
nRouter Unveils Cortex Air-Gapped ML Sidecar for Zero-Leakage LLM Defense
nRouter introduces Cortex — an air-gapped, zero-database ML sidecar communicating over internal mTLS 1.3 that intercepts prompt injections and redacts PII with zero token spend on blocked requests.
SUNNYVALE, CA, June 18, 2026 — nRouter, the enterprise LLM gateway platform built by CloudAct Inc., today announced the general availability of Cortex, an air-gapped machine learning sidecar purpose-built to deliver real-time prompt injection defense and data privacy sanitization with mathematical isolation.
As enterprise adoption of generative AI accelerates, Chief Information Security Officers (CISOs) and compliance leaders face an escalating dilemma: traditional guardrail solutions inspect user prompts by relaying them to third-party cloud APIs, introducing secondary data-leakage surfaces and adding significant latency. Cortex eliminates this trade-off by running entirely within the customer's private infrastructure perimeter.
Mathematical Air-Gap: Zero Database, Zero Egress
Unlike monolithic API proxies that intertwine routing logic with persistent storage, Cortex is architected around strict operational boundaries:
- Stateless in-memory execution: Cortex holds zero database credentials, maintains no persistent disk storage, and performs zero logging of prompt payloads.
- Zero external egress: Operating strictly within private network boundaries, the sidecar has no outbound internet connectivity. It cannot transmit sensitive telemetry or training samples outside the enterprise perimeter.
- Microsecond mTLS 1.3 communication: The nRouter Rust gateway core communicates with Cortex over mutual TLS 1.3 channels with hardware-accelerated cryptographic authentication.
Guaranteed Zero Spend on Blocked Injections
A core architectural invariant of nRouter is financial protection against adversarial abuse. In traditional proxy architectures, an application must call an expensive frontier model before realizing the output was compromised or toxic, burning costly token credits.
With Cortex integrated into the gateway preflight inspection chain:
- Every incoming prompt undergoes parallel heuristic and neural classification before provider dispatch.
- If a jailbreak, prompt injection, or malicious payload is identified, the request is immediately terminated at the gateway boundary.
- The platform holds $0 and bills $0 to the customer's ledger. Denied requests incur zero financial penalty.
Deep Privacy: Real-Time PII & Secret Redaction
Cortex executes high-throughput entity extraction and contextual masking at wire speed. Sensitive identifiers — including Social Security numbers, HIPAA-protected health records, payment card numbers, internal bearer tokens, and private API keys — are automatically sanitized or tokenized before upstream forwarding, ensuring compliance with global data protection mandates.
Executive Commentary
"Enterprise engineering teams cannot afford to choose between cutting-edge AI capabilities and data sovereignty," said Rama Surasani, Founder and CEO of nRouter. "Cortex proves that deep security and ultra-low latency are not mutually exclusive. By air-gapping our ML sidecar from external egress and pairing it with our zero-spend financial guarantee, we provide engineering teams with a bulletproof defense perimeter for autonomous AI workloads."
Availability
Cortex is available immediately to all nRouter customers across managed cloud and dedicated Virtual Private Cloud (VPC) deployments. For technical documentation, visit nrouter.ai/docs.
For media inquiries: sales@nrouter.ai. For enterprise architecture consultations: sales@nrouter.ai.
About nRouter
nRouter is a managed LLM gateway for teams that want one key, one bill, and zero provider configuration. Customers buy credits, call any major model through a single endpoint, and get cost tracking, routing, and safety controls built in. nRouter is built and operated by CloudAct Inc..
More from the press room

Enterprise Adoption Surges as nRouter Reaches SOC 2 Type II In-Progress Milestone and Launches Dedicated VPC Deployments
Accelerating enterprise adoption across fintech and digital health with institutional security, Postgres Row-Level Security isolation, and private VPC deployment options.

nRouter Expands Gateway to 100+ Multimodal Models
Unified enterprise inference across Google Vertex AI, Anthropic Claude, OpenAI, AWS Bedrock, Meta Llama, and DeepSeek with sub-millisecond provider failover and multimodal support.

nRouter Abolishes Broker Markups with Flat List-Price Settlement and FinOps FOCUS 1.4 Tokenomics
Eliminating the 15% broker markup tax: nRouter establishes 0% per-token markup, settling at exact upstream provider list prices with automated 3-way FinOps FOCUS 1.4 reconciliation.