Dokyumi
Two-stage AI pipeline · Mistral OCR + Claude

Describe your document.
Get an instant API.

Skip the AWS setup, the complex SDKs, and the per-page billing surprises. Define your extraction schema in plain English, get a dedicated API endpoint, and start parsing in minutes.

100 free extractions/month. No credit card required.

Or try the demo — no sign-up required →

10x
cheaper than pure vision models
< 2 min
from schema to live API
Any doc
PDFs, images, scans, photos
Zero infra
no AWS or GCP account needed

How it works

Other tools give you a raw text dump. Dokyumi gives you exactly the fields you asked for, validated, in JSON.

1

Define Your Schema

Describe your document type and the fields you want to extract in plain English. Invoices, bank statements, tax forms — anything. AI infers the schema for you.

2

Get Your API Endpoint

We generate a dedicated extraction endpoint scoped to your schema. One curl command, one API key, one POST request. You're done.

3

Get Structured JSON

Upload any document and receive validated, structured JSON back instantly. OCR caching means repeated docs cost nothing extra.

What people are parsing

Any document with a repeatable structure is a candidate. Here's what Dokyumi handles well.

🧾

Invoice Processing

Extract vendor name, invoice number, line items, totals, and due dates. Feed directly into your accounting system.

→ vendor, amount, line_items[], due_date
🏦

Bank Statements

Pull transactions, balances, account numbers, and date ranges from any bank's PDF format — no bank-specific integration required.

→ transactions[], opening_balance, closing_balance
📋

Insurance Claims

Extract claim numbers, policy details, loss descriptions, and coverage amounts. Automate intake without a human in the loop.

→ claim_number, policy_id, loss_date, amount
📊

Tax Documents

Parse W-2s, 1099s, Schedule Cs, and other tax forms into clean structured data. No more manual data entry for tax software.

→ wages, federal_tax_withheld, employer_ein
🏥

Medical Records

Extract diagnoses, medication lists, lab values, and provider info from clinical documents. Structured output ready for EHR import.

→ diagnoses[], medications[], provider, date
🚚

Logistics & Shipping

Bills of lading, customs declarations, packing lists. Extract origin, destination, cargo details, and weight — instantly.

→ shipper, consignee, cargo[], weight, hs_code

Have a document type that isn't listed here?

Built for developers

Production-ready from day one. No duct tape required.

⚡

Two-Stage AI Pipeline

Mistral OCR for text extraction, Claude for intelligent field mapping. 10x cheaper than pure vision models, faster too.

✅

Schema Validation

Zod-powered validation catches extraction errors before they hit your app. Confidence scores on every field so you know when to flag for review.

💾

OCR Caching

Identical documents skip OCR entirely on repeat extractions. Bulk processing the same batch daily? Pay once, cache forever.

🏷️

White-Label Portals

Create branded upload portals for your customers on Growth and Enterprise plans. Webhook delivery, email notifications, custom domain — all built-in.

How Dokyumi compares

Textract and Document AI are raw OCR engines. LlamaParse is for RAG pipelines. Dokyumi is the only one built specifically for structured data extraction with a schema you define.

FeatureDokyumiAWS TextractGoogle Doc AILlamaParse
No AWS/GCP account required✓✗✗✓
Custom extraction schema✓✗✗Partial
Dedicated API endpoint per schema✓✗✗✗
White-label upload portals✓✗✗✗
OCR result caching✓✗✗✓
Field confidence scores✓✓✓✗
Predictable flat-rate pricing✓✗✗✗
Free tier100/mo1K pages/mo1K pages/mo10K credits/mo

Comparison based on publicly available information as of March 2026. Pricing subject to change.

Built for the boring work that matters

Document parsing isn't glamorous. But bad data extraction kills products. Here's what teams use Dokyumi to solve.

“We were spending 40 hours a week manually entering invoice data. We needed something that gave us clean JSON, not a wall of OCR text we still had to parse ourselves.”

AP
Accounts Payable team
Mid-market logistics company

“We tried Textract first. The setup alone took two weeks and the output still needed post-processing to be usable. Dokyumi had us live in an afternoon.”

DS
Developer / Fintech startup
Bank statement processing

“The white-label portal feature is what got us. We could give clients a branded upload page and handle all the extraction behind the scenes without building anything custom.”

AG
Agency owner
Insurance document processing

Frequently asked questions

Everything you need to know before you start extracting.

What is Dokyumi?+
Dokyumi is a no-code document parsing API platform. You describe the fields you want to extract from your documents, and Dokyumi generates a dedicated API endpoint that handles OCR and structured data extraction automatically. It uses a two-stage AI pipeline: Mistral OCR for text extraction and Claude for intelligent field mapping.
How does Dokyumi compare to AWS Textract or Google Document AI?+
AWS Textract and Google Document AI are raw OCR engines — they return raw text or key-value pairs and require you to write significant post-processing code. Dokyumi is schema-first: you define exactly which fields you want (like vendor_name, invoice_total, due_date) and get clean, validated JSON back. No AWS or GCP account required. Setup takes under 2 minutes instead of weeks.
What document types does Dokyumi support?+
Dokyumi supports any document with a repeatable structure: invoices, bank statements, W-2s and 1099s, pay stubs, insurance claims (EOBs, declaration pages), medical records, legal contracts, leases, bills of lading, customs documents, and more. If the document has consistent fields, Dokyumi can extract them.
What file formats are supported?+
Dokyumi accepts PDF, JPEG, PNG, TIFF, and WEBP files up to 20MB. For best OCR results, documents should be at least 150 DPI. Both scanned documents and digital PDFs are supported.
How do I get started with Dokyumi?+
Sign up for free at dokyumi.com. You get 100 free extractions per month with no credit card required. Create your first extraction schema by describing your document in plain English (or use AI inference to auto-detect the schema), then use the generated API endpoint to start extracting data.
What is the pricing?+
Dokyumi offers a free tier with 100 extractions per month. Paid plans start at $79/month for the Starter plan (1,000 extractions, 10 schemas), $499/month for Growth (10,000 extractions, unlimited schemas, white-label portals), and $1,299/month for Enterprise with custom limits and SLAs. All paid plans include full API access, webhooks, and priority support.
What are white-label portals?+
White-label portals are branded upload interfaces you can give to your customers. Instead of exposing your API, you create a custom-branded page where clients upload documents. Dokyumi handles extraction in the background and delivers results via webhook or email. Available on Growth and Enterprise plans.
Does Dokyumi have an API I can call from my code?+
Yes. Dokyumi's core feature is its REST API. After creating a schema, you get a dedicated endpoint you can call with a POST request — file + schema slug. Returns structured JSON with extracted fields and per-field confidence scores. See the full API documentation at dokyumi.com/docs.

Your documents. Your schema. Your API.

Create your first extraction schema in under 2 minutes. 100 free extractions every month, no credit card required.