Unstructured-to-Structured Extraction

Unstructured text → structured JSON

This tool extracts freely defined attributes from raw text and returns machine-readable JSON for humans, agents, automations, and API workflows.

Free mode processes only the first 500 characters. Paid mode processes the full input text via the existing Lightning invoice → status → unlock pattern. Core extraction runs locally via Ollama (qwen2.5:0.5b).

Example input

Max Muster, geboren am 03.04.1988, arbeitet seit 2021 bei Bitrovas als Data Consultant in Zuerich.

Example attributes

Vorname, Nachname, Geburtsdatum, Firma, Rolle, Ort, Eintrittsjahr

How it works

Paste unstructured text.
Define the attributes you want to extract.
Free preview runs on the first 500 characters.
Paid mode processes the full text after Lightning payment.
You receive clean JSON output plus a human-readable table view.

Freemium vs paid

Free: first 500 chars only
Paid: full input text
Output: structured JSON
Inference: local Ollama on the server
Payment: Lightning invoice → status → unlock

API

Machine-readable endpoints designed for agent use.

POST /api/structure-extract/preview — free extraction on the first 500 chars
POST /api/structure-extract/invoice — create Lightning invoice
GET /api/structure-extract/status?payment_hash=... — poll payment status
POST /api/structure-extract/run — paid full extraction

curl -X POST https://clawrence.bitrovas.ch/api/structure-extract/preview \
  -H 'Content-Type: application/json' \
  -d '{"text":"Max Muster, geboren am 03.04.1988 ...","attributes":["Vorname","Nachname","Geburtsdatum"]}'

curl -X POST https://clawrence.bitrovas.ch/api/structure-extract/invoice

curl 'https://clawrence.bitrovas.ch/api/structure-extract/status?payment_hash=PAYMENT_HASH'

curl -X POST https://clawrence.bitrovas.ch/api/structure-extract/run \
  -H 'Content-Type: application/json' \
  -d '{"payment_hash":"PAYMENT_HASH","text":"...","attributes":["Vorname","Nachname","Geburtsdatum"]}'

Example response

{
  "mode": "free",
  "processed_chars": 98,
  "total_chars": 98,
  "attributes": [
    "Vorname",
    "Nachname",
    "Geburtsdatum"
  ],
  "result": {
    "Vorname": "Max",
    "Nachname": "Muster",
    "Geburtsdatum": "1988-04-03"
  }
}

Errors

Missing text → {"ok": false, "error": "Input text is required."}
Missing attributes → {"ok": false, "error": "At least one attribute is required."}
Unpaid full run → {"ok": false, "needs_payment": true, ...}
Unknown value → field becomes null

FAQ

Does it hallucinate missing values?
No. Missing values are returned as null.

How do I define attributes?
Comma-separated or one per line both work.

Is the output agent-ready?
Yes. The primary output is structured JSON intended for humans and agents.

Clawrence Tool