Flazz is a local-first AI desktop application that acts as your intelligent coworker. Unlike cloud-based AI assistants, Flazz keeps all your data on your machine while providing powerful AI capabilities with long-term memory.
Think of it as your personal AI that:
- Remembers everything - Automatically builds a knowledge graph from your emails, meetings, and notes
- Learns your patterns - Observes your habits and preferences to provide personalized assistance
- Creates its own skills - AI autonomously generates reusable workflows based on your tasks
- Stays private - All data lives in
~/Flazzon your machine - Connects to 100+ platforms - One Composio API key unlocks Gmail, Slack, GitHub, Notion, and more
- Adapts continuously - Evolves with your workflow without manual configuration
Flazz doesn't just store information—it actively learns from your behavior and builds intelligence over time.
Automatic Knowledge Extraction
- Processes emails, meetings, voice memos, and documents in the background
- Extracts people, organizations, projects, and topics automatically
- Creates bidirectional wiki-links between related entities
- Builds a navigable knowledge graph like Obsidian, but fully automated
Behavioral Learning
- Observes your work patterns and communication style
- Learns which tasks you perform frequently
- Remembers your preferences for tools, formats, and workflows
- Adapts responses based on your past interactions
Context Retention Across Sessions
- Maintains conversation history indefinitely
- Recalls details from weeks or months ago
- Connects new information with existing knowledge
- Never forgets important context
Smart Labeling & Filtering
- AI-powered categorization of emails and documents
- Automatically filters noise from important information
- Learns what matters to you over time
Flazz can create and refine its own capabilities without manual programming.
AI-Generated Skills When you perform a task repeatedly, Flazz notices and offers to create a reusable skill:
- "I see you often review pull requests. Should I create a 'PR Review' skill?"
- The AI writes the skill definition, templates, and instructions automatically
- Skills are stored as markdown files you can edit or version control
Skill Evolution
- Skills improve based on your feedback
- AI refines prompts and workflows automatically
- Learns from successful vs. unsuccessful executions
- Suggests optimizations and enhancements
Example: Auto-Generated Code Review Skill
---
name: code-review-assistant
description: Reviews code for security, performance, and best practices
category: development
auto_generated: true
created_from: 5 similar tasks on 2024-03-15
---
# Code Review Assistant
Analyzes code changes and provides:
- Security vulnerability detection
- Performance bottleneck identification
- Best practice recommendations
- Documentation quality assessment
Learned preferences:
- Prefers functional programming patterns
- Focuses on TypeScript strict mode compliance
- Emphasizes test coverageSkill Library
- Browse AI-generated and community skills
- Share skills across your team
- Import skills from templates
- Version control with automatic revision history
Connect to 100+ platforms with a single API key—no individual OAuth flows needed.
One Key, Unlimited Connections
# Add your Composio API key once
COMPOSIO_API_KEY=your_key_here
# Instantly access:
- Gmail, Outlook, Yahoo Mail
- Slack, Discord, Microsoft Teams
- GitHub, GitLab, Bitbucket
- Notion, Confluence, Google Docs
- Jira, Linear, Asana, Trello
- Salesforce, HubSpot, Pipedrive
- Google Calendar, Outlook Calendar
- Google Drive, Dropbox, OneDrive
- And 80+ more platformsUnified Workflow Automation
- Send emails, create calendar events, update tickets—all from one interface
- AI automatically chooses the right platform based on context
- Cross-platform workflows: "When I get a GitHub PR, notify me in Slack and add to my calendar"
No OAuth Hassle
- Traditional approach: Configure OAuth for each service individually
- Flazz + Composio: One API key, instant access to everything
- Composio handles authentication, rate limits, and API changes
Example Workflows
"Email the team about the deployment"
→ AI detects you use Gmail, sends via Composio
"Create a ticket for this bug"
→ AI knows your project uses Linear, creates ticket automatically
"Schedule a meeting with Sarah next week"
→ AI checks your Google Calendar, finds free slots, sends invite
Works with all major LLM providers out of the box:
- OpenAI (GPT-4, GPT-4o, o1, o3)
- Anthropic (Claude 3.5 Sonnet, Claude 4)
- Google (Gemini Pro, Gemini Flash)
- xAI (Grok)
- Groq, Mistral, Cohere, DeepSeek, and more
- Local models via Ollama
- MCP (Model Context Protocol) - Connect to any MCP server for custom tools
- Composio integration - 100+ platforms with one API key (see above)
- Built-in tools - File system, search, git, command execution, and more
- Custom agents - Create specialized AI agents for specific domains
- Markdown-first - All notes are plain markdown files
- Full-text search - Fast grep-based search across your workspace
- Version history - Git-backed version control for your memory
- Rich editor - TipTap-powered markdown editor with live preview
- 100% local-first - Your data never leaves your machine
- Workspace sandboxing - All file operations are scoped to
~/Flazz - Permission system - Explicit approval for sensitive operations
- No telemetry - Optional analytics only
- Node.js 18+ and npm/pnpm
- Git
# Clone the repository
git clone https://github.com/yourusername/flazz.git
cd flazz
# Install dependencies
pnpm install
# Build dependencies
pnpm run deps
# Start development server
pnpm run devThe app will open automatically. Your workspace will be created at ~/Flazz.
