# Project export: AlarmGemini - Agentic AI on Android This document was generated by HackStack to give an AI agent context about a hackathon project. Sections are labeled with their provenance; content marked as truncated was cut to keep this document small. ## Project metadata - Hackathon: UC Berkeley AI Hackathon 2025 - Tagline: Agentic android alarm app using Gemini API with voice input. Say 'set 5 backup alarms' - AI autonomously calculates intervals, resolves conflicts. First truly independent alarm assistant. - Devpost: https://devpost.com/software/alarmgemini - GitHub: https://github.com/somesh-bagadiya/AlarmGemini - Video: https://www.youtube.com/embed/en3SYS2HcoQ?enablejsapi=1&hl=en_US&rel=0&start=&version=3&wmode=transparent - Team: 1 GitHub contributor(s) — Somesh Bagadiya (2 commits) ## Devpost submission (written by the team) ### Overview 🚀 What It Does AlarmGemini demonstrates the future of conversational AI integration - showing how any app can implement intelligent chat and voice interactions using modern LLMs. The MVP Concept: This isn't just an alarm app - it's a proof-of-concept for conversational AI patterns that any application can adopt. Traditional App Interaction: User: Navigates through menus and forms App: Executes predefined actions Conversational AI Integration Examples: Natural Language Variations: User: "Set alarm at 7 in morning" → AI creates 7:00 AM alarm User: "Set alarm 5 min from now" → AI calculates current time + 5 minutes User: "Set 3 alarm at 7 in the morning 10 mins apart" → AI creates 7:00, 7:10, 7:20 AM Core Integration Patterns Demonstrated: Voice-to-Action Pipeline: Speech → AI Processing → App Functions Natural Language Understanding: Converting conversational commands to app actions Agentic Decision Making: AI autonomously calculates optimal solutions Contextual Responses: AI explains its reasoning and provides feedback 🛠 How We Built It Technology Stack: Android: Kotlin + Jetpack Compose + Material Design 3 AI: Gemini 2.0 Flash API (Direct REST integration) Voice: Android Speech Recognition + Text-to-Speech Architecture: MVVM with Coroutines and StateFlow Speech-to-Speech Pipeline: Key Technical Innovations: Natural Language Parsing: Handles variations like "7 in morning", "5 min from now", "10 mins apart" Contextual Time Processing: Converts relative times ("from now") using current system time Autonomous Interval Calculation: AI computes spacing for "3 alarms 10 mins apart" automatically Intelligent Command Recognition: Understands intent from informal speech patterns 💡 What We Learned Agentic AI is the Future: Moving beyond reactive chatbots to proactive AI that analyzes, decides, and acts autonomously represents a fundamental shift in human-computer interaction. Context is Everything: The difference between "set alarm" and "set backup alarms for important meeting" requires understanding semantic context, temporal awareness, and user intent. Voice + AI = Magic: When speech recognition, natural language processing, and text-to-speech work seamlessly together, the interaction feels truly magical. 💪 Challenges We Overcame 1. SDK Migration Crisis: Started with Google's deprecated Generative AI SDK that caused crashes. Pivoted to direct REST API integration using OkHttp. 2. Real-Time Voice Processing: Implemented optimized API calls and comprehensive visual feedback to ensure natural conversation flow. 3. Complex Command Parsing: Developed sophisticated prompt engineering and fallback patterns to handle multi-part voice commands reliably. 🎯 What Makes This Special This isn't about alarms - it's about demonstrating conversational AI integration patterns: Compared to existing approaches: Alarmi focuses on specific alarm features AI Voice Alarm provides basic voice reminders Most apps still rely on traditional UI/UX patterns AlarmGemini's Integration Patterns: Reusable Architecture: Speech-to-speech pipeline that any app can implement LLM Integration Framework: Shows how to connect Gemini API to app functions Conversational UX Paradigm: Demonstrates natural language as primary interface Agentic Behavior Templates: Patterns for autonomous AI decision-making in apps 🚀 Impact & Demo Value The Real MVP - Conversational AI Integration: Replicable Patterns: Any app can adopt these