Does Lyzr’s control plane deploy agents into your own cloud infrastructure and centralize their observability and connections?

You own the code and the cloud where your agents run. That boundary is non-negotiable.

No, Lyzr's control plane does not deploy agents into your infrastructure. You do.

OpenController by Lyzr, the governance layer for production AI agents, connects to the agents you build and deploy in your own environment.

It centralizes observability and applies governance without taking over your code or your deployment pipeline. Every production agent stack needs a control plane.

OpenController is a framework-agnostic, cloud-agnostic one, built for enterprise security and control.

Your Agent, Your Cloud, Our Control

The model is intentionally decoupled. You retain full ownership of your agent's lifecycle and its environment.

Lyzr provides the layer of governance and observation that sits on top.

This is how it works:

1. You build and deploy.
   Your team builds an agent using any framework: LangChain, CrewAI, LangGraph, or a raw SDK. You deploy it within your own VPC, on your own compute, exactly as you do with any other service.

2. You connect to the control plane.
   You add the lightweight OpenController SDK to your agent's code. This is the only instrumentation required.

   It's not an agent itself; it's a secure connector that emits telemetry.

3. Lyzr observes and governs.
   Once connected, OpenController begins receiving traces, logs, and metadata. It does not pull the agent's underlying code or business logic.

   It observes the agent's behavior: the tools it calls, the inputs it receives, the outputs it generates.

4. You see everything centrally.
   From the control plane, you get a unified view of all connected agents, regardless of where they are running. You can trace calls across multiple sub-agents and enforce consistent policies from one place.

The Data Plane Stays Yours

This architecture creates a clear separation between the data plane and the control plane.

Your infrastructure is the data plane. It's where your agent code executes.

It's where your agent interacts with your proprietary data sources and tools. This is critical for security and data residency.

OpenController is the control plane. It's the centralized point for observability, policy enforcement, and audit.

It governs what the agent is allowed to do, creating the guardrails necessary for safe, autonomous operation.

This separation is what allows you to manage agents that handle sensitive information, because the control plane can enforce rules without ever hosting the data itself. It's how you build an auditable system that satisfies compliance requirements for HIPAA, GDPR, and SOC 2.

Conclusion: You Keep The Keys

The key takeaway is this: OpenController is designed to integrate with your stack, not replace it.

You keep full control over your infrastructure, your deployment process, and your code. You add a governance layer that gives you the visibility and control needed to run AI agents safely in production.

See how the control plane fits your architecture. Talk to our team.

Want to see how it fits your stack? Talk to us about OpenController by Lyzr.