Open Source · MIT License

🦐 PicoClaw
Ultra-Lightweight AI Assistant

$10 Hardware · <10MB RAM · 1s Boot · Written in Go

The world's most efficient personal AI assistant. Run your own AI agent on a $10 RISC-V board with under 10MB of memory. 99% smaller than OpenClaw, 400x faster startup, truly portable across RISC-V, ARM64, and x86.

$10 Hardware Cost
<10MB Memory Usage
<1s Boot Time
400× Faster Startup
Get Started View on GitHub
Terminal 
$ picoclaw onboard
# ✅ Config initialized
$ picoclaw agent -m "What is 2+2?"
# 🦐 PicoClaw: The answer is 4!
Why PicoClaw?

Built for the Edge, Powered by AI

PicoClaw brings the full power of AI assistants to minimal hardware. No cloud dependency, no expensive servers — just a tiny binary that runs anywhere.

Ultra-Lightweight

<10MB memory footprint — 99% smaller than OpenClaw. The smallest full-featured AI assistant ever built.

<10MB RAM Usage

Minimal Cost

Runs on hardware as cheap as $10 — that's 98% cheaper than a Mac Mini. AI for everyone.

$10 Hardware Cost

Lightning Fast

400× faster startup time. Boot in under 1 second even on a 0.6GHz single-core processor.

<1s Boot Time

True Portability

Single self-contained binary across RISC-V, ARM64, and x86. One build, deploy anywhere.

3 Architectures

AI-Bootstrapped

95% agent-generated core with human-in-the-loop refinement. The AI built itself.

95% AI-Generated

Open Source

MIT licensed, community-driven. Clean, readable codebase. PRs welcome!

MIT License
Benchmark

How PicoClaw Compares

See how PicoClaw stacks up against other AI assistant frameworks in terms of resource usage, speed, and cost.

Comparison of AI Assistant Frameworks: PicoClaw vs OpenClaw vs NanoBot
Metric OpenClaw NanoBot PicoClaw 🦐
Language TypeScript Python Go
RAM Usage >1GB >100MB <10MB
Startup Time (0.8GHz core) >500s >30s <1s
Min Hardware Cost $599 (Mac Mini) ~$50 (Linux SBC) $10 (Any Linux Board)
Architecture Support x86 only x86, ARM RISC-V, ARM64, x86
Single Binary
PicoClaw performance comparison chart showing 99% less RAM usage and 400x faster startup compared to alternatives
In Action

See PicoClaw In Action

PicoClaw handles standard AI assistant workflows with lightning speed — from coding to memory management and web search.

PicoClaw Full-Stack Engineering demo showing code development, deployment, and scaling capabilities

Full-Stack Engineer

Develop, deploy, and scale applications directly from your AI assistant. PicoClaw writes code, manages files, and automates workflows.

  • Develop
  • Deploy
  • Scale
PicoClaw Logging and Planning Management demo showing scheduling, automation and memory capabilities

Logging & Planning

Schedule tasks, automate reminders, and maintain long-term memory. PicoClaw remembers your preferences and manages your workflow.

  • Schedule
  • Automate
  • Memory
PicoClaw Web Search and Learning demo showing discovery, insights and trend analysis

Web Search & Learning

Discover information, extract insights, and track trends. Integrated with Brave Search API for real-time web access.

  • Discovery
  • Insights
  • Trends

Real-Time Memory Footprint

Watch PicoClaw run with under 10MB of RAM on a $10 LicheeRV Nano board. That's less memory than a single browser tab uses.

PicoClaw running with under 10MB memory footprint on minimal hardware
Memory usage during operation
LicheeRV Nano - the $10 RISC-V board that can run PicoClaw AI assistant
LicheeRV Nano ($9.9 RISC-V board)
Deploy Anywhere

Supported Hardware

PicoClaw can be deployed on almost any Linux device! From $10 RISC-V boards to powerful servers — pick the right hardware for your use case.

Best Value

LicheeRV Nano

$9.9

E (Ethernet) or W (WiFi6) version. The perfect minimal home assistant platform.

  • RISC-V Architecture
  • WiFi6 or Ethernet
  • Minimal Home Assistant
Buy Now

NanoKVM

$30–$100

Standard or Pro version. Perfect for automated server maintenance and remote management.

  • KVM Functionality
  • Server Maintenance
  • Remote Management
Buy Now

MaixCAM / MaixCAM2

$50–$100

AI camera with built-in compute. Ideal for smart monitoring and computer vision applications.

  • Built-in Camera
  • AI Vision Capable
  • Smart Monitoring
Buy Now
Get Started

Install PicoClaw

Get up and running in under 2 minutes. PicoClaw supports multiple installation methods to fit your workflow.

Download Precompiled Binary

The fastest way to get started. Download the binary for your platform and run it directly.

