Overview & Presets

Templates are the core building blocks of Ocriva. A template defines exactly how the AI reads your documents and what data it should return. This guide walks you through every aspect of template creation and configuration.

---

What is a Template?

A Template is a reusable blueprint that instructs the AI on:

• What to extract — the specific fields or information to pull from a document
• How to extract it — which AI model to use and what instructions to follow
• What to return — the output format (JSON, CSV, PDF, DOCX, Text, XML, HTML, or Image)

Every document you process is run through a template. This makes extraction consistent and repeatable across hundreds or thousands of documents.

Template Components

Component	Description
Name & Description	Human-readable label and explanation
Project	The project this template belongs to
Schema	JSON Schema defining the fields to extract
Assistant Config	AI provider, model, instructions, temperature
Result Format	Output format and format-specific options
Data Files	Reference documents for RAG (Retrieval-Augmented Generation)

Template Lifecycle

A template becomes active automatically once you have configured both a model and instructions in the Assistant Config. Until then, it stays in draft state and cannot be used for processing.

---

Creating a Template

Step 1 — Navigate to Templates

Go to your Project dashboard and select the Templates tab in the left navigation. Click the Create Template button in the top-right corner.

Step 2 — Name and Description

Fill in the basic information:

• Name (required) — Use a descriptive name that identifies what documents this template processes. Examples:
  • Thai Tax Invoice 2024
  • Employee Expense Receipt
  • Product Specification Sheet
• Description (optional but recommended) — Explain the purpose and the type of documents it handles. This helps team members pick the correct template when uploading.

Step 3 — Project Assignment

Templates always belong to a specific project. The project context is set automatically when you create a template from within a project. If you have access to multiple projects, you can reassign the template from the template settings.

Step 4 — Choose a Result Format

This is the first configuration decision. The result format determines what form the AI returns extracted data in and which configuration options are available.

Format	Best For	Field Definition Required
JSON	API integrations, structured data, databases	JSON Schema (19 presets available)
CSV	Spreadsheets, Excel, data analysis	Column names
Text	Plain text summaries, simple extraction	Not required
PDF	Reports, printable documents	Not required
DOCX	Word documents, editing, sharing	Not required
XML	Legacy systems, data exchange	Not required
HTML	Web display, email content	Not required
Image	AI image generation from documents	Image options

Step 5 — Use a Preset (JSON only, optional)

If you selected JSON, you can optionally pick from 19 preset template examples that come with pre-configured extraction fields for common document types. See the Preset Templates section below for the full list of presets.

If no preset fits your use case, you can define a custom schema from scratch.

Step 6 — Define Fields (JSON and CSV)

• JSON: Define a JSON Schema with field names, types, and descriptions.
• CSV: Define column names and ordering.
• Text / PDF / DOCX / XML / HTML: No field definition needed — the AI returns free-text output.
• Image: Configure image generation options.

Step 7 — Set Up the AI Model

Choose an AI provider and model, write the extraction instructions, and set any advanced parameters.

Step 8 — Save and Test

Click Save. The template is now active. Use the Test button to run it against a sample document and verify the output before processing real files.

---

Preset Templates (JSON Quick Start)

When you select JSON as your result format, Ocriva offers 19 preset template examples with pre-configured extraction fields. These presets let you get started quickly without manually defining every field.

How to Use Presets

1. Create a new template and select JSON as the result format
2. Click Use Preset or Template Examples
3. Browse the available presets by category
4. Select a preset — its fields and schema will be auto-filled
5. Customize the fields as needed (add, remove, or modify)
6. Save your template

Tip: Presets are starting points — you can always add, remove, or modify fields after selecting a preset.

Available Presets

Category	Preset	Description
Financial	Invoice	Invoice number, vendor, customer, line items, amounts, tax
Financial	Receipt	Store name, items, total, payment method, date
Financial	Quotation	Quote number, items, pricing, validity, terms
Financial	Bank Statement	Account info, transactions, balance, period
Legal	Contract	Parties, terms, effective dates, signatures
Legal	Power of Attorney	Grantor, agent, powers granted, scope, expiration
Legal	Company Certificate	Company name, registration number, directors, address
Government	Identity Document	Name, ID number, date of birth, address, photo
Government	Civil Registration	Registration type, persons involved, dates, location
HR	Job Application	Applicant info, education, work experience, skills
Education	Exam Document	Student name, exam details, scores, grades
Education	Transcript	Student info, courses, grades, GPA, credits
Education	Academic Publication	Title, authors, abstract, keywords, references
Medical	Medical Certificate	Patient info, diagnosis, doctor, hospital, dates
Medical	Medical Records	Patient history, vitals, diagnosis, treatment
Medical	Prescription	Patient, medications, dosage, doctor, pharmacy
Handwriting	Handwritten Form	Form fields extracted from handwritten documents
Retail	Food Menu / Product Label	Items, prices, ingredients, nutrition, allergens
Logistics	Package Tracking	Tracking number, sender, receiver, status, dates