Get knowledge graph

curl --request GET \
  --url https://app.nex.ai/api/developers/v1/graph \
  --header 'Authorization: <api-key>'

{
  "nodes": [
    {
      "id": "12345",
      "name": "John Smith",
      "type": "person",
      "definition_slug": "person",
      "primary_attribute": "<string>",
      "created_at": "2023-11-07T05:31:56Z"
    }
  ],
  "edges": [
    {
      "id": "<string>",
      "source": "<string>",
      "target": "<string>",
      "label": "works_at",
      "definition_id": "<string>"
    }
  ],
  "relationship_definitions": [
    {
      "id": "<string>",
      "name": "<string>",
      "entity_1_to_2": "<string>",
      "entity_2_to_1": "<string>"
    }
  ],
  "context_edges": [
    {
      "id": "tri_42",
      "source": "12345",
      "target": "67890",
      "label": "works_with",
      "edge_type": "context",
      "confidence": 0.5
    }
  ],
  "insight_nodes": [
    {
      "id": "ei:42",
      "content": "John prefers email communication over phone calls",
      "type": "preference",
      "confidence": 0.5,
      "source": "entity_insight"
    }
  ],
  "insights": {},
  "total_nodes": 123,
  "total_edges": 123,
  "total_context_edges": 123
}

Graph

Get knowledge graph

Returns the full knowledge graph for the workspace, including CRM entity nodes, relationship edges, context edges (from ingested text), insight triplet edges (AI-extracted subject-predicate-object relationships), insight nodes, and insight link edges (semantic relationships between insights). Supports an optional limit query parameter to control the maximum number of entity nodes returned (default 1000, max 5000). Set Accept: text/html or ?format=html to receive an interactive visualization.

GET

graph

Get knowledge graph

curl --request GET \
  --url https://app.nex.ai/api/developers/v1/graph \
  --header 'Authorization: <api-key>'

{
  "nodes": [
    {
      "id": "12345",
      "name": "John Smith",
      "type": "person",
      "definition_slug": "person",
      "primary_attribute": "<string>",
      "created_at": "2023-11-07T05:31:56Z"
    }
  ],
  "edges": [
    {
      "id": "<string>",
      "source": "<string>",
      "target": "<string>",
      "label": "works_at",
      "definition_id": "<string>"
    }
  ],
  "relationship_definitions": [
    {
      "id": "<string>",
      "name": "<string>",
      "entity_1_to_2": "<string>",
      "entity_2_to_1": "<string>"
    }
  ],
  "context_edges": [
    {
      "id": "tri_42",
      "source": "12345",
      "target": "67890",
      "label": "works_with",
      "edge_type": "context",
      "confidence": 0.5
    }
  ],
  "insight_nodes": [
    {
      "id": "ei:42",
      "content": "John prefers email communication over phone calls",
      "type": "preference",
      "confidence": 0.5,
      "source": "entity_insight"
    }
  ],
  "insights": {},
  "total_nodes": 123,
  "total_edges": 123,
  "total_context_edges": 123
}

This endpoint requires the record.read scope on your API key.

Overview

Returns the full knowledge graph for your workspace. The graph combines CRM relationship data with AI-extracted knowledge from ingested context, providing a unified view of entities and their connections.

Node Types

Type	ID Format	Response Field	Description
CRM entity	`"12345"`	`nodes`	Resolved entities (people, companies, custom objects)
Ghost	`"ghost:123"`	`nodes`	Entities mentioned in context but not yet matched to a CRM record
Entity insight	`"ei:123"`	`insight_nodes`	AI-extracted insight scoped to a specific entity
Knowledge insight	`"ki:123"`	`insight_nodes`	Workspace-level AI-extracted knowledge

Edge Types

The context_edges array contains edges from multiple sources, distinguished by the edge_type field:

Edge Type	Description	Has Confidence
`context`	Relationships from ingested text (e.g., “involved_in”)	No
`triplet`	AI-extracted subject-predicate-object relationships (e.g., “works_with”, “reports_to”)	Yes
`insight`	Links connecting insight nodes to entity nodes	No
`insight_link`	Semantic relationships between insight nodes (`related_to` or `contradicts`)	Yes

