Document processing

Read the text out of a PDF, Word or Excel file — Lao included — and, if you ask, have it summarized, reduced to key points, translated, or turned into structured fields. It runs as an asynchronous job: enqueue it, then poll the job.

POST /v1/docproc/jobs API key

Fields

`file` * file	A PDF, Word (.docx) or Excel (.xlsx) file. Over 10 MB: upload it first with POST /v1/docproc/uploads, then pass sourceKey in a JSON body.
`task` extract \| summarize \| key_points \| translate \| structured	What to do with the text. Default: extract (the text only, no model).
`targetLanguage` string	For translate: the language to translate into, e.g. "English" or "Lao".
`schemaHint` string	For structured: the fields you want, e.g. "invoice number, date, total".

Enqueue a job

curl -X POST https://ai-api.vanguardinitiative.com/v1/docproc/jobs \
  -H "authorization: Bearer $API_KEY" \
  -F file=@report.pdf -F task=summarize

const form = new FormData();
form.append('file', file);
form.append('task', 'summarize');
const res = await fetch('https://ai-api.vanguardinitiative.com/v1/docproc/jobs', {
  method: 'POST',
  headers: { authorization: 'Bearer ' + API_KEY },
  body: form,
});
const { jobId } = await res.json();

import requests
with open('report.pdf', 'rb') as f:
    job_id = requests.post(
        'https://ai-api.vanguardinitiative.com/v1/docproc/jobs',
        headers={'authorization': f'Bearer {API_KEY}'},
        files={'file': f},
        data={'task': 'summarize'},
    ).json()['jobId']

Poll for the result

Poll GET /v1/jobs/{id} until status is success or error. The text is in result.text; the task’s answer (summary, key points, translation, fields) is in summary.markdown.

curl https://ai-api.vanguardinitiative.com/v1/jobs/$JOB_ID \
  -H "authorization: Bearer $API_KEY"

200 Finished job

{
  "status": "success",
  "job": {
    "id": "6d0b…",
    "serviceId": "docproc",
    "status": "success",
    "result": { "text": "…", "charCount": 18234, "pageCount": 12, "task": "summarize" },
    "summary": { "markdown": "## Summary\n…" }
  }
}