Skip to content

Repository files navigation

🧠 Jarvis Personal AI OS

An autonomous, privacy-first personal AI operating system for unified signal ingestion, cognitive signal understanding, multi-agent intelligence routing, and automated life management.


🌟 Overview

Jarvis Personal AI OS is a multi-agent artificial intelligence framework designed to process ambient real-world signals—ranging from financial transactions, bank SMS notifications, and GPay receipts, to action items, notes, and informational alerts.

By combining deterministic fast-path metadata extraction with local LLM cognitive capabilities (qwen2.5:1.5b), Jarvis transforms raw multi-source streams into structured, domain-specific ledgers and daily executive summaries.


📱 Mobile Collector Integration

Jarvis pairs natively with the mobile collection background app:

  • Android App Repository: 📲 pradeepgithubrepo/jarviscollector
  • Description: Runs as a lightweight Android service intercepting financial SMS messages and transaction notifications on-device, securely transmitting JSON signal envelopes to Supabase storage buckets for automated ingestion.

🗺️ High-Level System Architecture

flowchart LR
    subgraph Ingestion ["📱 1. Ingestion"]
        Mobile["Android Collector\n(jarviscollector)"] --> Bucket["Supabase Storage"]
        Statements["Bank PDFs / CSVs"] --> Bucket
        Bucket --> Consumer["Consumer Agent\n(Deduplication & Qualification)"]
    end

    subgraph Intelligence ["🧠 2. Cognitive Layer"]
        Consumer --> SUA["Signal Understanding Agent\n(SUA)"]
        SUA --> Router["Intelligence Router\n& Dispatcher"]
    end

    subgraph Agents ["🤖 3. Specialist Agents"]
        Router --> Fin["💳 Financial Agent"]
        Router --> Todo["📝 Todo Agent"]
        Router --> FYI["ℹ️ FYI Agent"]
        Router --> Brief["🗞️ Daily Briefing Agent"]
    end

    subgraph Ledgers ["💾 4. Persistence"]
        Fin --> L1[("financial_transactions")]
        Todo --> L2[("todos")]
        FYI --> L3[("fyi_entries")]
        Brief --> L4[("daily_briefings")]
    end
Loading

📚 Core Architecture Documentation Suite

Explore the detailed architecture, pipelines, and agent specifications under docs/architecture/:

Document Description Key Focus Areas
📥 01. Consumption Pipeline Data ingestion, file hashing, deduplication & qualification. jarviscollector, SHA-256 dedup, qualification scoring rules.
🧠 02. SUA & Signal Classification Signal Understanding & Analysis cognitive engine. Metadata bypass, local LLM parsing, unified contracts, routing.
🤖 03. Agent Ecosystem Visual guide & breakdown of all 8 Jarvis AI agents. Consumer, SUA, Financial, Financial Summary, Todo, FYI, Briefing, Lifecycle.
📐 04. End-to-End System Architecture Topology, database schemas & system lifecycle. Database ER diagrams, replay framework, backfill processing.

📑 Historical & Technical Documentation Index (docs/v2 & docs/v2.1)

The repository contains extensive technical deep-dives, migration blueprints, validation reports, and SQL schemas built during system development:

🔍 Expand Comprehensive Technical Index (20+ Specifications & Reports)

📥 Phase 1: Ingestion & Normalization (docs/v2/phase1a-c)

🎯 Qualification Layer (docs/v2/)

🧠 Phase 2: Understanding, LLM Benchmarks & Routing (docs/v2/phase2a-b & understanding_layer)

📝 Specialist Agents & Validation (docs/v2/)


🛠️ Project Structure

jarvis/
├── README.md                      # GitHub Repository Frontage
├── docs/
│   ├── architecture/              # Core System Architecture Suite
│   │   ├── 01_CONSUMPTION_PIPELINE.md
│   │   ├── 02_SUA_AND_SIGNAL_CLASSIFICATION.md
│   │   ├── 03_AGENT_ECOSYSTEM.md
│   │   └── 04_END_TO_END_SYSTEM_ARCHITECTURE.md
│   ├── v2/                        # Historical V2 Architecture & Phase Reports
│   └── v2.1/                      # Forensic Pipeline Audits
├── src/
│   ├── agents/                    # Specialist & Core AI Agents
│   │   ├── consumer/              # Ingestion, Parsers & Qualification
│   │   ├── sua/                   # Signal Understanding & Contract Builder
│   │   ├── financial/             # Personal Finance & Transaction Specialist
│   │   ├── financial_summary/     # Expenditure Analytics & Rollups
│   │   ├── todo/                  # Action Item & Reminder Specialist
│   │   ├── fyi/                   # Contextual Knowledge & Info Alerts
│   │   ├── daily_briefing/        # Morning & Evening Digest Synthesis
│   │   └── lifecycle/             # System Supervisor & Backfill Orchestrator
│   └── intelligence/              # LLM Clients, Dispatchers & Routing Engine
├── config/                        # Qualification Rules & Domain Rules
└── tests/                         # End-to-End Pipeline & Agent Test Suites

⚡ Quickstart & Setup

Prerequisites

  • Python 3.10+
  • uv (Fast Python package manager)
  • Ollama (with qwen2.5:1.5b model pulled for local offline execution)
  • Supabase PostgreSQL Database

Installation

# Clone repository
git clone https://github.com/pradeepgithubrepo/jarvis.git
cd jarvis

# Install dependencies using uv
uv sync

# Configure environment variables
cp .env.example .env

📖 Start Exploring Architecture Docs →

About

Jarvis Personal AI OS is an autonomous, privacy-first multi-agent system designed to ingest, classify, and understand ambient real-world signalsfrom bank whatsapp, SMS , bank statement receipts to tasks and alerts. Powered by local and remote LLMs and structured metadata fast-paths, it automates personal finance and life ledgers.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages