ninjadoc.ai

Document extraction your agents can cite.

Get structured PDF answers with bounding boxes, citation URLs, and MCP-ready outputs for auditable agent workflows.

Start Building

Document AI APIs compared for auditability

Most tools return extracted text. Ninjadoc returns answers with the source evidence an agent or reviewer needs to verify them.

ninjadoc.ai
Traditional OCR
Generic LLM API
1Natural language document questions
2Bounding boxes for every answer
3Citation URLs for source review
4MCP-ready for AI agents
5Audit-ready evidence JSON
6Transparent per-page pricing
7No templates or field training
8Developer-friendly REST API

Pay as you go.
No subscriptions.

250 credits per dollar. Credits never expire. Start with what you need, scale as you grow.

$5

1,250 credits

Perfect for trying out the API

625 pages of processing
Access to queries, and future APIs
Get Started
Most Popular
$25

6,250 credits

Great for regular document processing

3,125 pages of processing
Access to queries, and future APIs
Get Started
$100

25,000 credits

For high-volume processing needs

12,500 pages of processing
Access to queries, and future APIs
Get Started
$500

125,000 credits

Maximum value for large-scale operations

62,500 pages of processing
Access to queries, and future APIs
Get Started

Per-Page Pricing

Simple, predictable costs for every operation

Document Query

Ask natural language questions about any document

2credits/page

$0.01 per page

Agent-Ready

Your agents extract.
We prove it.

When AI makes a decision based on a document, you need more than an answer. You need proof. Ninjadoc returns exact source coordinates for every field, making every agent-driven extraction auditable by default.

Tool in your agent's pipeline

Call Ninjadoc from any agent framework. Structured JSON answers with citation URLs that travel with your data into every downstream system.

Every extraction is provable

Every field comes with page number and bounding box coordinates pointing to the exact location it was drawn from: text, image, checkbox, or signature.

Audit trail by default

Every extracted field comes with a citation URL linking directly to its source location in the document. When an auditor asks where the number came from, you have a link — not just a claim.

// Every answer includes a citation URL that links directly to the source in the document
{
  "index": 0,
  "question": "What is the total amount due?",
  "answer": "$4,250.00",
  "confidence": 0.99,
  "evidence": [{
    "page_index": 0,
    "evidence_texts": ["Total Amount Due: $4,250.00"],
    "citation_url": "https://api.ninjadoc.ai/citations/01JV.../0/page.jpg",
    "citation_url_cropped": "https://api.ninjadoc.ai/citations/01JV.../0/cropped.jpg",
    "located": [{
      "boxes": [{
        "normalized": { "x1": 680, "y1": 820, "x2": 890, "y2": 850 },
        "pixel":      { "x1": 420, "y1": 680, "x2": 540, "y2": 700 }
      }]
    }]
  }]
}
MCP Server· works with Claude, Cursor, and any MCP-compatible agent
{
  "mcpServers": {
    "ninjadoc": {
      "serverUrl": "https://api.ninjadoc.ai/mcp/nj_API_KEY"
    }
  }
}
Developer First

Integrated in minutes,
not months.

Our API is designed for developer happiness. RESTful, fully documented, and ready for production. Integrate with any language or framework in just a few lines of code.

Simple REST API
Language Agnostic
Comprehensive Documentation
24/7 Developer Support