Your phone already has everything it needs to help you. Contexta makes it act on it — on-device, privately, without asking.

📲 Download APK  ·  🎬 Watch Demo  ·  📄 Full Proposal

---

📖 Table of Contents

• Overview
• The Problem
• The Solution
• Screenshots
• System Architecture
• Intelligent Context Pipelines
• Tech Stack
• Project Structure
• Privacy & Performance
• Getting Started
• Team

---

🌟 Overview

Contexta is a Personal AI Operating System layer for Android smartphones. It acts as an autonomous agent that continuously reads your physical environment and digital schedule to proactively adjust your phone settings — saving you from the cognitive tax of constant manual toggles.

Whether you're stepping into a meeting, walking to a destination, or arriving home after work, Contexta observes, decides, and acts seamlessly in the background — with zero cloud dependency, zero latency, and zero compromise on privacy.

Stat	Value
⚡ Action Latency	< 100ms
🧠 Processing	100% On-Device
☁️ Cloud Calls	Zero
🔋 Battery Overhead	Negligible (WorkManager)
📱 Platform	Android 10+

---

⚠️ The Problem

Smartphones are inherently reactive. This creates friction at every step of a user's day:

• Manual Switching — Users must remember to silence their phones before meetings or raise brightness when stepping outdoors.
• Micro-Distractions — The average user makes ~40 minor setting adjustments per day, collectively fragmenting focus and wasting cognitive energy.
• Rigid Schedules — Tools like "Do Not Disturb" operate on fixed time windows, ignoring sudden real-world changes.
• Context Blindness — No native Android feature connects your calendar, movement, and location into a unified, proactive response.

---

💡 The Solution

Contexta introduces a continuous, on-device perception-action loop:

┌──────────┐     ┌──────────┐     ┌──────────┐     ┌──────────┐
│  OBSERVE │────▶│  DECIDE  │────▶│   ACT    │────▶│  LEARN   │
│ Calendar │     │ On-Device│     │  DND /   │     │ Override │
│ WiFi     │     │  Logic   │     │ Volume / │     │ Feedback │
│ Accel.   │     │  + ML    │     │  Apps    │     │  Loop    │
└──────────┘     └──────────┘     └──────────┘     └──────────┘

1. Observe — Ingests native Android signals: Calendar events, Accelerometer readings, WiFi SSID.
2. Decide — Classifies user context in real-time using on-device logic (Meeting, Walking, Driving, Home, Away).
3. Act — Triggers system intents: DND mode, sound profiles, app launches.
4. Learn — Adapts instantly to user overrides without retraining or cloud sync.

---

📱 Screenshots

Full Session Walkthrough — Cold Start → Meeting → Walking → Home

1. Initial Dashboard	2. Meeting Detected → DND ON	3. Walking → Maps & Music	4. Home Profile Active
Three detectors idle at launch. 0 actions, HOME OFF, DND OFF.	Calendar finds "Sprint Standup" (91% conf). DND instantly enabled.	Accelerometer variance > 0.8 (87% conf). Maps & Music offered.	SSID "MyHomeWiFi" matches (95% conf). HOME ON, DND ON.

5. Away Mode — Office WiFi	6. Home + DND — 5 Events	7. Meeting Re-Triggered — 6 Events	8. Full Activity Log — 8 Events
SSID "OfficeWiFi_5G" ≠ home. Device correctly stays AWAY.	HOME confirmed ON. DND still active from earlier meeting. 5 actions.	Calendar re-scans every cycle. Consistent DND re-trigger. 6 events.	Complete audit: 3× DND, 1× Walk→Music, 2× HOME, 2× AWAY.

---

🏗️ System Architecture

Contexta bridges a high-fidelity React Native frontend with a low-latency Android Java Native Engine.

