Skip to main content

Ontology

The ontology is your organization's knowledge graph: a continuously updated map of every entity in your infrastructure and the relationships connecting them.

What the ontology contains

SixDegree discovers entities from connected integrations and stores them as typed nodes in the graph. Relationships between those nodes are inferred automatically from the data each integration provides.

ElementExamples
EntitiesAccounts, opportunities, subscriptions, tickets, incidents, vendors, contracts, people, services
RelationshipsOWNS, RENEWS, ESCALATED_TO, BELONGS_TO, DEPENDS_ON
AttributesLast updated, owner, status, tags, and integration-specific metadata

As your business changes, the graph updates automatically. New opportunities, ticket escalations, and renewals reflect immediately. No manual documentation step required.

info

LLMs only know what they were trained on. To reason about your specific infrastructure without hallucinating, an LLM needs accurate, real-time context about your systems. The ontology provides exactly that.

The object-oriented model

Entities in the ontology work like objects in object-oriented programming. Tools are the methods that operate on them: retrieving ephemeral data or taking actions.

SalesforceAccount.getOpenOpportunities() // real-time data, not stored in graph
ZendeskTicket.escalate(reason) // action executed via molecule
PagerDutyService.getIncidents() // live fetch from PagerDuty API
GitHubRepository.createIssue(title) // write action

Only long-living resources are stored in the graph. Ephemeral data (open opportunities, active incidents, recent tickets) is fetched on demand through tools at query time. This keeps the graph lean while giving the AI a complete, grounded picture when it reasons.

tip

This separation matters for performance. The graph answers "what exists and how does it connect" instantly. Tools answer "what is happening right now" by calling the upstream API.

Why the ontology matters

Traditional documentation goes stale immediately. Wikis fall behind. Tribal knowledge lives in people's heads and leaves with them. The ontology solves this by treating discovery as infrastructure, not documentation.

RoleBenefit
Revenue opsSee every touchpoint behind an account before a renewal: open opportunities, support escalations, and product usage in one view.
SupportTrace an SLA breach to the incident behind it and the revenue it puts at risk, without hopping between tools.
OperationsKnow which vendors and contracts back each team, and where spend no longer maps to usage.
EngineeringSee what depends on the code you're changing before you ship, and who owns it.

Exploring the graph

The ontology appears as an interactive graph in your dashboard. Nodes represent entities; edges show relationships. Node color indicates entity type.

info
info

Screenshot needed: Full-page graph view: colored nodes for different entity types (Salesforce accounts blue, Zendesk tickets green, PagerDuty incidents orange), edges labeled with relationships, search and filter chips at the top.

:::

Click any node to open its detail panel: full attributes, grouped relationships, and a shortcut to open it in the AI chat.

info
info

Screenshot needed: Entity detail side-panel that slides in when you click a node: entity name, type, attributes, relationships grouped by type, and a "Show in chat" button.

:::

Patterns the graph reveals

  • Accounts with an open escalation and a renewal due soon
  • Hidden dependencies between revenue, support, and the systems behind them
  • Vendors and contracts no team used this quarter
  • Ownership gaps where no one is accountable for an account or service

Common uses

Account context

Ask the AI "show every touchpoint for this account: tickets, calls, and product usage" to get the full picture before a renewal call, without stitching together five tabs.

Escalation impact

Trace an SLA breach to the incident behind it and the revenue it puts at risk. The graph answers blast-radius questions in seconds, across support and revenue at once.

Spend review

Ask "which vendors are we paying for that no team used this quarter" to surface contracts and spend that no longer map to usage, in one traversal.

Finding owners

Every entity shows current ownership and accountable contacts, whether that's the account owner, the on-call responder, or the team behind a service.

Building a complete picture

The ontology grows as you connect more integrations. Each molecule adds a new layer:

Salesforce → accounts, opportunities, contacts
Zendesk → tickets, organizations, agents
Stripe → subscriptions, invoices, customers
NetSuite → vendors, contracts, spend
PagerDuty → services, on-call schedules, escalation policies
GitHub → repositories, pull requests, users, teams
warning

The ontology reflects only what your connected molecules can discover. An entity that exists only in a disconnected system will not appear in the graph and will not be visible to the AI.

Next steps