Result
Fields will appear here as JSON, exactly as the API returns them.
Document OCR · Straventa
Read the document.
Indonesian paperwork, returned as typed fields instead of a blob of text. Eleven document types ship with field schemas, extraction runs synchronously or as a job with webhooks, and you can run the whole thing inside your own infrastructure.
Try it
Send a document, read the fields
This runs against the same extraction service the product uses. Pick a document type, upload a file, and read back the typed fields.
What it reads
Seven Indonesian document types, with schemas
Each type declares its own field schema, so extraction returns named fields rather than text you have to parse. Need one we do not ship? Put your own schema and it survives redeploys.
ktp
KTP — Kartu Tanda Penduduk
NIK, nama, tempat & tanggal lahir, alamat, and the rest as named fields
kk
KK — Kartu Keluarga
Household record and its members
npwp_personal
NPWP — personal
Personal taxpayer number
npwp_badan
NPWP — badan
Corporate taxpayer number
nib
NIB — Nomor Induk Berusaha
Business identification number
akta
Akta — akta pendirian
Deed of establishment
sk_kemenkumham
SK Kemenkumham
Ministry decree on legal-entity status
siup_tdp
SIUP / TDP
Trading licence and company registration
faktur_pajak
Faktur Pajak
Tax invoice header and line values
rekening_koran
Rekening Koran
Bank statements as transaction rows
freeform
Freeform
Anything without a seeded schema — you supply the field list
How it compares
Four ways to read an Indonesian document
The global OCR platforms are excellent general engines that know nothing about a KTP. The local KYC vendors know the document but sell you a whole onboarding flow. Written for whoever has to justify the choice.
| Criterion | Straventa Document OCR | Google Document AI | AWS Textract | In-house Tesseract |
|---|---|---|---|---|
| Indonesian document types | Eleven seeded types — KTP, Kartu Keluarga, NPWP personal and badan, NIB, Akta, SK Kemenkumham, SIUP/TDP, Faktur Pajak, Rekening Koran, and a freeform type for anything without a schema. | General forms, invoices, and a set of identity parsers that do not cover Indonesian national documents. | General forms, tables, invoices, and IDs; no Indonesian national document types. | None. You author every field rule yourself. |
| What comes back | Named fields declared by the document type's schema — a KTP returns nik, nama, tempat_lahir, alamat and the rest, ready to use. | Entities and key-value pairs you map onto your own field names. | Blocks, key-value pairs, and tables you map yourself. | Raw text and coordinates. The parsing is your project. |
| Deployment | Straventa cloud, or self-hosted inside your own Kubernetes or Docker — same product. | Google Cloud only. | AWS only. | Anywhere — you operate it. |
| Where the document goes | On a self-hosted deployment, nothing leaves your environment. | Google Cloud regions. | AWS regions. | Your environment. |
| Retention of extracted data | Set by a platform-owned document-type registry: non-KYC output expires on a 7-day transient clock, compliance-bound output such as KTP stays on the statutory class. A tenant cannot move a KTP schema onto the short clock. | Your own lifecycle configuration; nothing stops a misconfiguration. | Your own lifecycle configuration; nothing stops a misconfiguration. | Whatever you built, if you built it. |
| Custom document types | PUT your own field schema per tenant; your version survives redeploys and platform seeding. | Custom processors, trained and hosted in the vendor's platform. | Custom queries and adapters within the vendor's platform. | Yours to write, and yours to maintain. |
| Fits the rest of the stack | Extraction feeds KYC/KYB onboarding and reconciliation on the same platform, under one contract and one login through Straventa Accounts. | An integration project. | An integration project. | An integration project. |
| Pricing model | Quoted per volume, or per deployment when self-hosted. Contact us. | Per page, per processor, on the vendor's public rate card. | Per page, by API, on the vendor's public rate card. | No licence — you pay in engineering time and accuracy work. |
Competitor rows describe each vendor’s publicly documented model at time of writing; confirm current terms with the vendor before making a decision.
How it is priced
Priced on the volume you actually run — or per deployment when it lives in your data centre
Billing model
Per volume, or per deployment
- Cloud extraction is quoted against monthly document volume and the document types you use.
- Self-hosted is licensed per deployment, so a high-volume year does not become a per-page bill.
- Compliance-bound documents carry their statutory retention regardless of tier — that is not a paid add-on.
- Billing is not yet active and no payment is collected today; talk to us and we will quote the shape that fits.
Bring us a stack of documents.
Send a representative sample and we will tell you the extraction accuracy you should expect on your own paperwork, before you commit to anything.