Bank statement extraction API
REST API to convert PDF bank statements into structured JSON. 99% accuracy, auto-refunded on failure.
$ curl https://api.bankstatementlab.com/v1/extractions \
-H "Authorization: Bearer bsl_live_..." \
-F "file=@statement.pdf"Bank statements extraction
Extract transactions, balances and metadata from any PDF.
Structured JSON output
Predictable schema across 10 supported languages.
99% accuracy
Auto-refunded on failure. No charge for failed extractions.
Sync & async modes
Synchronous for small PDFs, async + polling for large batches.
Code samples
Integrate in any language. Examples below.
$ curl https://api.bankstatementlab.com/v1/extractions \
-H "Authorization: Bearer bsl_live_..." \
-F "file=@statement.pdf"const form = new FormData();
form.append("file", fs.createReadStream("statement.pdf"));
const res = await fetch("https://api.bankstatementlab.com/v1/extractions", {
method: "POST",
headers: { Authorization: `Bearer $${process.env.BSL_KEY}` },
body: form,
});
const data = await res.json();import requests
with open("statement.pdf", "rb") as f:
r = requests.post(
"https://api.bankstatementlab.com/v1/extractions",
headers={"Authorization": f"Bearer {BSL_KEY}"},
files={"file": f},
)
data = r.json()$ch = curl_init("https://api.bankstatementlab.com/v1/extractions");
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_POST, true);
curl_setopt($ch, CURLOPT_HTTPHEADER, ["Authorization: Bearer " . getenv("BSL_KEY")]);
curl_setopt($ch, CURLOPT_POSTFIELDS, [
"file" => new CURLFile("statement.pdf"),
]);
$data = json_decode(curl_exec($ch), true);Response shapes
The data, column_names and transaction_count fields are only present when status equals "completed". Otherwise they are omitted from the JSON (absent, not null).
GET /v1/extractions/:id
200 OK — Extraction complete
HTTP/1.1 200 OK
Content-Type: application/json
{
"id": "ext_3f8a9b2c1d4e5f6a7b8c9d0e",
"status": "completed",
"file_name": "statement-march-2026.pdf",
"created_at": "2026-03-15T10:23:00.000Z",
"completed_at": "2026-03-15T10:23:18.000Z",
"page_count": 3,
"credits_charged": 3,
"api_key_id": "ak_abc123",
"column_names": ["Date", "Description", "Amount"],
"transaction_count": 42,
"data": {
"columns": ["Date", "Description", "Amount"],
"transactions": [
["2026-03-01", "VIREMENT SEPA", "1500.00"],
["2026-03-02", "CB CARREFOUR", "-45.20"]
]
}
}200 OK — Extraction in progress
HTTP/1.1 200 OK
Content-Type: application/json
{
"id": "ext_3f8a9b2c1d4e5f6a7b8c9d0e",
"status": "processing",
"file_name": "statement-march-2026.pdf",
"created_at": "2026-03-15T10:23:00.000Z",
"page_count": 3,
"credits_charged": 3,
"api_key_id": "ak_abc123"
}200 OK — Extraction failed
HTTP/1.1 200 OK
Content-Type: application/json
{
"id": "ext_3f8a9b2c1d4e5f6a7b8c9d0e",
"status": "failed",
"file_name": "statement-march-2026.pdf",
"created_at": "2026-03-15T10:23:00.000Z",
"completed_at": "2026-03-15T10:23:32.000Z",
"page_count": 3,
"credits_charged": 3,
"api_key_id": "ak_abc123",
"error": {
"type": "extraction_failed",
"message": "Extraction failed due to an internal error. Contact support if the issue persists."
}
}Polling pattern
When an extraction runs in async mode, the response contains status: "processing" until the worker has finished. Always check that status equals "completed" BEFORE reading the data, transaction_count and column_names fields — they are omitted from the JSON until extraction is complete.
Recommended client-side polling frequency: 5 to 10 seconds. The server-side worker runs every 10 seconds — polling faster will not speed up results.
GET /v1/extractions
200 OK — Mixed page (one element per state)
HTTP/1.1 200 OK
Content-Type: application/json
{
"data": [
{
"id": "ext_001",
"status": "completed",
"file_name": "march.pdf",
"created_at": "2026-03-15T10:00:00.000Z",
"completed_at": "2026-03-15T10:00:18.000Z",
"page_count": 3,
"credits_charged": 3,
"api_key_id": "ak_abc123",
"column_names": ["Date", "Description", "Amount"],
"transaction_count": 42
},
{
"id": "ext_002",
"status": "processing",
"file_name": "april.pdf",
"created_at": "2026-04-15T10:00:00.000Z",
"page_count": 5,
"credits_charged": 5,
"api_key_id": "ak_abc123"
},
{
"id": "ext_003",
"status": "failed",
"file_name": "may.pdf",
"created_at": "2026-05-15T10:00:00.000Z",
"completed_at": "2026-05-15T10:00:42.000Z",
"page_count": 2,
"credits_charged": 2,
"api_key_id": "ak_abc123",
"error": {
"type": "extraction_failed",
"message": "Extraction failed due to an internal error. Contact support if the issue persists."
}
}
],
"has_more": false
}Polling pattern
When an extraction runs in async mode, the response contains status: "processing" until the worker has finished. Always check that status equals "completed" BEFORE reading the data, transaction_count and column_names fields — they are omitted from the JSON until extraction is complete.
Recommended client-side polling frequency: 5 to 10 seconds. The server-side worker runs every 10 seconds — polling faster will not speed up results.
Built for
Accounting automation
Import client statements directly into your bookkeeping pipeline.
Lending due diligence
Verify income and cash flow from applicant bank statements.
Personal finance apps
Let users connect their statements without manual data entry.
Frequently asked questions
Send your API key as a Bearer token in the Authorization header. We use API keys — no OAuth in v1.
10 requests/second and 1,000 requests/hour per API key. Need higher limits? Contact us.
We target 99% uptime on a best-effort basis. No contractual SLA in v1 — reach out for enterprise commitments.
Failed extractions are automatically refunded to your credit balance. You only pay for successful pages.
PDF only in v1. Max 20 MB per file. Image and CSV support coming later.
English, French, Spanish, German, Italian, Portuguese, Japanese, Dutch, Korean, Hindi.