About Dokyumi
Dokyumi is a document extraction API that turns PDFs and scanned images into structured JSON shaped to the fields you define, for developers and operations teams who process invoices, bank statements, tax forms and other business documents.
What Dokyumi does
Custom extraction schemas
Describe the fields you need in plain English, build them in the schema editor, or upload one to three sample documents and let Dokyumi suggest the fields. You end up with a named schema that returns the same JSON shape every time, whatever the layout of the source document.
One extraction API
Send a PDF, JPEG, PNG, TIFF or WEBP file (up to 20MB) and a schema slug to POST /api/v1/extract with a bearer token. The response comes back in the same request, so any language that can make an HTTP call can use it.
Review signals on every result
Each result carries the extracted data, a model-reported confidence map, validation errors, and a completed or review status. Missing required values and fields below your confidence threshold go to review, so a person checks the uncertain ones before they reach your system.
White-label upload portals and webhooks
Give your own customers a branded page to upload documents instead of exposing your API. Submissions land in your dashboard and can be delivered to your system by webhook.
SIC classification for lenders
A dedicated classifier reads a business application and bank statements and returns a 2- and 4-digit SIC code with confidence, evidence and a needs-review flag. Lenders can try it on their own documents at dokyumi.com/sic without an account.
Workflow guides and starter schemas
Twenty-four process guides cover work like AP invoice intake, MCA bank statement intake, bills of lading, loss runs and rent rolls. Each one comes with a starter schema and the review steps that matter for that process, so a pilot starts from a working draft instead of a blank schema.
More detail: API documentation, workflow guides, SIC classifier.
What makes Dokyumi different
You name the fields; you do not parse a text dump
AWS Textract and Google Document AI are broad cloud document platforms that return text, tables and key-value data across many processor types. Dokyumi does one narrower job: it returns the fields in your schema, named the way your code expects them, in one response.
No cloud account to set up
Textract runs inside an AWS account and Document AI inside a Google Cloud project, with the IAM and billing setup that comes with each. Dokyumi needs a free signup and an API key; the first extraction can happen the same afternoon.
Flat plans instead of per-page metering
Textract and Document AI bill per page processed. Dokyumi is a flat subscription: $99/month buys 500 credits, one credit covers a document up to 5 pages, and when credits run out you get a clear quota error rather than an overage bill.
Review is part of the result contract
Every produced result exposes data, confidence and validation.errors plus a completed or review status. Dokyumi does not claim a universal accuracy number; it tells you which values to check, and confidence scores are model-reported, not calibrated probabilities.
Customer-facing portals on every plan
Branded upload portals come with every tier: 1 on Free, 5 on Starter, 25 on Growth and unlimited on Enterprise. The cloud OCR APIs above do not include a portal your customers can upload to; you would build that yourself.
Competitor descriptions are based on their public product and pricing pages, checked September 2026. If you only need one fixed form type parsed, a single-document parser may fit better; see which tool you need.
Who uses Dokyumi
- Software teams adding document upload and data extraction to their own product
- Accounts payable and finance operations teams processing vendor invoices
- Small-business lenders and MCA underwriters reading applications and bank statements
- Insurance operations teams handling claims and declaration pages
- Tax preparers and accounting firms working with W-2s, 1099s and returns
- Property managers and real estate teams extracting lease terms
- Healthcare billing teams working with EOBs and medical bills
Industry guides: use cases.
The team behind Dokyumi
Dokyumi was founded by John Arndt, who also runs Soxoa, a Sacramento, California AI consulting and software company. His background spans more than a decade across AI, cybersecurity, fraud and risk, privacy, payments and enterprise software.
John started Dokyumi in February 2026, alongside a set of single-form parsers for documents like W-2s, pay stubs and invoices. A fixed template works for one form; it does not help when you need a second document type, fields the template does not cover, or a way for your customers to upload files themselves. Dokyumi is the general version: the same OCR and extraction approach, with the fields coming from your schema.
