Claroty Claire: The First CPS-Native AI Security Agent for Critical Infrastructure

Claroty Claire Brings a CPS-Native AI Security Agent to Critical Infrastructure

Here's the thing about defending critical infrastructure: the machines we worry about most are rarely just computers. They're insulin pumps, water pumps, turbine controllers, and robotic arms on factory floors — cyber-physical systems where a security decision can affect something moving in the real world. That's exactly the gap Claroty is aiming at with Claire, introduced on May 28, 2026, and described as the industry's first CPS-native AI security agent. As a defender, I find that framing genuinely encouraging, and I want to walk you through why.

What "CPS-Native" Actually Means

Plenty of AI security tools were born in the world of laptops, servers, and cloud apps, then stretched to cover industrial and medical gear. Claire flips that order. It's purpose-built for cyber-physical systems from the ground up, designed to help organizations proactively defend mission-critical infrastructure across industrial, healthcare, and operational technology (OT) environments.

Being CPS-native matters because these environments speak a different language than IT. A patch that's routine in an office network might be forbidden on a medical device until the manufacturer approves it. Uptime and physical safety aren't nice-to-haves — they're the whole point. An AI agent that understands that distinction natively is far more useful than one bolted on after the fact.

The Knowledge Base Behind the Agent

What makes me optimistic about Claire isn't the "AI agent" label — it's what's underneath it. The agent is powered by a CPS-specific language model trained on what Claroty calls the largest CPS data lake, paired with more than a decade of accumulated domain expertise.

Trained on the Real Industrial and Medical World

The numbers here are the part worth sitting with. The underlying model draws on data from over 6,500 unique OEMs and medical device manufacturers. Its knowledge base spans deployments across more than 20,000 sites, 50-plus sectors, and 60-plus countries. And it's backed by threat research from Team82, Claroty's research group, with the broader platform serving 1,300-plus customers including 24 of the Fortune 100.

In plain terms: a defender using Claire isn't reasoning from generic security theory. The agent's answers are grounded in how these specific devices, in these specific sectors, behave in the field. For cyber-physical systems security, that real-world grounding is the difference between advice you can act on and advice you have to second-guess.

How Claire Helps Defenders Day to Day

I always like to translate capabilities into "what does this actually do on a Tuesday morning." Claire focuses on three constructive jobs.

Minimizing the Attack Surface

The agent helps teams shrink their exposure proactively — finding and prioritizing the soft spots before an adversary does. Reducing attack surface is unglamorous, foundational defensive work, and it's precisely the kind of task where an AI agent that knows your device landscape can save human operators real hours.

Protecting Safety and Uptime

Because Claire is CPS-native, it's built to maintain safety and uptime rather than treat them as obstacles. That's a meaningful design choice. In OT and healthcare security, the safest action sometimes isn't the most aggressive one, and an agent that respects that constraint earns trust faster.

Continuous Compliance Without the Spreadsheet Grind

This may be my favorite piece. Claire supports continuous compliance through automated asset mapping — lining up your inventory against regulatory frameworks and OEM-approved patch levels. Anyone who has assembled a compliance report by hand knows how much time that automated mapping gives back to a team.

Keeping Humans in the Driver's Seat

What I appreciate most is the philosophy. Claroty CEO Yaniv Vardi framed Claire as empowering human operators to make confident decisions, supported by trustworthy, tailored insights and agentic actions. That's the right north star. The goal isn't to remove the defender — it's to hand them a well-informed partner so the human stays in command with better information.

For the people protecting hospitals, factories, and utilities, a CPS-native AI security agent that prioritizes safety, trims attack surface, and automates the compliance slog is a genuinely constructive step forward. Defense is a craft, and tools like this one let skilled people spend their attention where it counts. I'll be watching how it lands in the field — but as launches go, this is the kind I like to write about.

Sources: Help Net Security — Claroty launches Claire, May 29, 2026; Industrial Cyber — Claroty launches Claire, a CPS-native AI security agent, May 28, 2026; PRNewswire — Claroty Introduces Claire, May 28, 2026