chat/voice integration techniques Framework Demonstration: Shows how to connect LLMs to app functionality UX Paradigm Shift: Natural language as primary interface, not just a feature Broader Implications for App Development: This project shows how any application can integrate conversational AI: E-commerce: "Find me a blue dress under $100 for a wedding" Finance: "Move $500 from savings to checking and pay my electricity bill" Healthcare: "Schedule my annual checkup and remind me about my medication" Productivity: "Create a project timeline for the Q2 launch with 5 milestones" Technical Integration Patterns: Voice-to-Function Pipeline: Speech → NLP → App Actions → Voice Response LLM-App Bridge: Converting natural language to structured app commands Agentic Decision Framework: AI reasoning patterns for autonomous app behavior Conversational State Management: Maintaining context across interactions Built with: Kotlin, Jetpack Compose, Gemini 2.0 Flash API, Android Speech Recognition, Text-to-Speech The MVP Demonstration: Natural language processing that handles real speech patterns: "Set alarm at 7 in morning" (informal grammar) "Set alarm 5 min from now" (relative time) "Set 3 alarm at 7 in the morning 10 mins apart" (complex multi-alarm) This shows how any app can implement conversational AI interactions! 🎤✨ ## README (from the GitHub repository) No README available. ## Detected evidence (automated analysis) Indexed codebase: 47 recognized source files, 131 KB. - Kotlin (language) — detected in the code - AI coding agent: Cursor — evidence: config files committed to the repository ## Codebase structure (from repository index) ### Files (60 of 60) ``` .cursor/rules/alarm-core.mdc .cursor/rules/alarmgemini-project.mdc .cursor/rules/chat-ai.mdc .gitignore .idea/.gitignore .idea/AndroidProjectSystem.xml .idea/codeStyles/codeStyleConfig.xml .idea/codeStyles/Project.xml .idea/compiler.xml .idea/deploymentTargetSelector.xml .idea/deviceManager.xml .idea/gradle.xml .idea/inspectionProfiles/Project_Default.xml .idea/migrations.xml .idea/misc.xml .idea/runConfigurations.xml .idea/studiobot.xml app/.gitignore app/build.gradle.kts app/plan.md app/proguard-rules.pro app/src/androidTest/java/com/example/alarmgemini/ExampleInstrumentedTest.kt app/src/main/AndroidManifest.xml app/src/main/java/com/example/alarmgemini/ai/SimpleNlp.kt app/src/main/java/com/example/alarmgemini/alarm/AlarmReceiver.kt app/src/main/java/com/example/alarmgemini/alarm/AlarmScheduler.kt app/src/main/java/com/example/alarmgemini/alarm/AlarmTools.kt app/src/main/java/com/example/alarmgemini/AlarmListScreen.kt app/src/main/java/com/example/alarmgemini/AlarmListViewModel.kt app/src/main/java/com/example/alarmgemini/AlarmRingActivity.kt app/src/main/java/com/example/alarmgemini/chat/ChatMessage.kt app/src/main/java/com/example/alarmgemini/chat/ChatScreen.kt app/src/main/java/com/example/alarmgemini/chat/ChatViewModel.kt app/src/main/java/com/example/alarmgemini/chat/VoiceChatScreen.kt app/src/main/java/com/example/alarmgemini/MainActivity.kt app/src/main/java/com/example/alarmgemini/MainViewModel.kt app/src/main/java/com/example/alarmgemini/services/McpService.kt app/src/main/java/com/example/alarmgemini/ui/theme/Color.kt app/src/main/java/com/example/alarmgemini/ui/theme/Theme.kt app/src/main/java/com/example/alarmgemini/ui/theme/Type.kt app/src/main/java/com/example/alarmgemini/UiState.kt app/src/main/res/drawable/ic_launcher_background.xml app/src/main/res/drawable/ic_launcher_foreground.xml app/src/main/res/mipmap-anydpi-v26/ic_launcher_round.xml app/src/main/res/mipmap-anydpi-v26/ic_launcher.xml app/src/main/res/values/colors.xml app/src/main/res/values/strings.xml app/src/main/res/values/themes.xml app/src/main/res/xml/backup_rules.xml app/src/main/res/xml/data_extraction_rules.xml app/src/test/java/com/example/alarmgemini/alarm/AlarmSchedulerTest.kt app/src/test/java/com/example/alarmgemini/ExampleUnitTest.kt build.gradle.kts dependencies.txt gradle.properties gradle/libs.versions.toml gradle/wrapper/gradle-wrapper.properties gradlew gradlew.bat settings.gradle.kts ``` ### Dependencies No dependency index available. ### Recent commits (newest first) - Update gradle.properties - Chat and voice working and at same stage - first commit ## Key source files (fetched