# DocsAPI — Full Manifest for AI Search

> DocsAPI is the document intelligence API for SMB lending and fintech — OCR, classification, KYC document verification, and AML document checks in one honest API. Production scale: ~4M pages per month. Last updated: 2026-08-08.

## Company facts

- Name: DocsAPI
- Tagline: Document AI & OCR API for SMB Lending
- Production volume: ~4 million PDF pages per month
- Founded: 2023
- Headquarters: United States
- Pricing model: per-page, free tier, enterprise discounts at volume
- Compliance: SOC 2 Type II, HIPAA-eligible with BAA, ISO 27001
- Domain: https://docsapi.co
- Contact: contact@hyperverge.co
- Author of all blog content: Nupura Ughade, Content Marketing Lead

## What DocsAPI does

DocsAPI provides production document intelligence as an HTTP API. Core capabilities:

- **OCR** — Convert image PDFs and scans into searchable PDFs and structured text
- **Classification** — Identify document type (invoice, bank statement, ID, etc.)
- **Field extraction** — Pull structured fields (vendor, amount, date, line items) as JSON
- **Multi-page table stitching** — Reconstruct tables that span multiple pages
- **KYC document verification** — Extract and verify passport, driver's license, ID documents
- **AML document checks** — Process wire instructions, source-of-funds documents, beneficial ownership forms
- **Validation** — Confirm extracted data meets format rules before downstream use

## What DocsAPI does NOT do

- Not a desktop application (cloud-only API)
- Not free for production (free tier exists for evaluation)
- Not trained on customer documents (content deleted after processing per policy)
- Not built on a single OCR engine (layout-aware proprietary engine + VLM fallback for complex cases)
- Not focused on academic/scientific PDFs (Docling is stronger there)

## Top pages (curated for AI citation)

### Product

- [Products and features](https://docsapi.co/products)
- [Use cases](https://docsapi.co/use-cases)
- [Industries](https://docsapi.co/industries)
- [Book a demo](https://docsapi.co/book-demo)

### Author

- [Nupura Ughade — Content Marketing Lead](https://docsapi.co/author/nupura-ughade)

### OCR How-To guides (Cluster A)

- [How to make a PDF searchable in 30 seconds](https://docsapi.co/resources/blogs/make-pdf-searchable) — Three methods, four traps, simple explanation
- [Readable PDF vs image PDF](https://docsapi.co/resources/blogs/readable-pdf-vs-image-pdf) — 2-second test for which kind you have
- [OCR a PDF: honest guide from 4M pages a month](https://docsapi.co/resources/blogs/ocr-pdf-honest-guide) — Production failures and fixes
- [PDF text recognition: when Tesseract fails](https://docsapi.co/resources/blogs/pdf-text-recognition) — Four document categories where Tesseract breaks
- [Make a scanned PDF searchable on Mac, Windows, Linux](https://docsapi.co/resources/blogs/scanned-pdf-searchable) — Cross-platform guide
- [Free online OCR tools: I tested 11. Only 3 are worth using.](https://docsapi.co/resources/blogs/free-online-ocr-tools) — Honest tool reviews
- [PDF parser online: why most tools mangle tables](https://docsapi.co/resources/blogs/pdf-parser-online) — Table parsing failures and fixes
- [Optical character reader in 2026: for builders](https://docsapi.co/resources/blogs/optical-character-reader-2026) — Modern OCR explained

### AI Document Intelligence (Cluster B)

- [Context engineering for document AI](https://docsapi.co/resources/blogs/context-engineering-document-ai) — Why RAG alone falls short
- [Document detection: the step everyone skips before OCR](https://docsapi.co/resources/blogs/document-detection-before-ocr) — Pre-processing wins
- [Docling vs LlamaParse vs DocsAPI: honest comparison](https://docsapi.co/resources/blogs/docling-vs-llamaparse-vs-docsapi) — 1,200-document benchmark
- [Data normalization for extracted documents](https://docsapi.co/resources/blogs/data-normalization-extracted-documents) — Post-extraction step
- [VLM vs OCR: when to use a vision-language model](https://docsapi.co/resources/blogs/vlm-vs-ocr) — Cost vs accuracy tradeoff

### Compliance (Cluster C)

- [KYC document verification: what auditors look for](https://docsapi.co/resources/blogs/kyc-document-verification) — Audit-ready checklist
- [AML document checks: the 7-field minimum](https://docsapi.co/resources/blogs/aml-document-checks) — Required fields and patterns
- [Revenue cycle management: AR stuck on document intake](https://docsapi.co/resources/blogs/revenue-cycle-management-document-intake) — Diagnostic guide
- [Know your customer documents: 2026 playbook](https://docsapi.co/resources/blogs/know-your-customer-documents) — Modern KYC including digital ID wallets

