Library Overview

The 9thSense Library is a curated set of 40+ pre-built document types that ship with every workspace. Each type encodes years of domain knowledge: which fields to extract, how to validate them, which model to invoke, and how to auto-detect the document from a raw file.

You pass a context_id to the extraction API and get back a structured JSON object. No prompt engineering. No schema design.

---

Categories at a glance

---

How each type works

Every library document type has four components:

context_id — The API identifier you pass when requesting extraction. Example: kyc_aadhaar, bank_statement, gst_certificate.

output_schema — The exact fields the model extracts. Each field has a name, type, and description. For example, Aadhaar extracts aadhaar_number, name, date_of_birth, address, gender, photo_present, and more.

validation_rules — Deterministic checks applied after extraction — the same rule family that powers Goal Rules. Rules include:

• required — field must be present and non-null
• regex — value must match a pattern (e.g. PAN must match ^[A-Z]{5}[0-9]{4}[A-Z]$)
• format — value must conform to a format like date_iso or uppercase
• mask — field is masked in logs and audit output (applies to Aadhaar number, account numbers, PAN)

Analysis type — Most types perform structured extraction (printed/handwritten text, photos, stamps, tables, mixed layouts). The Classification category instead runs fraud checks — deepfake detection, liveness, and synthetic document detection — returning a verdict + confidence rather than extracted fields.

---

Using a library type

Pass context_id to the extraction endpoint from any SDK or directly via the API.

import { NinthSense } from "@9thsense/sdk";

const client = new NinthSense({ apiKey: process.env.NINTHSENSE_API_KEY });

const result = await client.verify.document({
  doc_type: "kyc_aadhaar",
  file_url: "https://storage.example.com/docs/aadhaar.jpg",
});

console.log(result.output);
// {
//   aadhaar_number: "XXXX XXXX 5678",
//   name: "Priya Sharma",
//   date_of_birth: "1990-04-15",
//   gender: "F",
//   address: "123 MG Road, Bengaluru, Karnataka 560001",
//   photo_present: true,
//   qr_code_present: true
// }

Or via the REST API directly:

curl -X POST https://api.9thsense.ai/v1/verify \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "context_id": "kyc_aadhaar",
    "file_url": "https://storage.example.com/docs/aadhaar.jpg"
  }'

---

Creating custom document types

If the library does not cover a document type you need, you can create your own.

Option 1 — Define it manually:

POST /v1/document-types
{
  "context_id": "employee_id_card",
  "display_name": "Employee ID Card",
  "output_schema": {
    "employee_name": "string",
    "employee_id": "string",
    "department": "string or null",
    "valid_until": "YYYY-MM-DD or null"
  },
  "validation_rules": {
    "employee_name": { "required": true },
    "employee_id": { "required": true }
  }
}

Option 2 — Generate from samples: Upload two or more example documents to the discover endpoint and let 9thSense infer the schema automatically:

POST /v1/document-types/discover
{ "sample_urls": ["https://...", "https://..."] }

Option 3 — Generate from description:

POST /v1/document-types/generate
{ "description": "A car insurance policy page showing policy number, premium amount, insured vehicle, and validity period" }

Custom types follow the same API surface as library types — same extraction call, same validation rules.

---

Explore by category

• Identity Documents — Aadhaar, PAN, Passport, Driving License, Voter ID, Ration Card, Vehicle RC
• Financial Documents — Bank Statement, Cheque, Salary Slip, Form 16, ITR, Invoice, and more
• Business Documents — GST, CIN, UDYAM, FSSAI, Drug License, Board Resolution, and more
• Travel Documents — Flight Ticket, Hotel Booking, Visa
• Scene Understanding — Merchant verification, site safety, vehicle inspection
• Classification — Deepfake, synthetic document, and liveness checks
• Other — Marksheet, University Offer Letter, Election Affidavit