Insight Link Edges

insight_link edges represent semantic relationships discovered between insights. They connect two insight nodes (entity insights ei:* or knowledge insights ki:*) and always include a confidence score (0-1) representing semantic similarity. Labels:

related_to — The two insights are semantically related
contradicts — The two insights contain contradictory information

{
  "id": "il_42",
  "source": "ei:100",
  "target": "ei:200",
  "label": "related_to",
  "edge_type": "insight_link",
  "confidence": 0.87
}

Edge Priority

When both a context edge and a triplet edge exist between the same pair of nodes, generic context labels (involved_in, context) are suppressed in favor of the richer triplet edge. Non-generic context labels are kept alongside triplet edges.

Interactive Visualization

Request HTML format to receive an interactive graph visualization:

# Via query parameter
curl -s "https://app.nex.ai/api/developers/v1/graph?format=html" \
  -H "Authorization: Bearer YOUR_API_KEY"

# Via Accept header
curl -s "https://app.nex.ai/api/developers/v1/graph" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Accept: text/html"

Example

curl -s "https://app.nex.ai/api/developers/v1/graph?limit=100" \
  -H "Authorization: Bearer YOUR_API_KEY"

{
  "nodes": [
    {
      "id": "12345",
      "name": "John Smith",
      "type": "person",
      "definition_slug": "person",
      "primary_attribute": "John Smith",
      "created_at": "2026-01-15T10:00:00Z"
    },
    {
      "id": "67890",
      "name": "Acme Corp",
      "type": "company",
      "definition_slug": "company",
      "primary_attribute": "Acme Corp",
      "created_at": "2026-01-10T08:00:00Z"
    }
  ],
  "edges": [
    {
      "id": "1001",
      "source": "12345",
      "target": "67890",
      "label": "works_at",
      "definition_id": "5"
    }
  ],
  "relationship_definitions": [
    {
      "id": "5",
      "name": "works_at",
      "entity_1_to_2": "works_at",
      "entity_2_to_1": "employs"
    }
  ],
  "context_edges": [
    {
      "id": "tri_42",
      "source": "12345",
      "target": "67890",
      "label": "leads_project_at",
      "edge_type": "triplet",
      "confidence": 0.92
    },
    {
      "id": "ei_edge_100",
      "source": "ei:100",
      "target": "12345",
      "label": "preference",
      "edge_type": "insight"
    },
    {
      "id": "il_7",
      "source": "ei:100",
      "target": "ei:200",
      "label": "related_to",
      "edge_type": "insight_link",
      "confidence": 0.87
    }
  ],
  "insight_nodes": [
    {
      "id": "ei:100",
      "content": "Prefers async communication over meetings",
      "type": "preference",
      "confidence": 0.85,
      "source": "entity_insight"
    },
    {
      "id": "ki:50",
      "content": "Q1 planning cycle typically starts in December",
      "type": "fact",
      "confidence": 0.90,
      "source": "knowledge_insight"
    }
  ],
  "insights": {
    "12345": [
      {
        "content": "Prefers async communication over meetings",
        "confidence": 0.85,
        "type": "preference"
      }
    ]
  },
  "total_nodes": 250,
  "total_edges": 180,
  "total_context_edges": 3
}

Authorizations

Authorization

string

header

required

API key for authentication (format: "Bearer YOUR_API_KEY")

Query Parameters

limit

integer

default:1000

Maximum number of entity nodes to return

Required range: 1 <= x <= 5000

format

enum<string>

Response format. Use html to receive an interactive graph visualization instead of JSON.

Available options:

json,

html

Response

Knowledge graph data

Complete knowledge graph for the workspace

nodes

object[]

CRM entity nodes (people, companies, custom objects) and ghost nodes (entities mentioned in context but not yet resolved to a CRM record)

Hide child attributes

nodes.id

string

Node ID. CRM entities use numeric IDs (e.g., "12345"). Ghost nodes use the format "ghost:<context_id>".