-
Configure your LLM provider
- Open Settings → Models
- Add your API key for OpenAI, Anthropic, or any supported provider
- Select your preferred model
-
Start chatting
- Ask questions, request code, or have the AI help with tasks
- All conversations are saved in your workspace
-
Connect your data (optional)
- Add your Composio API key in Settings → Integrations
- Instantly connect Gmail, Slack, GitHub, Notion, and 100+ other platforms
- Set up email sync to build your memory graph automatically
- Connect calendar for meeting transcripts and scheduling
- Add voice memos for automatic transcription and knowledge extraction
Flazz automatically extracts structured knowledge from your communications:
Email from Sarah Chen → Creates/updates:
People/Sarah Chen.md
Organizations/Acme Corp.md
Projects/Q1 Integration.md
All entities are linked bidirectionally, creating a navigable knowledge graph.
Flazz can create skills automatically or you can define them manually. Skills are reusable AI workflows stored in ~/Flazz/memory/Skills/.
AI-Generated Skill Example: When you repeatedly perform similar tasks, Flazz offers to create a skill:
---
name: meeting-summarizer
description: Summarizes meeting transcripts with action items
category: productivity
auto_generated: true
learned_from: 12 meeting summaries
---
# Meeting Summarizer
Automatically extracts from meeting transcripts:
- Key decisions made
- Action items with owners
- Follow-up questions
- Next meeting topics
Your preferences (learned):
- Prefers bullet points over paragraphs
- Always includes attendee list
- Highlights deadlines in bold
- Sends summary to Slack #team channelManual Skill Creation: You can also create skills yourself:
---
name: code-reviewer
description: Review code for best practices
category: development
---
# Code Reviewer
Review the provided code for:
- Security vulnerabilities
- Performance issues
- Best practices
- Documentation quality
...Skill Features:
- Version control with automatic revision history
- Supporting files (templates, scripts, references)
- Shareable across team members
- AI continuously improves skills based on usage
Connect to 100+ platforms with a single API key:
{
"composio": {
"apiKey": "your_composio_api_key",
"enabledApps": [
"gmail",
"slack",
"github",
"notion",
"googlecalendar",
"linear"
]
}
}Available Platforms (100+):
- Communication: Gmail, Outlook, Slack, Discord, Teams, Telegram
- Development: GitHub, GitLab, Bitbucket, Jira, Linear
- Productivity: Notion, Confluence, Google Docs, Asana, Trello
- CRM: Salesforce, HubSpot, Pipedrive
- Calendar: Google Calendar, Outlook Calendar
- Storage: Google Drive, Dropbox, OneDrive
- And many more...
