Neo launches with $100 million for enterprise AI agent security
Neo has emerged from stealth with $100 million across seed and Series A funding to build a security control layer for enterprise AI agents and AI-enabled software. The company says its platform discovers agents, models, extensions and MCP servers, assesses privileges and configurations, attributes actions to people or software identities, and enforces policies over tool calls, workflows, data movement and API access. SecurityWeek and Globes independently reported the financing and product launch. The funding is confirmed, but the available sources do not include independent product testing, supported-integration lists, performance results or false-positive rates.
What Neo launched
Neo has emerged from stealth with $100 million across seed and Series A funding. The company says Andreessen Horowitz and Bessemer Venture Partners backed the rounds, with Craft Ventures and Merlin Ventures participating. Neo plans to use the capital to expand engineering and go-to-market teams.
Neo describes its product as a control layer for AI agents, AI-enabled applications and conventional enterprise software gaining agentic capabilities. SecurityWeek independently reported the launch and financing, while Globes reported the company's Boston headquarters and Tel Aviv development center.
The reported security controls
According to Neo and SecurityWeek, the platform combines four functions:
- •Continuous discovery of agents, models, extensions and MCP servers.
- •Posture assessment for excessive privileges and configuration weaknesses.
- •Attribution connecting an action to a user, agent or application identity.
- •Policy enforcement over tool calls, workflows, data movement and API access.
That scope reflects a real change in enterprise security. An agent can act through valid user permissions, invoke tools and move data without resembling a conventional standalone application. Inventory alone may therefore be insufficient if a security team cannot reconstruct which identity initiated a consequential action or stop a tool call that exceeded its intended boundary.
| Confirmed by retrieved sources | Not yet independently established |
|---|---|
| $100 million across seed and Series A rounds | Detection coverage across major agent frameworks |
| Launch investors and named founders | False-positive and false-negative rates |
| Product positioning around inventory, attribution and policy | Enforcement latency and failure behavior |
| Planned engineering and go-to-market expansion | Independent production benchmarks |
LDS assessment
The interesting technical claim is the combination of identity-to-action attribution with runtime enforcement. Controls applied only at a user interface can miss direct API invocation or downstream tool execution, so enterprise buyers should examine where Neo's policies are enforced and whether logs remain complete across multi-step workflows.
Evaluation should focus on identity-provider and agent-framework coverage, delegated credentials, audit-log quality, policy semantics, deployment architecture and what happens when the control service is unavailable. The launch establishes funding and product intent; it does not yet prove operational effectiveness.
Key Points
- 1Neo raised $100 million to build security controls for enterprise AI agents and AI-enabled software.
- 2The reported platform combines discovery, posture assessment, action attribution and runtime policy enforcement.
- 3Independent evidence of integration coverage, performance and false-positive rates is not yet available.
Scoring Rationale
A $100 million launch is a notable funding event in the emerging market for securing enterprise AI agents and their runtime actions. The product scope is directly relevant to security and platform teams, although independent technical validation is not yet available.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