from GitHub, selected and truncated for size) ### app/plan.md ```markdown # AlarmGemini - **Hackathon MVP: Agentic AI Alarm Assistant** ## 🏆 **HACKATHON CONCEPT** **Goal**: Demonstrate advanced agentic AI capabilities in an alarm app that thinks and acts autonomously **Timeline**: Hackathon demo-ready (focus on impressive AI features, not production polish) **Theme**: Show how AI can be truly "agentic" - making intelligent decisions, learning patterns, and acting proactively Following [hackathon AI project planning strategies](https://medium.com/@lalarukh.1992/day-3-of-30-i-used-ai-to-plan-a-hackathon-project-from-start-to-finish-ec1f989b8b5f), this MVP demonstrates **agentic AI that revolutionizes user interaction** through natural language understanding and autonomous decision-making. ## 🎯 **CORE DEMO FEATURES (What Judges Will See)** ### **1. Advanced Agentic Intelligence** 🤖 **"Set 5 alarms in 5 minutes"** → AI automatically calculates optimal 1.25-minute intervals **"Create backup alarms for my important meeting"** → AI proactively creates 3 smart backup alarms **"Set alarms starting at 6 AM with increasing intensity"** → AI plans progressive alarm sequence ### **2. Context-Aware Decision Making** 🧠 - **System time awareness**: "Set alarm in 2 hours" uses real current time - **Intelligent conflict resolution**: AI detects overlapping alarms and suggests optimizations - **Proactive suggestions**: "You usually snooze at 7 AM, want me to set it for 7:15 instead?" ### **3. Natural Language Mastery** 💬 - **Complex pattern recognition**: Understands multi-part commands - **Conversational memory**: Remembers context within conversation - **Graceful fallbacks**: Never fails - always finds a way to help ### **4. Autonomous Learning** 📈 - **Pattern detection**: Learns user habits and preferences - **Proactive recommendations**: Suggests optimizations without being asked - **Adaptive responses**: Gets smarter with each interaction ## 🚀 **HACKATHON DEVELOPMENT PHASES** ### **Phase 1: Core Agentic System** ✅ **COMPLETE** - [X] **Advanced SimpleNlp.kt**: Multi-alarm intelligence with `AgenticCommand` architecture - [X] **Smart ChatViewModel.kt**: Dual-layer AI processing with context awareness - [X] **Intelligent scheduling**: Automatic interval calculation and conflict detection - [X] **System time integration**: Real-time context for all operations - [X] **Demo-ready UI**: Translucent chat popup showcasing AI interaction ### **Phase 2: Hackathon Demo Enhancements** ⭐ **CURRENT FOCUS** #### **Week 1: Advanced Agentic Features** 🎪 **Goal**: Add impressive AI capabilities that wow judges - [ ] **Proactive AI Suggestions** - [ ] "You have 5 alarms at 7 AM this week, want me to optimize them?" - [ ] "Based on your pattern, I recommend setting a backup alarm" - [ ] Auto-suggest alarm improvements during chat - [ ] **Smart Pattern Recognition** - [ ] Detect user habits: "You always set alarms on Sunday night for Monday" - [ ] Learn preferences: "You prefer 10-minute intervals between backup alarms" [truncated — 7055 more characters] ``` ### build.gradle.kts ```kotlin // Top-level build file where you can add configuration options common to all sub-projects/modules. plugins { alias(libs.plugins.android.application) apply false alias(libs.plugins.kotlin.android) apply false alias(libs.plugins.kotlin.compose) apply false alias(libs.plugins.google.android.libraries.mapsplatform.secrets.gradle.plugin) apply false } ``` ### settings.gradle.kts ```kotlin pluginManagement { repositories { google { content { includeGroupByRegex("com\\.android.*") includeGroupByRegex("com\\.google.*") includeGroupByRegex("androidx.*") } } mavenCentral() gradlePluginPortal() } } dependencyResolutionManagement { repositoriesMode.set(RepositoriesMode.FAIL_ON_PROJECT_REPOS) repositories { google() mavenCentral() } } rootProject.name = "AlarmGemini" include(":app") ``` ### .idea/compiler.xml ```xml ``` ### .idea/studiobot.xml ```xml ``` ### .idea/AndroidProjectSystem.xml ```xml ``` ### .idea/migrations.xml ```xml ``` ### .idea/deviceManager.xml ```xml ``` ### .idea/misc.xml ```xml ``` ### .idea/deploymentTargetSelector.xml ```xml ``` [37 more indexed source files omitted to keep this export small. The full file list is in the Codebase structure section above.]