Skip to content

What is OpenClaw?

OpenClaw is an open-source AI agent runtime — the engine that connects a language model to tools, memory, sessions, and channels so an agent can actually do things, not just chat. It is the foundation that Pinchy builds on.

A language model on its own only produces text. A runtime turns it into an agent by wiring in:

  • Tools the model can call to take real actions.
  • Memory and sessions so a conversation has continuity.
  • Channels so the agent can be reached where people already are.
  • A plugin system so new capabilities can be added without forking the core.

OpenClaw is extensible and local-first, which makes it a favorite for individual power users who want a capable, self-hosted agent under their own control.

OpenClaw is built around the single powerful operator. For a team or company, critical pieces are missing:

  • No user management or per-user accounts.
  • No role-based access control over who can use which agent.
  • No audit trail of who did what, when.
  • No governance layer for permissions across many agents and people.

These are not flaws in the runtime — they are simply a different layer, the enterprise layer.

Pinchy is an AI agent platform built on top of OpenClaw. OpenClaw stays the runtime; Pinchy wraps it with multi-user accounts, role-based access control, per-agent permissions, a signed audit trail, and a single self-hosted deployment. You get OpenClaw's runtime power with the governance a team needs around it.

Learn more at the OpenClaw project and the OpenClaw documentation.

Running OpenClaw yourself? The free OpenClaw security self-check walks the common exposure opt-outs (gateway binding, authentication, containment) in six questions. It is a questionnaire, so you never paste a secret or any config, and your answers never leave your browser.