┌──────────────────────────────────────────────────────────────────────┐
│         REACT NATIVE FRONTEND  (Expo · TypeScript · UI/UX)           │
│                                                                      │
│   ContextDashboard   ActivityLog   OverridePanel   DetectorCards     │
├──────────────────┬───────────────────┬───────────────────────────────┤
│  CalendarBridge  │   MovementBridge  │         HomeBridge            │
│  (JSI / RN NM)   │   (JSI / RN NM)  │        (JSI / RN NM)          │
├──────────────────┴───────────────────┴───────────────────────────────┤
│          ANDROID NATIVE SENSOR FUSION ENGINE  (Java)                 │
│                                                                      │
│  ┌──────────────────┐  ┌──────────────────┐  ┌──────────────────┐   │
│  │  MeetingDetector │  │ MovementDetector │  │   HomeDetector   │   │
│  │  CalendarContract│  │  SensorManager   │  │   WifiManager    │   │
│  │  ±30min scan     │  │  XYZ variance    │  │   SSID match     │   │
│  └────────┬─────────┘  └────────┬─────────┘  └────────┬─────────┘   │
│           ▼                     ▼                      ▼             │
│  ┌──────────────────┐  ┌──────────────────┐  ┌──────────────────┐   │
│  │ MeetingModeCtrl  │  │ MovementAction   │  │ HomeProfileCtrl  │   │
│  │ DND / Silent API │  │ Maps · Music     │  │ Volume · Media   │   │
│  └──────────────────┘  └──────────────────┘  └──────────────────┘   │
│                                                                      │
│  ╔══════════════════════════════════════════════════════════════╗    │
│  ║         WorkManager  (battery-efficient background polling)  ║    │
│  ╚══════════════════════════════════════════════════════════════╝    │
└──────────────────────────────────────────────────────────────────────┘

Layer Breakdown

Layer	Technology	Responsibility
UI Layer	React Native 0.73, Expo, TypeScript	Dashboard, real-time logs, override controls
Bridge Layer	React Native Native Modules (JSI)	Bi-directional JSON sync between JS and Java
Detector Layer	Android Java	Reads raw sensors; outputs classified context
Action Layer	Android Intents & APIs	Executes system-level changes (DND, volume, app launch)
Scheduler	Android WorkManager	Battery-safe background polling; avoids aggressive Wakelocks
Backend	Spring Boot 3	Rule configuration, telemetry storage, future ML model serving

---

🧠 Intelligent Context Pipelines

Pipeline 1 · Meeting Context (Calendar NLP)

CalendarContract ──▶ ±30min window scan ──▶ Keyword match
                                              (Meeting / Call / Standup)
                                                    │
                                                    ▼
                                   ACTION_INTERRUPTION_FILTER_PRIORITY
                                        (DND ON · Ringer Silenced)

• Sensor: Android CalendarContract provider
• Window: ±30 minutes from current time
• Keywords: Meeting, Call, Standup, Interview, Review
• Confidence: Percentage match (e.g., 91%)
• Action: NotificationManager.ACTION_INTERRUPTION_FILTER_PRIORITY + silent ringer

---

Pipeline 2 · Kinetic Context (Accelerometer Math)

SensorManager (SENSOR_DELAY_NORMAL)
        │
        ▼
XYZ magnitude vector ──▶ Sliding variance window
                                │
               ┌────────────────┴────────────────┐
          var > 3.0                          var > 0.8
               │                                 │
               ▼                                 ▼
           DRIVING                           WALKING
       Launch Google Maps               Launch Music App

• Sensor: SensorManager.SENSOR_DELAY_NORMAL (TYPE_ACCELEROMETER)
• Math: variance = Σ(|magnitude − mean|²) / n over a sliding window
• Thresholds: Variance > 3.0 → Driving · Variance > 0.8 → Walking
• Confidence: Percentage (e.g., 87%)
• Action: Fires Android Intent to launch Maps (Driving) or Spotify/Music (Walking)

---

Pipeline 3 · Location Context (WiFi Geofencing)

WifiManager.getConnectionInfo()
        │
        ▼
Current SSID ──▶ Compare with stored Home SSID
                        │
          ┌─────────────┴──────────────┐
       MATCH                       NO MATCH
          │                            │
          ▼                            ▼
    Profile → HOME               Profile → AWAY
 (normalize volumes,          (office-appropriate
  lift app restrictions)        settings retained)

• Sensor: WifiManager.getConnectionInfo() — no GPS required
• Logic: String-matches current SSID against user-configured home network
• Confidence: Fixed 95% (SSID uniqueness assumption)
• Home Action: Normalizes ringer, lifts notification restrictions, applies comfort media settings
• Away Action: Retains DND-compatible office profile; conserves battery