Terminal 
# Download from GitHub Releases
# Choose your platform: linux-amd64, linux-arm64, linux-riscv64
wget https://github.com/sipeed/picoclaw/releases/latest/download/picoclaw-linux-amd64

# Make it executable
chmod +x picoclaw-linux-amd64

# Initialize
./picoclaw-linux-amd64 onboard

# Start chatting!
./picoclaw-linux-amd64 agent -m "Hello PicoClaw!"
Download Latest Release

Build from Source

Get the latest features and contribute to development. Requires Go 1.21+.

Terminal 
# Clone the repository
git clone https://github.com/sipeed/picoclaw.git
cd picoclaw

# Install dependencies
make deps

# Build for your platform
make build

# Or build for all platforms
make build-all

# Build and install globally
make install

Docker Compose

Run PicoClaw without installing anything locally. Perfect for servers and automated deployments.

Terminal 
# Clone and configure
git clone https://github.com/sipeed/picoclaw.git
cd picoclaw
cp config/config.example.json config/config.json
vim config/config.json  # Set your API keys

# Build and start
docker compose --profile gateway up -d

# Check logs
docker compose logs -f picoclaw-gateway

# Or run one-shot agent
docker compose run --rm picoclaw-agent -m "What is 2+2?"

Quick Configuration

Set your API key in ~/.picoclaw/config.json to get started:

~/.picoclaw/config.json 
{
  "agents": {
    "defaults": {
      "model": "glm-4.7",
      "max_tokens": 8192
    }
  },
  "providers": {
    "openrouter": {
      "api_key": "YOUR_API_KEY"
    }
  }
}

Supported LLM Providers

OpenRouter Zhipu Anthropic OpenAI Gemini Groq
Multi-Platform

Connect Your Favorite Chat App

Talk to PicoClaw through the messaging platforms you already use. Simple setup, powerful results.

Telegram

Easy Setup

Recommended. Just create a bot token via @BotFather and configure. Supports voice messages with Groq Whisper transcription.

Discord

Easy Setup

Create a Discord bot, enable Message Content Intent, and add to your server. Full chat support with allow-list security.

QQ

Easy Setup

Connect via QQ Open Platform with AppID and AppSecret. Optional user allow-list for access control.

DingTalk

Medium Setup

Create an internal app on DingTalk Open Platform. Uses Client ID and Client Secret for authentication.

Reference

CLI Commands

PicoClaw's command-line interface is simple and intuitive. Here are the most common commands.

picoclaw onboard Initialize config & workspace
picoclaw agent -m "..." Chat with the agent (one-shot)
picoclaw agent Interactive chat mode
picoclaw gateway Start the gateway for chat apps
picoclaw status Show current status
picoclaw cron list List all scheduled jobs
picoclaw cron add ... Add a scheduled job or reminder
FAQ

Frequently Asked Questions

What hardware does PicoClaw run on?

PicoClaw runs on any Linux device with as little as 10MB of RAM. This includes $10 RISC-V boards like the LicheeRV Nano, ARM64 devices, and standard x86 machines. It supports RISC-V, ARM64, and x86 architectures natively with a single self-contained binary.

How much memory does PicoClaw use?

PicoClaw uses less than 10MB of RAM — that's 99% less memory than OpenClaw (which requires over 1GB) and significantly less than NanoBot (100MB+). This makes it possible to run on the cheapest Linux boards available.

Is PicoClaw free and open source?

Yes! PicoClaw is completely free and open source under the MIT license. You can find the source code on GitHub at github.com/sipeed/picoclaw. PRs are welcome!

What chat platforms does PicoClaw support?

PicoClaw supports Telegram (recommended), Discord, QQ, DingTalk, and more chat platforms. You can interact with your AI assistant through any of these platforms. Telegram also supports voice messages with Whisper transcription via Groq.

How fast does PicoClaw boot?

PicoClaw boots in under 1 second, even on a 0.6GHz single-core processor. That's 400× faster than OpenClaw (500s+) and 30× faster than NanoBot (30s+). Thanks to Go's native compilation and zero runtime dependencies.

Which LLM providers are supported?

PicoClaw supports multiple LLM providers including OpenRouter (access to all models), Zhipu (best for Chinese users), Anthropic (Claude), OpenAI (GPT), Google Gemini, DeepSeek, and Groq (also provides free voice transcription). OpenRouter is recommended as it gives access to multiple models.

Can PicoClaw schedule tasks and reminders?

Yes! PicoClaw supports scheduled reminders and recurring tasks through the cron tool. You can set one-time reminders ("Remind me in 10 minutes"), recurring tasks ("Remind me every 2 hours"), and cron-based schedules ("Remind me at 9am daily").

How was PicoClaw built?

PicoClaw was inspired by NanoBot and refactored from the ground up in Go through a self-bootstrapping process. The AI agent itself drove the entire architectural migration and code optimization — 95% of the core code was generated by AI with human-in-the-loop refinement.

Ready to Run AI on $10 Hardware?

Join the community of developers running personal AI assistants on minimal hardware. Open source, MIT licensed, and ready to deploy.

Star on GitHub Join Discord