How It Works:
- Get a Composio API key from composio.dev
- Add it to Flazz settings once
- AI automatically uses the right platform based on your request
- No individual OAuth configuration needed
Example Usage:
You: "Email the team about tomorrow's deployment"
AI: Uses Gmail via Composio to send email
You: "Create a ticket for the login bug"
AI: Creates issue in your GitHub/Linear/Jira (whichever you use)
You: "Add this to my calendar"
AI: Creates event in Google Calendar via Composio
Example: AI automatically scheduling meetings and managing your calendar
Flazz stores configuration in ~/Flazz/.config/:
models.json- LLM provider settingsmcp.json- MCP server configurationscomposio.json- Composio API key and enabled platformssecurity.json- Permission and security settingslearning.json- Behavioral learning preferences and patterns
Flazz follows a clean layered architecture:
┌─────────────────────────────────────────┐
│ Renderer (React UI) │
│ - Chat interface │
│ - Knowledge browser │
│ - Settings & configuration │
└─────────────────────────────────────────┘
↕ IPC
┌─────────────────────────────────────────┐
│ Main Process (Electron) │
│ - Window management │
│ - IPC handlers │
│ - OAuth flows │
└─────────────────────────────────────────┘
↕
┌─────────────────────────────────────────┐
│ Core (Business Logic) │
│ - Agent runtime │
│ - Memory graph │
│ - Workspace operations │
│ - Integrations (MCP, Composio) │
└─────────────────────────────────────────┘
↕
┌─────────────────────────────────────────┐
│ Workspace (~/Flazz) │
│ - memory/ │
│ - .config/ │
│ - .trash/ │
└─────────────────────────────────────────┘
apps/renderer- React frontendapps/main- Electron main processapps/preload- Secure IPC bridgepackages/core- Application logicpackages/shared- Shared types and schemas
flazz/
├── apps/
│ ├── main/ # Electron main process
│ ├── renderer/ # React UI
│ └── preload/ # IPC bridge
├── packages/
│ ├── core/ # Business logic
│ └── shared/ # Shared types
├── assets/ # Icons and images
└── docs/ # Documentation
# Development
pnpm run dev # Start dev server
pnpm run deps # Build dependencies
# Building
pnpm run build # Build all packages
cd apps/main && pnpm run make # Package Electron app for current platform
# Quality
pnpm run lint # Run ESLint
pnpm run typecheck # Type check all packages
pnpm run test # Run testsSee Building and Releasing for detailed instructions on:
- Building for Windows, macOS, and Linux
- Setting up automated releases via GitHub Actions
- Code signing for Windows and macOS
- Creating installers and portable versions
Quick build for current platform:
pnpm install
pnpm run deps
cd apps/renderer && pnpm run build
cd ../main && pnpm run build && pnpm run makeOutput will be in apps/main/out/make/
- Define contracts in
packages/shared - Implement logic in
packages/core - Add IPC handlers in
apps/main - Build UI in
apps/renderer
See AGENTS.md for detailed architecture guidelines.
We welcome contributions! Please read our Contributing Guide for details on:
- Setting up your development environment
- Code style and architecture guidelines
- Commit message conventions
- Pull request process
- Testing requirements
Quick start for contributors:
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Make your changes following our architecture guidelines
- Run tests and linting (
pnpm run lint && pnpm run test) - Commit your changes with conventional commits
- Push to your branch (
git push origin feature/amazing-feature) - Open a Pull Request using our PR template
For detailed development setup, see Development Setup.
Security is a top priority. Please review our Security Policy for:
- Supported versions
- How to report vulnerabilities
- Security best practices
- Known security considerations
To report a vulnerability, email security@flazz.app or create a private security advisory on GitHub. Do not open a public issue for security vulnerabilities.
Flazz is licensed under the Apache License 2.0.
Copyright 2024-2026 Flazz Contributors
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
- Local-first - Your data stays on your machine, not on OpenAI/Anthropic servers
- Long-term memory - Builds a persistent knowledge graph that grows over time
- Multi-provider - Not locked to one LLM provider, switch anytime
- Extensible - Add custom tools, skills, and integrations
- Learns your habits - Adapts to your workflow automatically
- 100+ platform integrations - One Composio key vs. no integrations
- General purpose - Not just for coding, handles emails, meetings, notes, tasks
- Memory system - Remembers context across sessions, not just current file
- Knowledge workspace - Integrated note-taking and knowledge management
- Privacy-first - No code sent to third parties, everything local
- Auto-generated skills - AI creates reusable workflows from your patterns
- Platform integrations - Connect to Gmail, Slack, Notion, etc.
- Native AI integration - Built from the ground up for AI, not bolted on
- Automatic knowledge extraction - No manual note-taking required
- Agent runtime - Complex multi-step workflows and tool execution
- Tool ecosystem - MCP and Composio integrations out of the box
- Behavioral learning - AI learns your preferences and adapts
- Self-improving - Creates and refines skills automatically
- 100% local - Your data never leaves your machine
- Open source - Full transparency and customization
- Multi-LLM - Use any provider, not locked to one
- Deeper integrations - Composio enables 100+ platforms vs. limited integrations
- Skill system - Reusable workflows that improve over time
- No subscription - Pay only for LLM API usage, not the app
- Mobile companion app
- Real-time collaboration
- Advanced memory search (vector embeddings)
- Plugin marketplace
- Cloud sync (optional, encrypted)
- Voice interface
- Multi-language support
- Issues - GitHub Issues
- Discussions - GitHub Discussions
- Twitter - @flazz_ai
Built with amazing open source projects:
- Electron - Cross-platform desktop apps
- React - UI framework
- Vercel AI SDK - LLM integration
- TipTap - Rich text editor
- Radix UI - UI primitives
- Model Context Protocol - Tool integration standard
Made with love by the Flazz team