PicoClaw

PicoClaw is an MIT-licensed personal AI assistant written in Go for small computers and embedded-style deployments. Its official site emphasizes a single-binary design, less than 10 MB of memory in the project’s published measurements, and startup under one second on the cited test class. Release assets cover several Linux architectures and package formats, macOS, Windows, FreeBSD, and Android; the project also documents Raspberry Pi, RISC-V, ARM, x86_64, Termux, Docker, and cloud-server use. These resource numbers are project-reported targets rather than independent benchmarks.

The software is free to download and self-host. Users provide access to a supported hosted model API or a compatible local endpoint, so inference charges depend on that provider. PicoClaw connects an assistant to chat services and includes local configuration, memory, scheduled commands, tools, and gateway or API-oriented operation. Channel and provider availability changes between releases, so deployments should be checked against the selected build and platform.

Top features

• Compact Go deployment: a compiled program targets low-memory machines without requiring a large application stack.
• Broad architecture coverage: official materials list x86_64, multiple ARM variants, RISC-V, MIPS, Android, Docker, and common small-board environments.
• Messaging channels: integrations include Telegram, Discord, Slack, DingTalk, Feishu, WeCom, LINE, QQ, and additional services.
• Model-provider choice: connects to multiple commercial APIs, OpenAI-compatible endpoints, and local Ollama-style deployments.
• Assistant functions: memory, web-oriented tools, scheduled commands, cron automation, and planning support recurring personal workflows.
• Gateway and API use: the project advertises backend operation for integrations through its gateway, REST, and related protocol support.
• Published binaries: GitHub releases provide checksums and platform-specific archives or packages, including Android bundles.

Use cases

• Run a chat-connected assistant on a Raspberry Pi, RISC-V board, home server, or inexpensive Linux machine.
• Deploy a local assistant backend where memory and startup overhead are constrained.
• Connect scheduled reminders, summaries, or monitoring tasks to an existing messaging account.
• Experiment with hosted and local model providers from the same small gateway.
• Package an assistant for Android or field hardware without operating a full desktop stack.