Neo Security

Neo Security: a control plane for the software that acts on its own

Neo (neo.ai) is an enterprise security platform for agentic software: it discovers, monitors, and controls the AI agents, MCP servers, browser extensions, and AI-enabled applications running across an organization's endpoints. Built by SentinelOne veterans — CEO Nick Warner built SentinelOne's go-to-market organization and took the company public as COO in 2021, co-founder Shlomi Salem led its detection engineering for over a decade, and Eran Shirazi co-founded EasySend after leading vulnerability research in IDF Unit 8200 — Neo emerged from stealth on July 20, 2026 with $100M in launch funding led by Andreessen Horowitz and Bessemer Venture Partners.

The premise: EDR was built for malware, not for sanctioned software that autonomously reads data and takes actions. Gartner projects agentic capabilities in enterprise apps jumping from 5% in 2025 to 40% by the end of 2026, and security teams currently have no inventory of it, let alone control.

Key Features

• Live inventory of agentic and traditional software across every endpoint
• Posture analysis that flags misconfigurations and excessive permissions
• Runtime enforcement with granular, LLM-powered policy controls
• Real-time attribution of every action to a human or agent actor
• Endpoint-level interception — allow, block, or hold actions for human approval
• On-device analysis with no kernel module, no reboot, and no network data egress; SOC 2 Type II

Ideal Use Case

A security operations team at a company where employees are installing AI agents, IDE assistants, and MCP servers faster than IT can catalog them — and where the CISO needs to answer "what agentic software runs here, what can it touch, and who is accountable for each action" before an incident forces the question.

How Neo Security differentiates

Neo positions as a control layer alongside EDRs like CrowdStrike and SentinelOne rather than a replacement — the focus is sanctioned-but-ungoverned agentic software, not malware. Its attribution model distinguishes human from agent actors per action, and deployment avoids kernel modules and reboots entirely. Production credibility: $100M at launch from a16z, Bessemer, Craft Ventures, and Merlin Ventures — one of 2026's largest security launches — with a founding team drawn from SentinelOne, Wiz, and Palo Alto Networks.

FAQ

What does Neo Security actually monitor? AI agents, MCP servers, browser extensions, and AI-enabled applications on endpoints — including what they access and what actions they take.

Does Neo replace an EDR like CrowdStrike? No. It sits alongside EDR as a dedicated control plane for agentic software, a category classic endpoint tools were not designed to govern.

How is Neo deployed? On-device analysis with no kernel module and no reboot; the company states data never leaves the endpoint.

How much does Neo cost? No public pricing — it is an enterprise sales motion with demos through neo.ai.

tl;dr

Neo Security is the SentinelOne-alumni answer to shadow AI: full inventory, posture intelligence, and runtime control over every AI agent and MCP server on enterprise endpoints, launched with $100M from a16z and Bessemer.

Related

Looking for more options? Browse the Security & Governance directory or read our best AI security tools listicle. Neo Security is also tracked on Crunchbase.