🧠 HARU — The Intelligence Assistant

A fully local, privacy-first desktop studying assistant that makes you smarter, faster, and more capable. Built with Tauri + SolidJS + Rust + Multiple Local AI Models.

---

🚀 Tech Stack

Frontend

Backend

AI / ML

---

⚡ Overview

HARU is a local AI assistant powered by offline pipelines.

Core functionalities include:

• Chats & Voice Conversations
• Dynamic Courses and Material Syllabuses
• Resources Search and AI Mindmaps
• Flashcards & Learning Tools
• Time Tracking & Productivity Analytics
• Notes & Whiteboard Workspace
• Videos & Media Management
• Goals & Milestones Tracking
• RAG & Web Search
• File Tools & Utilities
• Plugins & Brain-training Activities

Everything runs 100% locally — zero cloud, zero tracking, zero limits.

---

🖼️ Screenshots

Achievements System	Brain Activity Visualization	Smart Calculator Tool
AI Chat Interface	Course Viewer	Definition Explorer
Document Manager	Flashcards Analytics	Flashcards Study Mode
Goals & Milestones	Library View	New Course Wizard
Notes System	Plugins Manager	Pomodoro Focus Tool
Personal Progress Overview	Quick Notes Popup	Advanced Settings
Appearance Settings	Behavior Customization	Downloads Settings
Local Paths Configuration	Course Syllabus Page	Time Tracking Dashboard
Utilities Panel	File Transcription Tool	Folder Transcription
Translation Tool	Typing Trainer	Videos Library
Voice Assistant	Whiteboard Workspace

---

🧠 Core Algorithms

Feature	Algorithm	Description
Flashcards	Adaptive Interval SRS	Adjusts difficulty automatically
RAG	Multi-pass Vector Retrieval	High-accuracy context windows
Time Tracking	Attention Metrics	Measures focus & app usage
Notes	Semantic Graphing	Topic clustering
Goals	Milestone Dependency Graph	Auto-generated progress path
Voice	Streamed Whisper.cpp	Realtime local transcription
Typing	Stabilized WPM Scoring	Filters noise & errors
Tools	Multi-pipeline Workers	OCR, translation, embeddings

---

💽 Installation Guide

1. Install Dependencies

pnpm install

2. Run the App

pnpm tauri dev

---

🧩 Backend Workers (Using haru.bat)

HARU uses multiple FastAPI microservices.

Commands

haru.bat chat
haru.bat rag
haru.bat web
haru.bat tts
haru.bat stt
haru.bat misc
haru.bat home
haru.bat voice

What each one does

Command	Description	Port
chat	Main LLM chat logic	5000
rag	Document search, embeddings	5001
web	Web utilities (summaries, scrapers)	5002
tts	Text-to-speech worker	5003
stt	Whisper speech-to-text	5004
misc	OCR, file tools, misc features	3999
home	Dashboard counters	4999
voice	Voice conversation pipeline	5005

Recommended setup

haru.bat chat
haru.bat rag
haru.bat stt
haru.bat tts
pnpm tauri dev

---

🤖 Local LLM Server (Using llm.bat)

Run the multimodal model (Gemma Vision)

llm.bat v

Run Gemma chat model

llm.bat g

Run Qwen Thinking

llm.bat t

Default (Gemma 3 4B IT)

llm.bat

Runs llama-server.exe with:

• GPU acceleration (-ngl 99)
• Large context windows
• Multimodal mmproj loaded correctly

🛡️ License & Usage

© 2025 Mohaned. All rights reserved.
Unauthorized use, reproduction, modification, or distribution of this project or its source code is strictly prohibited.

⚠️ IMPORTANT: This project is beta, private, and fully restricted.