Lotti

Your AI‑powered context manager — a private, local‑first assistant for your tasks, notes, and audio.

Lotti is an open-source personal assistant that helps you capture, organize, and understand your work and life through AI-enhanced task management, audio recordings, and intelligent summaries—all while keeping your data entirely under your control.

Read more on Substack | Project Background

Table of Contents

• Why Lotti?
  • Example Use Cases
• Core Features
  • AI-Powered Intelligence
  • Comprehensive Tracking
  • Privacy & Control
• AI Provider Configuration
• Getting Started
  • Installation
  • Beta Testing
  • Development
• Documentation
• Use Cases
  • For Developers
  • For Knowledge Workers
  • For Personal Growth
• Contributing
• Technical Stack
• Philosophy
• License
• Acknowledgments

Why Lotti?

Most AI-powered tools require you to upload and store your personal data on their servers, creating privacy risks and vendor lock-in. Lotti takes a different approach:

• Complete data ownership: Your information stays on your devices. When you opt into cloud inference, European‑hosted, no‑retention providers are available
• Configurable AI providers per category: Choose between OpenAI, Anthropic, Gemini, Ollama (local), or any OpenAI-compatible provider on a per-category basis
• Privacy-first design: You control exactly what data gets shared with AI providers—only for specific inference calls via your API keys
• No vendor lock-in: Your data remains portable and accessible, independent of any subscription

Example Use Cases

• Pick up a task from last week — see your last notes, time spent, and a one‑paragraph recap
• Record a quick voice note — later it’s transcribed and turned into a checklist
• Ask “What did I finish in June?” — get a dated list with brief summaries

Core Features

Currently, Lotti's AI capabilities are focused on task management and productivity. Habit tracking is fully functional but will receive AI enhancements in future updates.

🤖 AI-Powered Intelligence

• Smart Summaries: Automatically generate summaries of tasks, capturing key points and progress
• Audio Transcription: Transcribe recordings using either local Whisper (OpenAI's open weights model, 99 languages supported) or cloud providers with audio capabilities like Gemini Flash/Pro
• Context Recap: Resume a task with a one‑screen recap of your latest notes, time, and progress
• Intelligent Checklists: Transform rambling audio notes into actionable checklists
• Chat with Your Data: Ask questions about your tasks, learnings, and achievements across any time period

📝 Comprehensive Tracking

• Tasks: Full lifecycle management (open, groomed, in progress, blocked, done, rejected)
• Audio Recording: Capture thoughts, progress notes, and brain dumps
• Time Tracking: Record time spent on tasks and projects
• Journal Entries: Written reflections and documentation
• Habits: Define and monitor daily habits and routines
• Health Data: Import from Apple Health and other sources
• Custom Metrics: Track anything that matters to you

🔐 Privacy & Control

• Local-Only Storage: All data is permanently stored only on your devices and never in the cloud
• Encrypted Sync: End-to-end encrypted synchronization between your devices (desktop/laptop and mobile) using Matrix (requires a Matrix account — self-hosted or public homeserver)
• Selective AI Usage: Configure AI providers per category—keep sensitive data completely local with Ollama but use state‑of‑the‑art (frontier) cloud models when appropriate
• Your API Keys: When you choose cloud AI, data is shared only for that specific inference call. Please review the respective provider's terms and privacy policy to understand how they handle your data
• GDPR-Compliant Options: European-hosted AI providers with no data retention policies available for enhanced privacy
• Built for on‑device: Designed for the era when local AI inference becomes standard

AI Provider Configuration

Lotti supports multiple AI providers, configurable per category:

• Cloud Providers: OpenAI, Anthropic Claude, Google Gemini
• Local Inference: Ollama for complete privacy (requires capable hardware)
  • Full functionality available with local models like Qwen3 (8B), GPT-OSS (20B/120B), Gemma3 (12B/27B)
  • Combined with local Whisper for speech recognition, enables 100% offline AI capabilities
• OpenAI-Compatible: Any provider with OpenAI-compatible APIs
• European Options: GDPR-compliant hosted alternatives

Configure different providers for different aspects of your life—use cutting-edge models for work projects while keeping personal reflections completely private with local inference. With sufficient hardware, you can run everything locally without any cloud dependency.

Getting Started

Installation

See DEVELOPMENT.md for setup and development workflow.

Beta Testing

• Build it yourself: for iOS, macOS, Android, Linux, Windows
• iOS/macOS: TestFlight builds are available for select users, will be available more broadly in due course
• Linux: See tar.gz files on GitHub releases - will also be available via Flatpak soon

Development

• Install Flutter (instructions) — FVM recommended; repo includes .fvmrc
• Install dependencies: make deps
• Static analysis: make analyze
• Tests: make test • Coverage report: make coverage
• Code generation: make build_runner • Localization: make l10n
• Run locally: macOS fvm flutter run -d macos • others flutter run -d <device>

See DEVELOPMENT.md for detailed development setup.

Documentation

• Manual - How to use Lotti
• Background Story - The inspiration and evolution of Lotti
• Architecture - Technical design and AI integration
• Privacy Policy - Our commitment to your privacy
• Contributing - How to help and our standards

Use Cases

For Developers

• Track project progress with automatic context recovery
• Document decisions and learnings with searchable audio notes
• Generate sprint summaries and retrospectives from your task data

For Knowledge Workers

• Maintain focus with AI-powered context switching
• Build a searchable knowledge base from daily work
• Track time and generate reports across projects

For Personal Growth

• Monitor habits and health metrics
• Reflect on achievements and learnings over time
• Keep a multilingual audio journal

Contributing

See CONTRIBUTING.md.

Technical Stack

• Frontend: Flutter (iOS, macOS, Android, Windows, Linux)
• AI Integration: Multiple providers with streaming support, including Ollama for 100% private local inference
• Audio: Local Whisper (OpenAI's open weights model) or cloud providers with multimodal audio support
• Storage: Local SQLite, no cloud storage
• Synchronization: End-to-end encrypted sync using Matrix infrastructure (requires a Matrix account)
• Testing: Comprehensive unit and integration tests

Philosophy

Lotti represents a different approach to AI-powered productivity:

1. Your data stays yours: No company should own your thoughts and experiences
2. AI as a tool, not a service: Use AI capabilities without subscription lock-in
3. Privacy by design: Choose exactly what to share, when, and with whom
4. Future-focused: Built for the coming era of powerful local AI

License

Lotti is open source under LICENSE.

Acknowledgments

Special thanks to the Flutter team, OpenAI for the Whisper model, and all contributors who believe in privacy-respecting AI tools.

---

Building in public • Follow development here on GitHub • Read updates on Substack