SentinelOne Bets Big On AI Security At OneCon 2025

According to CRN, SentinelOne unveiled a sweeping set of AI security products at its OneCon 2025 conference in Las Vegas this week, building on its $180 million acquisition of Prompt Security from August and its $225 million Observo AI deal from September. The company launched three generally available AI security products for employees, code assistants, and custom applications, plus a beta product for agentic AI governance. They also debuted Wayfinder managed services in partnership with Google Cloud and expanded Purple AI with agentic auto-investigations now in preview. CEO Tomer Weingarten believes many traditional cybersecurity tools will become “irrelevant” as AI agents reshape the security landscape.

The AI security acquisition spree

SentinelOne isn’t just talking about AI security—they’re buying their way into the market. The $180 million Prompt Security acquisition gives them immediate capabilities across 15,000 AI services, which is frankly impressive coverage. But here’s the thing: buying startups is one thing, integrating them seamlessly is another. The fact that they already have three products generally available suggests they moved fast on integration. The beta product for agentic AI using Model Context Protocol is particularly interesting because MCP is becoming the standard way AI agents discover tools. Basically, they’re positioning themselves at the infrastructure level where AI agents operate, which is smarter than just bolting on another security layer.

Where agentic security gets real

The Purple AI expansions are where this gets technically fascinating. “In-line Agentic Auto-investigations with dynamic reasoning” sounds like marketing speak, but what it means is AI that can actually reason through security incidents from start to finish. We’re talking about systems that don’t just flag threats but understand context, assess impact, recommend responses, and even create detection rules autonomously. And the integration with Singularity Hyperautomation suggests they’re building a closed-loop system where detection leads directly to response. The challenge? Trust. How comfortable will security teams be letting AI make these decisions without human oversight? It’s a massive cultural shift that might be harder than the technology itself.

The managed services play

Wayfinder represents a significant pivot toward services—combining their AI with “elite-level” human experts and Google’s threat intelligence. This is smart because pure automation has limits in cybersecurity. The four-tier approach shows they understand different customers need different levels of hand-holding. But look, the real story here might be the Google Cloud partnership. SentinelOne gets access to Google’s massive threat data, which makes their AI smarter, while Google gets a serious enterprise security partner. It’s a win-win that could pressure other cloud providers to make similar moves. The question is whether this services approach will cannibalize their product sales or create new revenue streams.

The bigger platform picture

Weingarten’s comments about making traditional security tools “irrelevant” reveal the grand ambition here. SentinelOne wants to be the central nervous system for security operations, not just another point solution. The open-source Purple AI MCP Server on GitHub is particularly strategic—it encourages developers to build on their platform rather than competitors’. But can they actually deliver on the promise of real-time, autonomous security? The technology is clearly advancing, but we’ve heard similar promises before. The difference now is that agentic AI might actually be capable enough to make it happen. Still, replacing a decade’s worth of security infrastructure won’t happen overnight, no matter how good the AI gets.