# Canyon Road

> Execution-Layer Security for AI workloads. Control what AI can access, run, and connect to — at runtime. Supervised copilots on endpoints. Unsupervised agents everywhere else. One control plane.

Canyon Road builds runtime security for AI agents and copilots. The product family is Beacon (supervised AI on endpoints), AgentSH (unsupervised agents in CI, containers, pipelines, and dev environments), and Watchtower (centralized policy and audit control plane).

## Pages

- [Home](https://www.canyonroad.ai/index.md): Overview of Canyon Road's execution-layer security platform.
- [How it works](https://www.canyonroad.ai/how-it-works/index.md): How Canyon Road enforces policy at runtime — allow, prompt, block, redirect.
- [Execution-Layer Security](https://www.canyonroad.ai/execution-layer-security/index.md): What execution-layer security is and why it matters for AI workloads.
- [Use cases](https://www.canyonroad.ai/use-cases/index.md): Concrete enterprise scenarios for AI runtime control.
- [FAQ](https://www.canyonroad.ai/faq/index.md): Common questions about Canyon Road, Beacon, AgentSH, and Watchtower.
- [Contact](https://www.canyonroad.ai/contact/index.md): How to reach Canyon Road.

## Products

- [agentsh](https://www.canyonroad.ai/products/agentsh/index.md): Open-source runtime for AI agent security in CI, containers, pipelines, and dev environments.
- [Beacon](https://www.canyonroad.ai/products/beacon/index.md): AI endpoint visibility and control for supervised copilots.
- [Watchtower](https://www.canyonroad.ai/products/watchtower/index.md): Enterprise control plane for AgentSH and Beacon — policy, audit, kill switch.

## Reference

- [Sitemap](https://www.canyonroad.ai/sitemap.md): Full machine-readable site map.
- [AGENTS.md](https://www.canyonroad.ai/AGENTS.md): Guidance for AI agents accessing this site.