What Canyon Road does

Canyon Road Technologies Inc. is a privately held, early-stage US software company founded in 2026 that builds runtime security software for AI agents and AI workloads. The company has 1–10 employees and operates from canyonroad.ai, distributing its core runtime, AgentSH, as a free open-source single binary on GitHub.

The company's product platform is organized around what it terms "Execution-Layer Security" — enforcement at the operating-system syscall layer (using eBPF/LSM on Linux, ESF/FUSE-T on macOS, and minifilter on Windows) rather than at the prompt or proxy layer. The platform comprises three products: AgentSH, an open-source runtime that intercepts file, network, and process operations of headless AI agents in CI/CD, containers, and dev environments; Beacon, a paid endpoint agent that monitors and controls supervised desktop AI tools (Claude, Cursor, ChatGPT, Claude Code) on employee macOS and Windows machines with human-in-the-loop approvals; and Watchtower, an enterprise control plane providing centralized policy distribution, approval routing to Slack/email/SMS, SIEM export to Splunk/Sentinel/QRadar, OpenTelemetry forwarding, SSO/RBAC, and a fleet-wide emergency kill switch. Policies are defined as code (YAML/JSON), versioned in git, and tested in CI. Adjacent capabilities include MCP (Model Context Protocol) security with version/hash pinning and cross-server attack-pattern detection, a Secrets Manager with third-party vault integration, and @agentsh/secure-sandbox — a TypeScript SDK with adapters for Vercel AI SDK, E2B, Daytona, Cloudflare Workers, and Blaxel.

The business model is a freemium-to-enterprise flywheel: AgentSH is free and open-source (no direct revenue); Beacon offers a Free tier for individuals, a Team tier for up to 5 endpoints, and an Organization tier with SIEM integration; Watchtower offers Team and Enterprise tiers with multi-policy groups, configurable retention, SSO/RBAC, and dedicated support. GTM combines product-led growth through GitHub and npm with enterprise field sales for Watchtower and higher Beacon tiers, supplemented by content marketing (founder-authored blog, agentsh.org documentation) and category-education efforts around "Execution-Layer Security." SOC 2 Type 2 and ISO 27001 compliance alignment is self-attested but certification status is not confirmed. No funding rounds, named customers, revenue figures, or geographic operating footprint are disclosed.

Name

Canyon Road

Legal name

Canyon Road Technologies Inc.

Website

https://canyonroad.ai

Company type

Private

Founded year

2026

Operating status

Operating

Headcount range

1–10 employees

Short description

Canyon Road builds runtime security software for AI agents and AI workloads, enforcing least-privilege policy at the operating-system syscall layer. Its three-product platform — open-source AgentSH, paid Beacon endpoint security, and Watchtower enterprise control plane — serves security teams, DevOps/platform engineers, and compliance organizations deploying supervised and unsupervised AI.

Ownership category

akta.pro rank

Product category

AI Runtime Security Software

NAICS

Security Systems Services (except Locksmiths) (561621), Security Systems Services (56162)

SIC

Services-Computer Programming Services (7371)

akta.pro primary industry

Model Monitoring, Drift & Behavioral Anomaly Detection (HDAAAKAF)

akta.pro secondary industries

Responsible AI, Security & Privacy Platforms (Safety, Guardrails, PII) (HDAEANAG), Responsible AI, AI Governance & Compliance Services (BPAEAHAJ)

Keywords

• AI agent runtime security
• Execution-layer security
• Endpoint AI governance
• Syscall-level policy enforcement
• AI workload monitoring

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Personnel, Marketing or Sales, Operations

Revenue model

1. AgentSH Open Source: Free, open-source runtime security for AI agents. Available as a single binary on GitHub. No direct revenue; drives awareness and adoption that converts to paid enterprise products.
2. Beacon Endpoint Security (Paid Tiers): Beacon offers Free tier (1 endpoint), Team tier (up to 5 endpoints), and Organization tier (unlimited endpoints with advanced features). Revenue from endpoint security subscriptions.
3. Watchtower Enterprise Control Plane: Enterprise subscription with Team and Enterprise tiers. Enterprise includes multi-policy groups, 30/90/365-day retention, SIEM integration, SSO/RBAC, and dedicated support. Revenue from fleet-wide security management.
4. Enterprise Onboarding Services: Enterprise onboarding and deployment support mentioned as a service offering for Beacon for endpoints and Watchtower fleet management. Professional services revenue stream.

