Home

ONCard Wiki

This Wiki will have detailed data about the app. Browse within the pages to know more.

You may refer to:

Setup Guide • Performance • Accounts • AI Models

Technical pages:

How FTC Works | How Cards Work | How MCQ Marking Works

What Is ONCard?

Table of Contents

• Overview
• Who is it for?
• The Core Idea
• Key Features
  • One-Input Card Generation
  • Files to Cards
  • Virtual Teacher with Real-Time Grading
  • Follow-Up Teaching Chat
  • Ask AI
  • Subject-Path Organization
  • Adaptive Learning (NNA System)
  • MCQ Generation
  • Multiple Student Profiles
• How It Works Under the Hood
• What Makes ONCard Different

Overview

ONCard (Open-source Neural-Accelerated Cards) is a fully offline, AI-powered study app designed to help students learn faster — without giving up privacy or paying for subscriptions. It combines a flashcard-style study system with a local AI "virtual teacher" that can create study content from your notes, grade your answers in real time, and explain concepts you're struggling with — all running on your own machine through Ollama.

Who is it for?

ONCard is built for students of any level who want to:

• Turn their notes, slides, or textbook material into structured study cards.
• Practice with instant, AI-graded feedback and deeper explanations.
• Build a growing, organized library of study material sorted by subject and topic.

The Core Idea

Most AI study tools live in the cloud — your data leaves your device, you hit rate limits, and you often pay a monthly fee. ONCard flips that around. Everything runs locally on your hardware. There are no subscriptions, no hidden tiers, no artificial usage limits, and no cloud dependency. Once the initial AI models are downloaded, you can study completely offline.

Key Features

One-Input Card Generation

Type a single question or topic from your notes and ONCard fills in the rest — title, hints, a detailed answer, difficulty rating, subject category, and subtopic. The AI does the heavy lifting so you can focus on studying instead of setup.

Files to Cards

Drop in PDFs, PowerPoint files, or images and ONCard extracts study questions directly from the visual content. Great for turning lecture slides or textbook pages into practice material without typing anything manually.

Virtual Teacher with Real-Time Grading

Write your answer to a card question and get it graded on a 0–10 scale. You receive structured feedback: what you got right, what went wrong, and what to improve. It works like having a teacher sitting next to you.

Follow-Up Teaching Chat

After your answer is graded, you can ask follow-up questions directly in the app. The AI keeps the context of your card and grading session, so explanations are relevant to what you just studied.

Ask AI

Type /ai or #ai in the search bar to open a free-form AI chat. The assistant can answer general questions, explain concepts, and search across your existing card library using semantic (meaning-based) search. You can customize the AI's tone — warm, funny, sarcastic, Shakespearean, and more.

Subject-Path Organization

Cards are sorted into a hierarchical subject tree (Mathematics → Algebra → Equations, Computer Science → Languages → Python, etc.). A topic sidebar lets you filter and browse by area, so large subjects stay manageable.

Adaptive Learning (NNA System)

ONCard doesn't just show you random cards. It tracks your performance, flags topics you're struggling with, and adjusts your study session accordingly. If you repeatedly fail a topic, it generates targeted reinforcement questions to help you close the gap. Over time, the app gets smarter about your weak areas and surfaces related practice material automatically.

MCQ Generation

Bulk-generate multiple-choice questions from your cards for quick practice sessions.

Multiple Student Profiles

Each profile is separate with its own card library, study history, and settings. Profiles capture basic info like grade and hobbies so the AI can tailor its language and examples to you.

Wikipedia Search

Students can search Wikipedia by typing /wiki in the search bar and typing the subject needed to search. eg: >> Photosynthesis not: >> What is Photosynthesis?

This might work, but it tends to be unstable and might not be able to be searched.

How It Works Under the Hood

ONCard runs on Ollama, a local AI engine you install alongside the app. By default, it uses:

• Gemma4:e2b (Google) — for generating cards, grading, chat, and explanations.
• Nomic Embed Text v2 MoE (Nomic AI) — for semantic search, topic clustering, and adaptive learning features.

Optional larger models (qwen3.5/other gemma4 models) can be installed for better reasoning quality or tool-calling features. There's also an optional cloud mode if your hardware can't handle local inference.

All your data — cards, study history, profiles — is stored in local SQLite databases on your machine. Nothing leaves your device unless you explicitly enable cloud mode.

What Makes ONCard Different

	Traditional AI Study Apps	ONCard
AI inference	Cloud-based (your data is sent to servers)	Fully local via Ollama
Cost	Subscription or pay-per-use	Free and open-source (MIT license)
Offline use	No (or very limited)	Yes, entirely
Privacy	Data stored on company servers	Everything stays on your machine
Usage limits	Rate limits, token caps	No artificial limits
Adaptive learning	Basic spaced repetition	Semantic weakness detection + targeted reinforcement
Open source	Rarely	Yes, fully

To provide a refined and personalized learning experience, ONCard redefines the principles of recommendation systems. At its core is NNA — a unified intelligence layer that powers and seamlessly integrates all ONCard features.