An AI that lives on your Mac, learns everything about you, and becomes more of yourself over time. It remembers. It automates. It evolves.
Free & open source • Your data stays on your device • No accounts required
Get updates → rappter.beehiiv.comThree steps. No terminal. No setup headaches.
Download the app. A 🦖 appears in your menu bar. Click it.
Chat with your rappter. Tell it about yourself. It remembers everything.
Over days and weeks, it learns your patterns, preferences, and voice. It becomes your Twin.
Not features. Superpowers — in plain English.
Tell it once, it knows forever. Preferences, contacts, decisions, context.
Every morning: weather, calendar, priorities — spoken aloud or as a notification.
Chat in the menu bar, on Telegram, Slack, Discord — wherever you already are.
It teaches you what it can do. One tip per day for 30 days. Click the notification to try it.
While you sleep, it consolidates its memory. Deduplicates. Prunes. Wakes up sharper.
New features arrive automatically. No manual updates, no app store.
Nothing leaves your machine. No cloud. No accounts. Truly private.
Need a researcher? A coder? An ops engineer? Switch personas in one click.
For Developers
Serious infrastructure for serious builders.
Each agent is one file. No routing layers, no config files, no dependency injection. Read it, understand it, modify it.
Works out of the box with Ollama. No OpenAI key required. Add keys later for GPT-4, Claude, Gemini, or Groq if you want.
Everything runs on your machine. Your data never leaves. Works offline with local models.
SQLite-backed memory that survives restarts. Semantic search across all conversations.
Ollama, OpenAI, Anthropic, Google Gemini, Groq. Switch with one config change.
CLI, Telegram, Discord, Slack, iMessage, WhatsApp, REST API, WebSocket, and more.
Real-time bidirectional communication. Build dashboards, mobile apps, or integrations.
Drop a .js or .py file in the agents folder. It's live. Hot-reload included.
TypeScript for the framework. Python for AI/ML agents. Both first-class citizens.
Your agent has a personality. It evolves based on usage. Feed it knowledge, watch it grow.
Define personality, expertise, tone, and goals. Share templates on ClawHub.
Overnight memory consolidation. Deduplicates, prunes, and summarizes. Wakes up sharper.
Daily tips delivered as notifications. One feature per day. Users learn by doing.
Built-in update agent. Checks, downloads, and applies updates automatically.
Runs headless as a system service. Cron jobs, monitoring, automation — always on.
Every agent is a single file. Read it, fork it, remix it.
Foundation for all agents. Chat, reason, respond.
Executes system commands safely with sandboxing.
Store, search, and recall persistent memories.
All-in-one personal assistant with tools.
Self-improving agent that rewrites itself.
Fetches, parses, and summarizes web content.
Full browser automation with Playwright.
Curated tech news from Hacker News.
Send messages across any connected channel.
Text-to-speech with multiple voice engines.
Manages conversation sessions and context.
Schedule recurring tasks with cron expressions.
Generate and analyze images with AI models.
Overnight memory consolidation and pruning.
30-day onboarding, one tip at a time.
Self-updating: check, download, apply.
Daily briefing: weather, calendar, priorities.
How we compare to other agent frameworks.
| Feature | openrappter | OpenClaw | LangChain |
|---|---|---|---|
| API Keys Required | None (Ollama) | Required | Required |
| Cost to Start | $0 | Pay per token | Pay per token |
| Privacy | 100% local | Cloud only | Depends |
| Setup Time | 1 command | ~30 min | Hours |
| Persistent Memory | Built-in | No | Add-on |
| Agent Registry | 22 built-in | Marketplace | DIY |
| Open Source | MIT | Partial | MIT |
| Community Hub | ClawHub | GPT Store | No |
| Channels | 15+ | API only | DIY |
| WebSocket Gateway | Built-in | No | No |
Every agent follows the same lifecycle. No exceptions.
execute(kwargs)
│
├── slosh(query) ── gather context
├── merge upstream_slush ── chain data
├── perform(kwargs) ── your logic
└── extract data_slush ── pass to next
Every agent inherits execute(), which handles the full lifecycle.
Override perform() with your logic.
Data flows between agents through data_slush — structured context that "sloshes" through the pipeline.
No orchestrator needed. Agents chain naturally.
Starts as a rappter. Becomes more of yourself over time.
Open source. 22 agents, full framework, no limits.
Initial consultation + demo. Adapted to YOUR use case.
Dedicated support team. Continuous optimization.
Hardware + software yours. Complete data sovereignty.
A Digital Twin is an AI that learns your preferences, communication style, knowledge, and routines. Over time it becomes a reflection of you — capable of drafting emails in your voice, remembering context from months ago, and automating tasks the way you'd do them.
We learn about your workflow, tools, and goals. Then we configure your Digital Twin with the right agents, memory structure, and integrations. You get a live demo before committing to anything.
Yes. Everything runs locally on your machine. No data is sent to any cloud service. If you use local models (Ollama), nothing ever leaves your device. Even with cloud LLM providers, your memory stays on-device.
You keep everything. The software, the data, the agents — it's all on your machine. The Community edition is free and open source forever. Paid tiers add support and customization, not lock-in.
Absolutely. Download the free Community edition and use it as long as you want. The paid tiers are for people who want a fully customized, professionally configured Digital Twin.
Choose your path.
One command. Full framework. Start building.
Stay in the loop 🦖
Subscribe to updates →