Memory APIs

Memory APIs power institutional knowledge graphs for departments and lists — create memory graphs, upload data, ingest connector content, generate graphs from search results, and chat over graph knowledge.

Overview

The Memory service lets you:

List and manage departments – List knowledge graphs (departments/lists), fetch by graph_id, and create new department graphs
Upload data – Ingest CSV or other files into a graph via multipart upload
Ingest connectors – Pull content from Enterprise Search connectors (e.g. Google Drive, Notion) into a graph
Generate memory – Build a knowledge graph from company or people search list records
Chat – Ask questions against a graph and receive answers with related nodes and edges

Base URL: https://graph-kb.chordian.ai

Authentication: Bearer token. Send your API key in the header: Authorization: Bearer <api-key>.

OpenAPI specification: https://graph-kb.chordian.ai/openapi.json

List Departments

List all knowledge graph instances (departments and list-backed graphs) for the authenticated tenant.

GET /memory/departments

Authentication: Authorization: Bearer <api-key>

Successful Response (200 OK)

{
  "departments": [
    {
      "graph_id": "string",
      "name": "string",
      "description": "string",
      "created_at": "string",
      "updated_at": "string",
      "node_count": 0,
      "edge_count": 0,
      "source_type": "string",
      "status": "string"
    }
  ]
}

Field	Type	Description
`departments`	array<object>	All graph instances for the tenant.
`departments[].graph_id`	string	Unique graph identifier (use as `list_id` for upload, generate, and chat).
`departments[].name`	string	Display name of the knowledge graph.
`departments[].description`	string	Human-readable description.
`departments[].created_at`	string	ISO timestamp when the graph was created.
`departments[].updated_at`	string	ISO timestamp when the graph was last updated.
`departments[].node_count`	integer	Number of nodes in the graph.
`departments[].edge_count`	integer	Number of edges in the graph.
`departments[].source_type`	string	Origin type, e.g. `department` or `list`.
`departments[].status`	string	Graph readiness, e.g. `empty` or `ready`.

Example

curl "https://graph-kb.chordian.ai/memory/departments" \
  -H "Authorization: Bearer <api-key>"

Get Department by ID

Fetch a single knowledge graph by graph_id.

GET /memory/departments/{graph_id}

Authentication: Authorization: Bearer <api-key>

Path Parameters

Parameter	Type	Required	Description
`graph_id`	string	✅	Graph identifier (same as `list_id` on related endpoints).

Successful Response (200 OK)

{
  "graph_id": "string",
  "name": "string",
  "description": "string",
  "created_at": "string",
  "updated_at": "string",
  "node_count": 0,
  "edge_count": 0,
  "source_type": "string",
  "status": "string"
}

Field	Type	Description
`graph_id`	string	Unique graph identifier.
`name`	string	Display name of the knowledge graph.
`description`	string	Human-readable description.
`created_at`	string	ISO timestamp when the graph was created.
`updated_at`	string	ISO timestamp when the graph was last updated.
`node_count`	integer	Number of nodes in the graph.
`edge_count`	integer	Number of edges in the graph.
`source_type`	string	Origin type, e.g. `department` or `list`.
`status`	string	Graph readiness, e.g. `empty` or `ready`.

Example

curl "https://graph-kb.chordian.ai/memory/departments/{graph_id}" \
  -H "Authorization: Bearer <api-key>"

Create Department

Create a new department knowledge graph.

POST /memory/departments/create

Authentication: Authorization: Bearer <api-key>

Request Body

{
  "name": "string",
  "description": "string",
  "source_type": "string"
}

Field	Type	Required	Description
`name`	string	✅	Display name for the new knowledge graph.
`description`	string	❌	Description of the graph’s purpose.
`source_type`	string	❌	Origin type (e.g. `department`).

Successful Response (200 OK)

{
  "graph_id": "string",
  "tenant_id": "string",
  "list_id": "string",
  "name": "string",
  "description": "string",
  "created_at": "string",
  "updated_at": "string",
  "node_count": 0,
  "edge_count": 0,
  "source_type": "string",
  "status": "string"
}

Field	Type	Description
`graph_id`	string	New graph identifier.
`tenant_id`	string	Tenant that owns the graph.
`list_id`	string	List identifier (often matches `graph_id`).
`name`	string	Display name.
`description`	string	Description.
`created_at`	string	ISO timestamp when created.
`updated_at`	string	ISO timestamp when last updated.
`node_count`	integer	Initial node count (typically `0`).
`edge_count`	integer	Initial edge count (typically `0`).
`source_type`	string	Origin type.
`status`	string	Initial status (e.g. `empty`).

Example

curl -X POST "https://graph-kb.chordian.ai/memory/departments/create" \
  -H "Authorization: Bearer <api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "string",
    "description": "string",
    "source_type": "string"
  }'

Upload Data

Upload a file and start document ingestion into a knowledge graph. Processing runs in the background.

POST /memory/upload

Authentication: Authorization: Bearer <api-key>

Request Body (multipart/form-data)

Field	Type	Required	Description
`file`	file	✅	Data file to ingest (e.g. CSV).
`list_id`	string	✅	Target graph `graph_id` / list identifier.

Successful Response (200 OK)

{
  "jobId": "string",
  "status": "string",
  "message": "string"
}

Field	Type	Description
`jobId`	string	Background job identifier for ingestion.
`status`	string	Job state (e.g. `queued`).
`message`	string	Human-readable status message.

