Agentic AI Security Moves From Afterthought To Enterprise Priority
Agentic AI security and governance refers to the policies, control systems, and monitoring tools that manage autonomous software agents as they reason, call tools, and execute workflows inside enterprise environments, preventing unsafe actions while still allowing these agents to perform meaningful business tasks at scale. For years, enterprises treated AI as contained chatbots. That illusion is gone. Autonomous AI deployment is now reaching into finance, operations, and software development, where agents can act with real permissions and touch mission‑critical data. Without AI agent control infrastructure, these systems behave like highly skilled interns with admin access and no supervision. The key takeaway: agents are not experimental toys anymore; they are production actors, and enterprises that deploy them without a strong control layer are accepting unpriced operational, compliance, and security risk.
Neo: Building The Real-Time Control Layer For Agentic Software
Neo has emerged from stealth with USD 100 million (approx. RM460 million) in funding to build security and control infrastructure for agentic software. That number alone shows how investors now view enterprise AI governance: as a core platform bet, not a side feature. Neo’s thesis is blunt. Enterprise security was built for predictable applications and human users, but agentic software can reason, invoke tools, and execute multistep workflows using valid user permissions in ways traditional controls do not see. So Neo positions itself as an "Agentic Software Control" company, offering a real‑time control layer that inventories every AI agent, plugin, browser extension, and autonomous capability, then monitors what they can access and blocks risky behavior before it happens. The opinionated stance here is correct: without this sort of continuous oversight, enterprises are effectively blind to how AI is reshaping data movement and authority inside their own systems.
Inside Neo’s Governance Stack: Inventory, Risk, and Native Enforcement
Neo’s design makes an important point: agentic AI security is not one feature, but an entire governance stack. Its Neoverse inventory tracks AI agents, applications, plugins, Model Context Protocol servers, and traditional software gaining autonomous capabilities, then layers capability and risk intelligence on top to assess what each tool can do and which systems or data it can reach. This is enterprise AI governance in practice, not theory. Security teams can identify over‑permitted agents and enforce policies on tool calls, API access, data movement, and agentic workflows tailored to roles and identities. Crucially, Neo adds attribution and native enforcement: every action is tied to the initiating human, agent, or application, and the platform can block malicious models or redirect out‑of‑bounds prompts directly, without handing off to legacy tools. In a world where Gartner expects agentic capabilities in 40% of enterprise apps by the end of 2026, up from 5% in 2025, this kind of integrated control layer is not optional—it is the new perimeter.
Arrakis: From Experimental Chatbots To Production-Grade AI Agents
Arrakis has also emerged from stealth, raising USD 38 million (approx. RM175 million) in a few months, including a USD 30 million (approx. RM138 million) Series A led by Blossom Capital. Backers include leaders from OpenAI and Datadog, which signals that autonomous AI deployment is now viewed as infrastructure, not consulting. Arrakis focuses on industrial companies that have experimented with AI chatbots but struggle to deploy AI that can carry out business tasks effectively across aerospace, energy, logistics, manufacturing, construction, and telecommunications. Its model‑agnostic platform and forward‑deployed AI engineers help enterprises design, build, and scale AI agents into key operations, with customers already including NYSE‑listed firms in energy and logistics. This approach acknowledges a hard truth: most enterprises do not need another chat interface; they need accountable agents embedded in workflows, with clear guardrails, auditability, and deployment discipline from day one.
A New Infrastructure Category: AI Agent Control and Governance
Seen together, Neo and Arrakis are clear evidence that AI agent control infrastructure is becoming its own enterprise category. Neo is targeting the security gap with a real‑time control layer that monitors how agentic software behaves and allows organizations to govern its behavior in real time. Arrakis is focused on getting AI agents into production safely at scale for industrial champions that must compete globally. Both bet on the same pain point: enterprises need guardrails and oversight mechanisms before deploying autonomous agents, or they risk untracked data flows and opaque decision chains. Investors are signaling strong confidence that agentic AI governance will sit alongside identity, observability, and network security as foundational infrastructure: "Agentic AI is creating a new security category" and execution will determine who defines this next era. The conclusion is straightforward—if your agents can act, they must also be governed, and you will buy platforms to do it.
