Neo raises $100M to put a control layer over enterprise agents
A nine-figure raise purely to secure and govern agent fleets shows agent security is now its own venture category. GaaS deployers have to budget for it.
- Neo launched from stealth with $100 million in combined seed and Series A, led by Andreessen Horowitz and Bessemer, with Craft and Merlin Ventures.
- The company builds a real-time control layer giving SecOps inventory, posture intelligence, attribution, and policy over AI agents, apps, browsers, and identities.
- Neo cites Gartner's projection that agentic apps rise from about 5 percent of enterprise apps in 2025 to 40 percent by end of 2026, expanding the attack surface.
Investors just put $100 million behind the idea that agent fleets are a security problem before they are a productivity one. Neo, founded by veterans of SentinelOne, Wiz, and Palo Alto Networks, launched out of stealth on July 20 with a combined seed and Series A led by Andreessen Horowitz and Bessemer Venture Partners, with Craft Ventures and Merlin Ventures participating. The pedigree is the pitch: a founding team drawn from three of the most successful security exits of the last decade, aimed squarely at the mess enterprises are about to make with agents.
What Neo is building
Neo calls itself an "Agentic Software Control" company. In plainer terms, it is building a real-time control layer that gives SecOps teams four things over their agents: an inventory of what exists, posture intelligence on how exposed each one is, attribution for what did what, and policy control to constrain behavior. Crucially, that scope is not limited to agents. It spans AI-enabled applications, browsers, identities, and traditional software, which is Neo's way of saying the boundary between an agent and everything else it touches is exactly where the risk lives.
The capital funds go-to-market and engineering. You can read the launch details on GlobeNewswire and The Next Web.
The number behind the number
Neo's thesis leans on a Gartner projection it cites: agentic applications climbing from about 5 percent of enterprise apps in 2025 to a projected 40 percent by the end of 2026. If that curve is even roughly right, every enterprise is about to run a fleet of semi-autonomous software with credentials, browser access, and the ability to act, and almost none of them have a way to see or govern it. That gap is the whole market. An eightfold jump in agent share in a year is an eightfold jump in attack surface, and it arrives faster than most security teams can staff for.
The broader point is that securing agents is turning into its own venture category rather than a feature bolted onto existing tools. It rhymes with the capability gaps we covered in the AISI open-weight cyber report: the offensive and deployment side of agents is outrunning the controls meant to contain it, and capital is flowing to close the distance.
What this means for the agent economy
A nine-figure raise to do nothing but inventory, attribute, and police agents tells you the buyer's next line item. For anyone deploying Agentic AI as a Service, the cost of the agent is no longer the whole cost. Governance and control are becoming a mandatory adjacency, the way endpoint security became unavoidable once every employee had a laptop. Neo is betting that the company running fifty agents will pay to know what all fifty are doing, and $100 million says the market agrees.