SchemaExtractSchemaExtract home
Open extractor
Skip to extractor

Document → JSON, on your schema

Turn invoices, packing lists, and bills of lading into JSON you can ship.

Upload a PDF or image, pick a preset or paste a JSON schema, and get typed fields back — vendor, totals, line items, packages, containers. OpenRouter-compatible free models when a cloud key is set; otherwise local Ollama if it is running; otherwise a labeled mock filler.

Open the extractorRead the API
Presets
Invoice · packing · bank · BoL
Files
PDF, PNG, JPG, WebP
Output
JSON + CSV

How it works

Three steps. No training set.

SchemaExtract is for operators who already know the shape of the data. You are not fine-tuning a model — you are declaring the contract.

  1. 01

    Drop a document

    PDF or image. Text extract plus optional page rasterization for vision. First three pages. Discarded after the response.

  2. 02

    Lock the schema

    Choose a preset or paste your own JSON Schema / example object. Extra keys are never invented.

  3. 03

    Ship the JSON

    Pretty result, copy, download .json, or flatten line items to CSV. Same contract as POST /v1/extract.

Presets

Start from a document you already have.

Four shipping and finance schemas. Open a guide, or jump straight into the extractor.

invoice
Invoice
Commercial invoices: vendor, invoice number, date, currency, total, and line items.
  • vendor
  • invoice_number
  • date
  • currency
  • total
  • line_items
packing_list
Packing list
Packing lists: shipper, consignee, marks, and packages with quantity and weight.
  • shipper
  • consignee
  • marks
  • packages
bank_statement
Bank statement
Simple statements: account holder, period, and transactions with amount and balance.
  • account_holder
  • period_start
  • period_end
  • transactions
bol
Bill of lading
Ocean BoL: number, parties, vessel, ports, freight terms, and container rows.
  • bol_number
  • shipper
  • consignee
  • vessel
  • port_of_loading
  • port_of_discharge
  • containers

Extractor

Playground

Same path the API uses. Cloud keys win if set; otherwise a reachable Ollama daemon is used. Mock filler is last and labeled as such. Files never leave the request.

Document

File preview appears here after you choose a document.
Custom

Object root only. Max 50 top-level keys. JSON Schema or a plain example object both work.

Cloud and mock post to POST /v1/extract. Ollama uses POST /v1/extract/jobs and polls so long local runs survive a proxy timeout. 402 on quota, 429 on burst limit.

Result

Pretty JSON plus copy, download, and CSV flatten.

No extraction yet

Missing fields come back as null. Extra keys are never added.

Pricing

Free 20 · Starter $29 · Pro $99

Start free. Quotas are enforced on POST /v1/extract. Stripe Checkout and the Customer Portal turn on when STRIPE_SECRET_KEY and price IDs are set — otherwise the paid CTAs stay disabled and polished.

Free

$0

20 extracts / month. Mock or your own model key.
  • All four presets + custom schema
  • Web extractor + JSON/CSV
  • Demo mode without billing

Included with the playground.

Starter
Most desks

$29

500 extracts / month. For a single ops desk.
  • API keys
  • Email magic-link workspace
  • Stripe Checkout when configured
Pro

$99

5,000 extracts / month. Usage overage billed later.
  • Higher quota
  • Customer portal
  • Priority path (same API)