Export API v2: bulk object exports

Use Export API v2 to download bulk CSV data for objects in your Enterpret workspace. Discover the available objects, choose a fixed time window, and download every file returned across all pages.

This API exports changes processed in the requested windows. It does not provide a complete point-in-time snapshot or a feed of deleted-record IDs.

If you are migrating a v1 client, review Data Export API v1: legacy reference. V2 object and time-window requests are a different contract from v1 filtered feedback requests and query-ID polling.

Objects and data

Authentication

Use a Data Export API token for the intended Enterpret workspace. The token must have the dataExport scope. Send it in the Authorization: Bearer YOUR_DATA_EXPORT_TOKEN header on each API request. Keep the token and returned download URLs private.

When available to your role, open Settings → API Access → Data Export API to generate or copy the appropriate token. If these controls are unavailable, ask your Enterpret Owner/Manager or Enterpret team.

Base URL

https://api.enterpret.com/export

Every example below includes the full URL. Replace all uppercase placeholders before running a request.

Endpoints

1. List Objects

POST /external/v2/objects

Example request

curl --request POST 'https://api.enterpret.com/export/external/v2/objects' \
  --header 'Authorization: Bearer YOUR_DATA_EXPORT_TOKEN' \
  --header 'Content-Type: application/json' \
  --data '{}'

Illustrative response

Example values below are placeholders, not a workspace object dictionary.

{
  "objects": [
    {
      "id": "OBJECT_ID_FROM_LIST",
      "name": "Example object",
      "description": "Example description"
    }
  ]
}

2. Export Object Data

POST /external/v2/export

Timestamp format

Use UTC timestamps on whole-hour boundaries, such as 2026-10-01T00:00:00Z. RFC3339 timestamps and the minute-precision UTC form 2026-10-01T00:00Z are accepted. Both bounds are rounded down to the hour. If endTime is omitted, it defaults to the start of the current UTC hour.

Time windows and bounds

The start is inclusive and the end is exclusive. For example, 2026-10-01T00:00:00Z to 2026-10-02T00:00:00Z requests one day. Start must be earlier than end after rounding, and a request may span no more than one calendar year. This is a limit on the interval length, not a statement about how far back data is available.

These windows follow processing time in Enterpret, rather than the original feedback date. An update to old feedback can therefore appear in a later processing window.

Response

files contains short-lived signed CSV download URLs. cursor, when present, identifies where to continue checking the requested range. A page can contain no files and still return a cursor. Empty arrays can be omitted from the JSON.

Example request

curl --request POST 'https://api.enterpret.com/export/external/v2/export' \
  --header 'Authorization: Bearer YOUR_DATA_EXPORT_TOKEN' \
  --header 'Content-Type: application/json' \
  --data '{
    "objectID": "OBJECT_ID_FROM_LIST",
    "startTime": "2026-10-01T00:00:00Z",
    "endTime": "2026-10-02T00:00:00Z"
  }'

Illustrative response

{
  "files": [
    "SIGNED_CSV_URL_FROM_RESPONSE"
  ],
  "cursor": "CURSOR_FROM_RESPONSE"
}

The URL and cursor above are placeholders. Use the actual returned values unchanged.

Retrieve all pages

1. Download every URL in files promptly.

2. If the response has a nonempty cursor, repeat the export request with that cursor and the same objectID, startTime and endTime.

3. Continue until the cursor is absent or empty, even if a page has an empty or absent files array.

curl --request POST 'https://api.enterpret.com/export/external/v2/export' \
  --header 'Authorization: Bearer YOUR_DATA_EXPORT_TOKEN' \
  --header 'Content-Type: application/json' \
  --data '{
    "objectID": "OBJECT_ID_FROM_LIST",
    "startTime": "2026-10-01T00:00:00Z",
    "endTime": "2026-10-02T00:00:00Z",
    "cursor": "CURSOR_FROM_PREVIOUS_RESPONSE"
  }'

A page returns up to 10 files. File boundaries can cover hours, days, weeks or months; do not expect one file per hour. A cursor means there is more range to check, not necessarily more nonempty files.

Read the CSV files

Use a CSV parser that handles quoted values, commas and line breaks. Read each file’s header and map columns by name. Columns depend on the object and workspace, and can differ across files as the schema changes. Validate field types, missing values and relationship fields before loading them into your destination.

Do not treat the example object descriptions as a fixed data dictionary. Repeated updates can cause the same record ID to appear in different files or windows.

Errors and troubleshooting

Handle the HTTP status and retain the returned error details for troubleshooting. Error bodies vary between authentication and request handling; do not assume every failure has the same JSON wrapper.

If an export has no files, first check the exact listed object ID and the processing window. An empty result alone does not establish that no source records exist or that all recent processing is complete.

Request pacing

Use bounded retries with backoff and avoid repeatedly requesting the same range while waiting for data. Contact Enterpret Support for throughput or availability requirements for your integration.

Recommended workflow

1. List the workspace’s objects and choose the exact object ID.

2. Try a small, fixed, completed time window and inspect all returned CSV headers.

3. Download every page and file, keeping the request bounds unchanged.

4. Plan how your destination will handle retries, repeated records, schema changes, updates and deletions before scheduling a recurring export.

Frequently asked questions

Where is the data dictionary?

List Objects provides the available object IDs, names and descriptions. Inspect a sample export’s headers to see the actual columns. Ask Enterpret Support to confirm field meanings or types you need for your integration.

How do I join objects?

Relationship fields depend on the object schema. Confirm the actual exported keys and their meaning before building joins; do not infer them solely from similar field names.

How should I identify records?

Exported rows include record_id. Keep the workspace and object identity with the record ID in your destination. Do not assume IDs are globally unique across workspaces or object types, or that each ID appears only once across an entire export.

Are exports a snapshot of the latest state?

No. Files contain surviving changes folded within their processing windows. Records unchanged during the requested windows may be absent. Historical files can also be rebuilt after a repair. The API does not promise an immutable point-in-time snapshot or exactly-once delivery.

Are deletions included?

Deleted-record IDs are not returned. Deletions can affect which rows survive in a compacted file, but a later deletion does not send a tombstone telling your destination to remove an earlier imported row. Incremental exports alone therefore cannot keep deletions synchronized. Confirm an appropriate reconciliation approach with Enterpret Support if you need a replica.