Dokyumi is a solo-founder company. John designs, builds and operates the product himself, using AI-assisted development tools, and there is no separate support team: when you email, John reads it.
Find John on LinkedIn, at johnarndt.net, GitHub and soxoa.com/john-arndt.
How Dokyumi works
Who you work with. John Arndt, directly, for product questions, schema help, Enterprise scoping and data-deletion requests.
How to reach us. Email hello@dokyumi.com. Expect a reply within one business day. Integration details live in the API documentation.
Turnaround. Signup is self-serve and takes minutes. API extractions return in the same request, typically in seconds for short documents. Enterprise quotes follow a review of your sample documents and expected volume.
Getting started
- Sign up free, no card required. You get 25 credits a month to test on your own files.
- Create a schema: describe your fields, build them in the editor, or upload one to three samples for suggested fields.
- Run representative documents through the dashboard and check results against the originals.
- Create an API key and call POST /api/v1/extract, or share a white-label upload portal with your customers.
- Upgrade from dashboard settings when you need more credits, schemas or portals. Enterprise starts with a conversation and your sample documents.
Key facts
| Company Name | Dokyumi (operated by Soxoa LLC) |
|---|---|
| Type | Document extraction software (SaaS and API) |
| Founded | 2026 |
| Founder | John Arndt |
| Headquarters | Sacramento, California, USA |
| Website | dokyumi.com |
| Core Offering | Schema-defined extraction API that returns structured JSON with confidence and validation details |
| Pricing | Free (25 credits/mo); Starter $99/mo or $990/yr (500 credits); Growth $499/mo or $4,990/yr (3,000 credits); Enterprise quoted. 1 credit = 1 document up to 5 pages; 50-page limit per document on self-serve plans. |
| Contract Terms | Monthly or annual subscription billed in advance; change plan or interval any time; no overage billing; Enterprise by contract |
| Services | Custom schemas, AI schema suggestions, extraction API, review signals, white-label upload portals, webhooks, SIC classification |
| Communication | Email (hello@dokyumi.com), reply within 1 business day; API documentation at dokyumi.com/docs |
| Notable Clients | None publicly named |
| Customers Served | Early-stage; not publicly disclosed |
| Projects Delivered | Not publicly disclosed |
| Competitors | AWS Textract, Google Document AI, LlamaParse |
| Social | LinkedInjohnarndt.netGitHubJohn Arndt at Soxoa |
Frequently asked questions
Who is behind Dokyumi?
Dokyumi was founded in 2026 by John Arndt and is operated by Soxoa LLC, an AI consulting and software company in Sacramento, California. John builds and runs the product himself with AI-assisted tooling, and he answers support email directly.
How much does Dokyumi cost?
Free includes 25 credits a month. Starter is $99/month (500 credits) and Growth is $499/month (3,000 credits), with annual billing at $990 and $4,990 a year. Enterprise is quoted for higher volume or documents over 50 pages.
Do I need an AWS or Google Cloud account?
No. You sign up at dokyumi.com, create an API key, and call one HTTPS endpoint. There is no cloud project, IAM role or storage bucket to configure.
Am I locked into a contract?
Self-serve plans are monthly or annual subscriptions billed in advance, and you can upgrade, downgrade or switch billing interval at any time. Charges for completed extractions are not refunded. Enterprise terms are set per contract.
What happens to the documents I upload?
Documents are processed when you submit them and stored in encrypted Cloudflare R2 storage, and OCR results are cached by file hash. Self-service deletion is not available yet; email hello@dokyumi.com to request deletion.
Does Dokyumi verify that a document is genuine?
No. Dokyumi extracts what the document says; it does not check authenticity or confirm that the claims in it are true. Route review results and anything consequential to a person before you act on it.
How fast will I hear back from support?
Email hello@dokyumi.com and expect a reply from John within one business day. Enterprise support terms are set in the contract.
Try it on your own documents: start free or review plans and limits. See the security page and privacy policy for data handling.