / 6,000 characters.

Fill in:YOUR DATAAPP OR PRODUCTTHINGS THAT BECOME NODESRELATIONSHIPSNODE AND EDGE PROPERTIESCONNECTION QUESTIONS TO ANSWERMAXIMUM DEPTHUSER, WORKSPACE, OR TENANTIMPORTANT PATHS, DEAD ENDS, AND ACCESS CASES

Good foundation · 65/100

Traditional SDK / API

Trace a path through your network

TypeScript
const related = await ultrahost.graph.traverse({
  from: { type: "user", id: user.id },
  through: ["knows", "works_at"],
  depth: 3,
})

Key takeaway

A Graph Database helps your application understand and explore relationships, not just individual pieces of information.

Next primitive

Custom Functions