Panguard AI, Inc.
Security is a right, not a luxury. Panguard AI builds the missing gatekeeper for AI agents, providing a skills audit and runtime guard to protect against agentic threats.
At a Glance
- Enterprise Security Teams
- AI Agent Developers
- Regulated Industries (Banking, Govt)
- Sovereign AI Initiatives
AI Tools by Panguard AI, Inc.
(1)
PanGuard AI
AI Agent Security Firewall
Discussions
No discussions yet
Be the first to start a discussion about Panguard AI, Inc.
Latest News
Palo Alto Unit 42 Confirms Panguard Research on Malicious Agent Skills
Panguard AI Releases ATR Detection Benchmarks: 97.2% Garak Recall
Launch of MCP Server Scanning: Detecting Threats Before Installation
OWASP Agentic Top 10 Coverage: ATR Ships Full Mapping for All Categories
Products & Services
A command-line interface for scanning, auditing, and guarding AI agents at runtime.
An open, executable standard (MIT licensed) for detecting AI agent threats across 10 categories including prompt injection and tool poisoning.
A collective immunity network that shares new threat rules across all Panguard users to protect against emerging attacks.
An enforcement engine that monitors agent behavior 24/7 and contains hijack attempts.
Market Position
Panguard focuses specifically on the 'Skills Audit' and 'Runtime Guard' layer for AI agents, addressing a gap left by traditional cloud and endpoint security companies like Palo Alto Networks.
Leadership
Founders
Adam Lin (林冠辛)
Solo founder of Panguard AI. Previously a Senior Software Engineer at Verizon. 27-year-old Taiwanese developer who built the Agent Threat Rules (ATR) standard.
Executive Team
Adam Lin
Solo Founder
Senior Software Engineer with experience in security and agentic AI. Creator of the ATR standard.
Founding Story
Started by Adam Lin after observing that AI agents were being given full system access (file read, code execution) without any review process. Panguard was created to be the 'missing review layer' for AI agents, similar to the App Store or NPM audits.
Business Model
Revenue Model
Open core/Freemium model with paid tiers for enterprise compliance evidence, sovereign deployments, and OEM licensing.
Pricing Tiers
Forever free, MIT licensed. Includes 768 ATR rules and pga CLI for self-hosting.
One-time fee for 90 days. Founder-led delivery aimed at proving security during enterprise/bank reviews.
Includes signed audit-ready evidence, airgap support, and per-nation deployments.
Custom tier for large-scale government or sovereign AI deployments.
Target Markets
- Enterprise Security Teams
- AI Agent Developers
- Regulated Industries (Banking, Govt)
- Sovereign AI Initiatives
- Securing AI agents in enterprise environments
- Auditing third-party MCP skills and tools
- Protecting against prompt injection and tool poisoning
- Providing audit-ready evidence for AI Act compliance