Coming Soon

aidr

AI Detection and Response

A network-layer control plane for AI agent egress traffic. Intercepts every outbound connection, evaluates it against your policy, logs the decision, and blocks anything not explicitly allowed.

aidr audit log

The Problem

AI agents make outbound network calls. You have no idea what they're doing.

EDR

Covers endpoints. Doesn't know the call came from an AI agent.

LLM Gateways

Only see calls routed through them. Blind to tool calls, browsing, and MCP requests.

AIDR

Fills the gap. Network-layer visibility and control for every AI agent on the host.

What It Does

One proxy. Total visibility.

01

Intercepts

Every outbound HTTP/S connection an AI agent makes — no code changes, no SDK to integrate.

02

Evaluates

Each connection against rules you define. Allow, block, or default-deny by destination.

03

Logs

Every decision in a structured, append-only audit trail. Process name, PID, destination, verdict.

04

Blocks

Anything not explicitly allowed. Default-deny means unknown destinations never reach the internet.

For Security Teams

Stop AI agents from becoming an attack surface.

Get the same class of control over AI traffic that firewalls gave you over network traffic.

  • ✗Block prompt injection exfiltration attempts
  • ✗Detect agents calling attacker-controlled infrastructure
  • ✗Enforce policy-as-code on every AI workload
  • ✗Audit trail for compliance and incident response

For Platform Engineers

Make AI agent behavior observable and auditable.

Deploy once. Get visibility across every agent on the host — no instrumentation required.

  • ✓Structured logs for every request an agent makes
  • ✓Process name, PID, and destination captured automatically
  • ✓No instrumentation — deploy once, visibility across all agents
  • ✓Policy-as-code that's reviewable and version-controlled

Be first to know when we launch.

We're building the control plane for AI agent egress. Join the waitlist for early access.