Quick Start

Upload a document and retrieve structured data in under two minutes.

---

01 — Upload a document

Send your file as multipart/form-data. Supported formats: PDF, Word, Excel, images, and emails.

cURL

curl -X POST https://api.feld.ai/v1/documents \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "file=@invoice.pdf" \
  -F 'metadata={"vendor": "Acme Corp"}'

Python

import requests

response = requests.post(
    "https://api.feld.ai/v1/documents",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    files={"file": ("invoice.pdf", open("invoice.pdf", "rb"))},
    data={"metadata": '{"vendor": "Acme Corp"}'},
)
document = response.json()
print(document["id"])  # e.g. "d7a8f3e1-..."

Response — 201 Created

{
  "type": "document",
  "id": "d7a8f3e1-4b2c-4e5a-9f1d-6c8b3a2e7d4f",
  "name": "invoice.pdf",
  "status": "pending",
  "page_count": null,
  "created_at": "2025-01-15T10:30:00Z"
}

Save the id — you'll need it for the next steps.

---

02 — Wait for processing

The document moves through pending → processing → completed. Poll the status endpoint until it's ready.

cURL

curl https://api.feld.ai/v1/documents/{document_id} \
  -H "Authorization: Bearer YOUR_API_KEY"

Python

import time

doc = {"status": "pending"}
while doc["status"] not in ("completed", "failed"):
    time.sleep(2)
    doc = requests.get(
        f"https://api.feld.ai/v1/documents/{document_id}",
        headers={"Authorization": "Bearer YOUR_API_KEY"},
    ).json()

tipp

Most documents are processed within 10–30 seconds.

---

03 — Extract structured data

Once the document is completed, trigger an extraction job with your schema.

Indexing required

Summary, extraction, classification, and splitting jobs all require the document to be indexed (status completed) before they can run. If you submit a job while the document is still indexing, the job waits for indexing to finish, but may fail if indexing does not complete within the processing window. For time-sensitive workflows, upload documents in advance so indexing happens before you need the results.

cURL

curl -X POST https://api.feld.ai/v1/extractions/jobs \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "document_id": "{document_id}",
    "schema_id": "{schema_id}"
  }'

Python

job = requests.post(
    "https://api.feld.ai/v1/extractions/jobs",
    headers={
        "Authorization": "Bearer YOUR_API_KEY",
        "Content-Type": "application/json",
    },
    json={
        "document_id": document_id,
        "schema_id": schema_id,
    },
).json()
print(job["id"])  # job_id for polling

---

04 — Retrieve results

Poll the job until status is completed, then read the extracted fields:

cURL

curl https://api.feld.ai/v1/extractions/jobs/{job_id} \
  -H "Authorization: Bearer YOUR_API_KEY"

Python

result = requests.get(
    f"https://api.feld.ai/v1/extractions/jobs/{job_id}",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
).json()
print(result["data"])

---

Next steps

• Authentication — API key management and security
• Error handling — Error types and status codes
• API Reference — Full endpoint documentation