POST /document/zero-shot

Runs zero-shot field discovery against a single document — the model inspects the file and proposes the fields it finds, with no prior template required. Use it when you are bootstrapping a new template, exploring an unknown document family, or validating what the AI is able to pick up before formalizing a schema.

📷 Paste here: screenshot of the Playground → Zero-shot screen on the platform, so the reader can compare the API output with the UI output side by side.

Endpoint

POST https://api-dev.xtrakt.ai/document/zero-shot

Authentication

Send your API key as a Bearer token. See Getting Started for how to generate one.

Authorization: Bearer xtkt_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxx

Request

Content-Type: multipart/form-data

Field	Type	Required	Description
file	file	✅	The document to analyze.

Constraints

• Allowed extensions: .pdf, .png, .jpg, .jpeg, .tiff, .tif, .bmp, .doc, .docx.
• Max size: 10 MB.
• Rate limit: 50 requests/minute per API key.

Response

200 OK — data contains the discovery id plus a metadata payload listing the fields the model detected:

{
  "data": {
    "id": "f2c1a3e4-7b8d-4c9e-9012-34567890abcd",
    "metadata": {
      "discoveredFields": [
        { "name": "supplier",       "value": "Acme Industries",   "confidence": 0.96 },
        { "name": "invoiceNumber",  "value": "INV-2026-00123",     "confidence": 0.94 },
        { "name": "issueDate",      "value": "2026-05-12",         "confidence": 0.92 },
        { "name": "total",          "value": "1542.30",            "confidence": 0.89 }
      ]
    },
    "createdDate": "2026-05-19T13:48:02Z"
  },
  "meta": { "httpStatusCode": 200, "messages": [] },
  "hasErrors": false
}

The exact list of discovered fields depends entirely on the document's content. There is no guarantee of which fields will appear from one call to the next.

Example — curl

curl -X POST "https://api-dev.xtrakt.ai/document/zero-shot" \
  -H "Authorization: Bearer xtkt_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxx" \
  -F "file=@./unknown-document.pdf"

Example — Python

import requests

API_KEY = "xtkt_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxx"
BASE_URL = "https://api-dev.xtrakt.ai"

with open("unknown-document.pdf", "rb") as fp:
    response = requests.post(
        f"{BASE_URL}/document/zero-shot",
        headers={"Authorization": f"Bearer {API_KEY}"},
        files={"file": ("unknown-document.pdf", fp, "application/pdf")},
    )

response.raise_for_status()
print(response.json()["data"]["metadata"]["discoveredFields"])

Example — Node.js

import fs from "node:fs";

const API_KEY = "xtkt_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxx";
const BASE_URL = "https://api-dev.xtrakt.ai";

const form = new FormData();
form.append("file", new Blob([fs.readFileSync("./unknown-document.pdf")]), "unknown-document.pdf");

const response = await fetch(`${BASE_URL}/document/zero-shot`, {
  method: "POST",
  headers: { Authorization: `Bearer ${API_KEY}` },
  body: form,
});

const { data } = await response.json();
console.log(data.metadata.discoveredFields);

Errors

HTTP status	Reason
400 Bad Request	Missing file, empty file, disallowed extension, or file over 10 MB.
401 Unauthorized	Missing/invalid/revoked API key.
403 Forbidden	Tenant has no active subscription for the zero-shot capability.
429 Too Many Requests	API key exceeded the 50 req/min limit.