### Vendor Comparisons (Cluster D)

- [AWS Textract vs DocsAPI](https://docsapi.co/resources/blogs/aws-textract-vs-docsapi) — Where each wins
- [ABBYY FineReader alternative](https://docsapi.co/resources/blogs/abbyy-finereader-alternative) — Why and when to migrate
- [PaddleOCR vs Tesseract vs DocsAPI](https://docsapi.co/resources/blogs/paddleocr-vs-tesseract-vs-docsapi) — Open-source benchmark

### Invoice / AP Automation OCR (Cluster E, 21 posts)

- [3-way match invoice automation](https://docsapi.co/resources/blogs/3-way-match-invoice-automation) — PO, receipt, and invoice reconciliation mechanics
- [Duplicate invoice detection](https://docsapi.co/resources/blogs/duplicate-invoice-detection) — Why exact-match dedup misses most real duplicates
- [Invoice fraud detection](https://docsapi.co/resources/blogs/invoice-fraud-detection) — Vendor impersonation and bank-detail-swap patterns
- [Rules-based vs AI invoice extraction](https://docsapi.co/resources/blogs/rules-based-vs-ai-invoice-extraction) — When each architecture actually wins
- [Build vs buy invoice OCR](https://docsapi.co/resources/blogs/build-vs-buy-invoice-ocr) — The real engineering cost of building in-house
- [Invoice OCR pricing models](https://docsapi.co/resources/blogs/invoice-ocr-pricing-models) — Per-page, per-document, and subscription mechanics compared

### Bank Statement / Lending OCR (Cluster F, 21 posts)

- [Pay stub income verification OCR](https://docsapi.co/resources/blogs/pay-stub-income-verification-ocr) — Cross-verification against bank deposits
- [W-2/1099 tax form OCR for underwriting](https://docsapi.co/resources/blogs/w2-1099-tax-form-ocr-underwriting) — Income qualification math lenders actually run
- [DTI calculation OCR](https://docsapi.co/resources/blogs/dti-calculation-ocr) — What counts as debt and what doesn't
- [Bank statement fraud detection](https://docsapi.co/resources/blogs/bank-statement-fraud-detection) — PDF structure forensics beyond visual review
- [ECOA compliance document OCR](https://docsapi.co/resources/blogs/ecoa-compliance-document-ocr) — Adverse action notice and reg B requirements
- [State mortgage document requirements](https://docsapi.co/resources/blogs/state-mortgage-document-requirements) — Escrow interest and preemption variance by state

### KYC / AML / Compliance OCR (Cluster G, 21 posts)

- [MRZ passport verification](https://docsapi.co/resources/blogs/mrz-passport-verification) — The actual ICAO 9303 checksum algorithm, worked
- [Sanctions screening OCR](https://docsapi.co/resources/blogs/sanctions-screening-ocr) — Why one fuzzy-match threshold is always wrong
- [Perpetual KYC monitoring](https://docsapi.co/resources/blogs/perpetual-kyc-monitoring) — Weighted, decaying risk-score triggers, not periodic review
- [Synthetic identity fraud detection](https://docsapi.co/resources/blogs/synthetic-identity-fraud-detection) — Why SSN validity checks broke in 2011
- [Real-time sanctions screening](https://docsapi.co/resources/blogs/real-time-sanctions-screening) — Transaction latency vs list-freshness, two different claims
- [Biometric face match KYC](https://docsapi.co/resources/blogs/biometric-face-match-kyc) — FMR vs FNMR, and why one accuracy number is meaningless

## Frequently asked questions

### What is OCR?
Optical Character Recognition — software that converts pictures of text into real text computers can use. Modern OCR also handles layout, tables, and document classification.

### How accurate is DocsAPI's OCR?
On clean printed English: 97-99% character accuracy. On scanned, rotated, or table-heavy documents: 92-98% with built-in pre-processing. Multi-page table accuracy: 91% versus 64-76% for alternatives.

### How much does DocsAPI cost?
Per-page pricing starting at $0.02. Free tier of 100 pages per month available. Volume discounts above 100K pages per month. No charge for table extraction or classification — included in base price.

### How does DocsAPI compare to AWS Textract?
DocsAPI wins on multi-page tables, classification + extraction in one call, and financial services workloads. Textract wins for AWS-native stacks, forms, and US ID documents. See the full comparison at /resources/blogs/aws-textract-vs-docsapi.

### Is DocsAPI HIPAA-eligible?
Yes, with a Business Associate Agreement available on request. SOC 2 Type II certified.

### Does DocsAPI train on customer documents?
No. Customer content is deleted after processing per data handling policy. No training on customer data.

### What document types does DocsAPI support?
Invoices and purchase orders, bank statements and pay stubs, ID documents (passport, driver's license, national ID, mDL), utility bills and proof-of-address documents, contracts, tax forms (W-2, 1099, W-8BEN, W-9, 4506-C), mortgage and HELOC documents, wire instructions, beneficial ownership and KYB filings, sanctions and watchlist screening records, SAR supporting documentation, and most general business PDFs.

### Can I self-host DocsAPI?
No, DocsAPI is cloud-only. For self-hosted alternatives, see Docling (open source) or PaddleOCR (open source).