Pricing tiers

Go-to-market motion2 records

Distribution channels5 records

Marketing channels5 records

Core offering

Canyon Road provides Execution-Layer Security for AI workloads through a three-product platform. AgentSH is an open-source runtime that intercepts system calls (files, network, processes, secrets) at the kernel level to enforce policy on AI agents and copilots before actions execute. Beacon monitors and controls supervised AI tools (Claude, Cursor, ChatGPT) on employee endpoints (macOS/Windows) with per-app visibility and human-in-the-loop approvals. Watchtower is the enterprise control plane providing centralized policy management, approval routing, SIEM export, and a fleet-wide kill switch for both Beacon and AgentSH deployments.

Product overview

Canyon Road offers Execution-Layer Security for AI workloads through a three-product platform. The core portfolio consists of: (1) AgentSH — an open-source runtime that secures unsupervised headless AI agents in CI, containers, and dev environments by intercepting system calls and enforcing least-privilege policy at execution time; (2) Beacon — an endpoint agent that monitors and controls supervised AI tools (Claude, Cursor, ChatGPT, Claude Code) on employee macOS and Windows machines with per-app visibility, runtime policy, and human-in-the-loop approvals; and (3) Watchtower — the enterprise control plane that governs both Beacon and AgentSH from one place, distributing policies, routing approvals, exporting to SIEM, and providing a fleet-wide emergency kill switch. The three products share a unified policy engine and single audit trail, with policies defined as code (YAML/JSON) and enforcement happening locally at execution time while governance flows from the central Watchtower plane.

Differentiator

Problem solved

Functional benefit

Products and services

• AgentSH
• Beacon
• Watchtower
• @agentsh/secure-sandbox

Quantifiable outcome

• Fleet halted in 1.2 seconds across 847 agents
• +2 more outcomes

Segments4 records

Ideal customer profiles4 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration7 records

AI capability5 records

Feature9 records

Scale indicators3 records

Recent moves5 records

Expansion highlights5 records

Direct peers

• Astrix Security: Astrix Security focuses on securing AI agents and non-human identities, providing runtime controls for third-party AI integrations and API-connected automations. Direct competitor in AI agent runtime security.
• Lakera: Lakera provides AI security for LLM-powered applications, including prompt injection defense and runtime guardrails for agentic systems. Comparable as a runtime AI security platform for enterprises deploying LLM/agent workloads.
• HiddenLayer: HiddenLayer offers an AI security platform covering model protection, runtime monitoring, and adversarial defense for ML and AI systems. Comparable scope in AI/ML security and behavioral monitoring.
• Noma Security: Noma Security delivers AI security and lifecycle management for enterprise AI applications, including agent governance and runtime policy enforcement. Direct competitor targeting enterprise AI governance buyers.
• PromptArmor: PromptArmor provides runtime AI security specifically designed for AI agents and LLM applications, including prompt injection prevention and tool-call monitoring. Highly comparable niche focus on agent runtime defense.

Emerging players

• Pillar Security: Pillar Security provides AI security posture management and runtime controls for enterprise AI applications and agents. Emerging player with overlapping scope in AI runtime protection.

Broad incumbents

• Cloudflare: Cloudflare is a major cloud security and edge platform with emerging AI workload controls, including bot management and Workers AI safeguards. Broad incumbent that could bundle AI agent security into its existing platform.
• CrowdStrike: CrowdStrike is the EDR/XDR market leader with an extensive endpoint footprint and is actively adding AI workload protection capabilities. Broad incumbent that could subsume AI agent endpoint security into Falcon.
• Microsoft Defender for Cloud / Security Copilot: Microsoft Defender and Security Copilot provide cloud workload protection and AI security features bundled with Azure and the Microsoft enterprise stack. Broad incumbent with deep enterprise distribution.
• SentinelOne: SentinelOne is an endpoint and cloud security platform expanding into AI workload monitoring and agentic security. Broad incumbent with overlapping endpoint AI visibility use cases to Beacon.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

• GitHubgithub.com/canyonroad/agentsh
• LinkedInlinkedin.com/company/canyon-road

Compliance2 records

Revenue estimate

Valuation estimate

Number of profiles

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Canyon Road