Example

curl -X POST "https://graph-kb.chordian.ai/memory/upload" \
  -H "Authorization: Bearer <api-key>" \
  -F "file=@<path-to-file>.csv" \
  -F "list_id=<graph_id>"

Ingest Connector Data

Ingest content from Enterprise Search connectors into a knowledge graph. Processing runs in the background.

Use Get Connector Status (GET /manage/admin/connector/status) in Enterprise Search APIs to list configured connectors. Set source_types to an array of connector.source values (for example ["google_drive"]) to ingest specific sources, or pass null to ingest all sources.

POST /memory/ingest/connectors

Authentication: Authorization: Bearer <api-key>

Request Body

{
  "list_id": "string",
  "source_types": null,
  "mode": "string",
  "graph_name": "string"
}

Field	Type	Required	Description
`list_id`	string	✅	Target graph identifier.
`source_types`	array<string> \| null	✅	Connector sources to ingest (e.g. `["google_drive", "notion"]`). Must be `null` to ingest all sources — do not omit this field.
`mode`	string	❌	Sync mode: `full` or `incremental` (default: `full`).
`graph_name`	string	❌	Display name; auto-creates a graph instance when set.

Successful Response (200 OK)

{
  "jobId": "string",
  "status": "string",
  "message": "string"
}

Field	Type	Description
`jobId`	string	Background job identifier.
`status`	string	Job state (e.g. `queued`).
`message`	string	Human-readable status message.

Example

curl -X POST "https://graph-kb.chordian.ai/memory/ingest/connectors" \
  -H "Authorization: Bearer <api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "list_id": "string",
    "source_types": null,
    "mode": "string",
    "graph_name": "string"
  }'

Generate Memory

Generate a knowledge graph from a company search list (rows and columns from a saved list).

Call Get Lists (GET /company-search/getLists) in Company Search APIs, choose a list from data, then pass that list’s rows as records and its columns as columns in this request body. Set listId to the selected list’s id.

POST /memory/generate

Authentication: Authorization: Bearer <api-key>

Request Body

{
  "listId": "string",
  "workflowCategory": "string",
  "records": [
    {
      "cells": {
        "column_id": {
          "value": "string",
          "status": "string",
          "metadata": {},
          "expandable": false,
          "_id": "string"
        }
      },
      "step_source": "string",
      "metadata": {
        "source": "string",
        "_id": "string",
        "uniqueID": "string"
      }
    }
  ],
  "columns": [
    {
      "id": "string",
      "header": "string",
      "type": "string"
    }
  ]
}

Field	Type	Required	Description
`listId`	string	✅	Target graph / list identifier (`graph_id`).
`workflowCategory`	string	❌	Workflow category hint (may be empty).
`records`	array<object>	✅	Search result rows with `cells`, `step_source`, and `metadata`.
`columns`	array<object>	✅	Column definitions (`id`, `header`, `type` e.g. `fixed` or `dynamic`).

Each item in records[].cells is keyed by column id from columns. Cell objects include value, status, metadata, expandable, and _id. Column type is typically fixed or dynamic.

Successful Response (200 OK)

{
  "jobId": "string",
  "status": "string",
  "message": "string"
}

Field	Type	Description
`jobId`	string	Background job identifier for graph generation.
`status`	string	Job state (e.g. `queued`).
`message`	string	Human-readable status message.

Example

curl -X POST "https://graph-kb.chordian.ai/memory/generate" \
  -H "Authorization: Bearer <api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "listId": "string",
    "workflowCategory": "string",
    "records": [
      {
        "cells": {
          "column_id": {
            "value": "string",
            "status": "string",
            "metadata": {},
            "expandable": false,
            "_id": "string"
          }
        },
        "step_source": "string",
        "metadata": {
          "source": "string",
          "_id": "string",
          "uniqueID": "string"
        }
      }
    ],
    "columns": [
      {
        "id": "string",
        "header": "string",
        "type": "string"
      }
    ]
  }'

Memory Chat

Ask a natural-language question against a knowledge graph. Returns an answer plus graph nodes and edges used in the response.

POST /memory/chat

Authentication: Authorization: Bearer <api-key>

Request Body

{
  "listId": "string",
  "question": "string"
}

Field	Type	Required	Description
`listId`	string	✅	Graph identifier to query (`graph_id`).
`question`	string	✅	Natural-language question.

Successful Response (200 OK)

{
  "answer": "string",
  "nodes": [
    {
      "id": "string",
      "label": "string",
      "properties": {}
    }
  ],
  "edges": [
    {
      "from": "string",
      "to": "string",
      "label": "string"
    }
  ]
}

Field	Type	Description
`answer`	string	Natural-language answer from the graph assistant.
`nodes`	array<object>	Graph nodes referenced in the response (`id`, `label`, `properties`).
`edges`	array<object>	Relationships between nodes (`from`, `to`, `label`).

Node label values include types such as Person, Company, Position, Location, and ScanReport. properties vary by label (e.g. name, linkedin_url on Person).

Example

curl -X POST "https://graph-kb.chordian.ai/memory/chat" \
  -H "Authorization: Bearer <api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "listId": "string",
    "question": "string"
  }'

Next Steps

Explore Enterprise Search APIs for connector-backed ingestion
Use Company Search APIs and People Search APIs to build lists for graph generation

Enterprise Search APIs MCP Server