Neo emerged from stealth with $100 million in seed and Series A funding, led by Andreessen Horowitz and Bessemer Venture Partners with participation from Craft Ventures and Merlin Ventures. The company is focused on securing agentic software as businesses adopt applications that can reason, invoke tools, and execute workflows with a user's permissions.
Former SentinelOne president and COO Nick Warner is Neo's chief executive, alongside product and engineering leaders with backgrounds in detection engineering, vulnerability research, and enterprise software. The company has Boston and Tel Aviv operations.
Neo describes itself as a real-time control layer that inventories agentic software, evaluates access and configuration risk, attributes actions to people and agents, and enforces policy over tool calls, APIs, and data movement. The operating premise is that an approved application can still create new risk when autonomous features inherit broad permissions and chain actions across systems. Its product maintains visibility into agents, models, extensions, and MCP servers before risky activity occurs.
The funding reflects investor conviction that agent adoption is moving faster than traditional security architecture. Markets Insider cited a Gartner estimate that agentic capabilities could reach 40% by the end of 2026, up from 5% at the end of 2025. SiliconANGLE frames Neo's opportunity as a control layer for enterprise AI agents.
TL;DR
Neo raised $100 million to build security controls for enterprise AI agents and agentic software. Its platform combines inventory, risk assessment, attribution, and policy enforcement for autonomous capabilities operating inside approved applications. The larger signal is that agent adoption is creating a security-visibility gap that established controls were not designed to handle.
What this means for your operation
Metrotechs analysis: The decision is not whether every business needs Neo. It is whether your organization can identify and govern autonomous capabilities already entering the software stack.
If approved vendors are adding agentic features, security teams need to know which agents exist, what data and tools they can reach, whose permissions they inherit, and whether actions can be traced back to a person, agent, or application. A failure to answer those questions is a control gap regardless of which product eventually addresses it.
Questions to establish the baseline:
- Inventory clarity: Can current tools produce a reliable catalog of agents, agentic extensions, and embedded autonomous features?
- Attribution and audit: Can an API call, data access, or tool invocation be traced to the originating person and agent?
- Policy enforcement: Do controls address autonomous behavior, or do they assume that every action begins with a human?
- Vendor roadmap: Which approved software providers are adding agents, and how will those features change existing permissions and data access?
What remains unknown: Neo has just emerged from stealth, so the cited coverage does not establish broad production adoption, integration effort, performance, or independent customer outcomes. The Gartner estimate indicates market direction, not a universal deadline for every company.
Next step: Inventory agentic capabilities in approved applications and document what current security controls can observe, attribute, and block. That evidence will show whether agent-control tooling is a near-term requirement or a category to monitor.