---

🛠 Tech Stack

Area	Technology
Mobile Frontend	React Native 0.73, Expo, TypeScript
UI Design	Custom Glassmorphism · Tailwind-style utilities
Native Engine	Android Java (API 29+)
Background Jobs	Android WorkManager
Sensor APIs	CalendarContract, SensorManager, WifiManager
System APIs	NotificationManager, AudioManager, Intent
Backend	Spring Boot 3, Java 17
Build	Gradle 8, EAS Build

---

📁 Project Structure

Contexta/
├── android/                      # Android native module (Java)
│   └── app/src/main/java/
│       ├── detectors/
│       │   ├── MeetingDetector.java
│       │   ├── MovementDetector.java
│       │   └── HomeDetector.java
│       ├── controllers/
│       │   ├── MeetingModeController.java
│       │   ├── MovementActionController.java
│       │   └── HomeProfileController.java
│       └── bridges/
│           ├── CalendarBridge.java
│           ├── MovementBridge.java
│           └── HomeBridge.java
├── frontend/                     # React Native / Expo app (TypeScript)
│   ├── components/
│   │   ├── ContextDashboard.tsx
│   │   ├── DetectorCard.tsx
│   │   ├── ActivityLog.tsx
│   │   └── OverridePanel.tsx
│   └── services/
│       ├── CalendarService.ts
│       ├── MovementService.ts
│       └── HomeService.ts
├── backend/                      # Spring Boot 3 (Java)
│   └── src/main/java/
│       ├── config/
│       ├── controller/
│       └── service/
├── assets/
│   ├── logo.png                  # Contexta brand logo
│   └── screenshots/              # App screenshots (8 screens)
│       ├── dashboard_initial.jpeg
│       ├── meeting_cold_start.jpeg
│       ├── walking_detection.jpeg
│       ├── home_detection_home.jpeg
│       ├── home_detection_away.jpeg
│       ├── home_profile_active.jpeg
│       ├── meeting_detection_active.jpeg
│       └── activity_log_full.jpeg
├── docs/                         # Architecture diagrams
├── Contexta-Release.apk          # Prebuilt release APK
├── Contexta_Demo.mp4             # Demo walkthrough video
├── Contexta_Openclaw.pdf         # Full hackathon proposal
└── README.md

---

🛡️ Privacy & Performance

Principle	Implementation
Zero Cloud Processing	All detection logic runs 100% on-device. No calendar events, GPS data, or sensor telemetry are ever transmitted externally.
No Persistent Storage	Sensor readings are processed in-memory and discarded immediately after classification.
Negligible Battery Impact	Android WorkManager with batched tasks and passive listeners — no aggressive CPU Wakelocks.
Sub-100ms Latency	Native Java execution eliminates JS-bridge round-trips for time-critical actions.
Offline Ready	Works with zero internet connection. All intelligence is pre-compiled on the device.

---

🚀 Getting Started

Prerequisites

• Node.js 18+
• Android Studio (Flamingo or later)
• Java 17
• Expo CLI: npm install -g expo-cli

Run the Frontend

git clone https://github.com/AnanthAkshay/Contexta.git
cd Contexta/frontend
npm install
npx expo run:android

Build the Native Android Module

cd Contexta/android
./gradlew assembleRelease

Run the Backend

cd Contexta/backend
./mvnw spring-boot:run

Install the Prebuilt APK

adb install Contexta-Release.apk

Grant permissions on first launch: Calendar, Physical Activity, and Nearby WiFi Devices.

---

👨‍💻 Team

Team Beta Onepiece — M.S. Ramaiah Institute of Technology, Bengaluru

Name	Role
Akshay A	Team Lead · Frontend (React Native)
Aaditya V	Backend (Spring Boot)
Tejas M	UI/UX Design
H M Pranav	Database & Integration

Submitted for the OpenClaw Hackathon by Samsung Prism 2026 — Daily Utility Track.

---

Made with ❤️ in Bengaluru  ·  M.S. Ramaiah Institute of Technology

⚡ On-device  ·  🧠 No cloud  ·  📴 Offline ready

.....