Example:

"12345"

nodes.name

string

Display name (primary attribute value, or a generated name for unnamed entities)

Example:

"John Smith"

nodes.type

string

Entity type name (e.g., "person", "company", "ghost")

Example:

"person"

nodes.definition_slug

string

Slug of the entity definition

Example:

"person"

nodes.primary_attribute

string

Value of the entity's primary attribute

nodes.created_at

string<date-time>

When the entity was created

edges

object[]

CRM relationship edges between entity nodes (e.g., person works_at company)

Hide child attributes

edges.id

string

Edge ID

edges.source

string

Source node ID

edges.target

string

Target node ID

edges.label

string

Relationship label (e.g., "works_at", "reports_to")

Example:

"works_at"

edges.definition_id

string

ID of the relationship definition

relationship_definitions

object[]

Definitions of relationship types in the workspace

Hide child attributes

relationship_definitions.id

string

Definition ID

relationship_definitions.name

string

Display name

relationship_definitions.entity_1_to_2

string

Label from entity 1 to entity 2 (e.g., "works_at")

relationship_definitions.entity_2_to_1

string

Inverse label from entity 2 to entity 1 (e.g., "employs")

context_edges

object[]

Edges derived from ingested context, AI-extracted triplets, insight-entity links, and insight-to-insight semantic relationships. Each edge has an edge_type indicating its source.

Hide child attributes

context_edges.id

string

Edge ID. Prefixed by type: "ctx_" for context, "tri_" for triplet, "ei_edge_" for entity insight edges, and "il_" for insight_link edges.

Example:

"tri_42"

context_edges.source

string

Source node ID. Can be an entity ID, ghost ID ("ghost:123"), or insight node ID ("ei:123" or "ki:123").

Example:

"12345"

context_edges.target

string

Target node ID

Example:

"67890"

context_edges.label

string

Edge label. For context edges: the relationship type. For triplets: the predicate (e.g., "works_with"). For insight edges: the insight type. For insight_link edges: "related_to" or "contradicts".

Example:

"works_with"

context_edges.edge_type

enum<string>

Source type of the edge

Available options:

context,

triplet,

insight,

insight_link

context_edges.confidence

number<double> | null

Confidence score (0-1). Present on triplet edges (extraction confidence) and insight_link edges (similarity score). Null/absent for context and insight edges.

Required range: 0 <= x <= 1

insight_nodes

object[]

Insight nodes representing AI-extracted knowledge (entity insights and workspace-level knowledge insights)

Hide child attributes

insight_nodes.id

string

Insight node ID. Entity insights use "ei:", knowledge insights use "ki:".

Example:

"ei:42"

insight_nodes.content

string

The insight text content

Example:

"John prefers email communication over phone calls"

insight_nodes.type

string

Insight type (e.g., "preference", "behavior", "fact")

Example:

"preference"

insight_nodes.confidence

number<double>

Confidence score (0-1)

Required range: 0 <= x <= 1

insight_nodes.source

enum<string>

Source type: "entity_insight" (scoped to an entity) or "knowledge_insight" (workspace-level)

Available options:

entity_insight,

knowledge_insight

insights

object

Per-entity insight summaries, keyed by entity ID

Hide child attributes

insights.{key}

object[]

Hide child attributes

insights.{key}.content

string

The insight text content

insights.{key}.confidence

number<double>

Confidence score (0-1)

Required range: 0 <= x <= 1

insights.{key}.type

string

Insight type

total_nodes

integer

Total number of entity nodes in the workspace (may exceed the returned count due to the limit parameter)

total_edges

integer

Total number of CRM relationship edges in the workspace

total_context_edges

integer

Total number of context edges returned (includes all edge types)

Last modified on March 19, 2026

Stream insights (SSE)List integrations

API Documentation

Schema

Objects

Records

Relationships

Lists

Tasks

Notes

Timeline

Search

Context

Insights

Graph

Integrations

AI Lists

Compounding

Lists

Objects

Schema

Records

Context

AI Lists

Relationships

Timeline

Tasks

Notes

Insights

Search

Graph

Integrations

Compounding

​Overview

​Node Types

​Edge Types

​Insight Link Edges

​Edge Priority

​Interactive Visualization

​Example

Authorizations

Query Parameters

Response

Overview

Node Types

Edge Types

Insight Link Edges

Edge Priority

Interactive Visualization

Example