> Documentation index: https://beezr.io/developers/api/llms.txt

# Keep another system in sync

## Keep the names and ids

Copy module and field System names from the Designer. Keep the Beezr record `id` alongside the corresponding record in your system so updates target the intended record.

## Make individual writes safe to retry

For create and update, attach an `Idempotency-Key` and retain it with the exact request body until you know the result. The same key and body return the stored response. A changed body with the same key returns `IDEMPOTENCY_KEY_REUSED`.

Use [Update a record](/developers/api/reference/update-record/) for copyable requests in cURL, JavaScript and Python.

If the API reports `IDEMPOTENCY_RESULT_UNKNOWN`, reconcile the record before starting another write. Don't change keys to force an unknown operation to run again.

## Batch when you know the work

Bulk create, update and delete handle 1–100 items per call and count as one request. Items run separately, in order. Inspect `failed` and `failures`; reconcile successes before resubmitting failed items. Bulk endpoints don't support idempotency, so don't blindly repeat an entire bulk create after an uncertain result.

```curl
curl --fail-with-body --silent --show-error \
  'https://services.beezr.io/generic/api/v1/records/bulk_update' \
  -H "X-API-Key: $BEEZR_API_KEY" \
  -H 'Content-Type: application/json' \
  --data '{
  "module": "cm_equipment_requests_base",
  "updates": [{"id":482913,"data":{"priority":1}},{"id":482914,"data":{"priority":1}}]
}'
```

## Apply backpressure

Keep at most 2 calls in flight. Wait `Retry-After` seconds on retryable 429s. Stop on `MONTHLY_QUOTA_EXCEEDED` until the 1st. Use the [retry helpers](/developers/api/limits/) and inspect business-rule messages before repeating HTTP 500 writes.

## Read back what you need

Use v4 query or get with `fields` to keep reads small. Query returns pages; continue while `offset + returned < total`. See [Paging](/developers/api/paging/) and [Safe retries](/developers/api/idempotency/).