VZU Insights · VZU primer · 2026-07-10
Agentic OS, explained.
A primer on VZU Agentic OS: the runtime, the building blocks, the blueprints, the governance, the deployment modes.
The essay
Agentic OS, explained.
Netwit Agentic OS is the VZU runtime. It is a production agent platform with multi-agent execution, MCP-native integrations, RBAC enforced at the agent level, and deployment in cloud, hybrid, or air-gapped on-prem modes. This is a primer on what the runtime is, what it does, and how it is built.
The runtime is built around three primitives: goals, tools, and memory. A goal is what the operator wants done. A tool is what the agent can call to do the work (every tool is exposed as an MCP server). A memory is what the agent knows between calls (the conversation history, the working state, the durable record). The runtime takes a goal, plans the steps, dispatches the agents, observes the results, and replans.
Multi-agent execution means the runtime can split a single goal across multiple agents. The three patterns are relay (A passes to B, B to C), parallel (A, B, C run at the same time, a coordinator merges), and supervisor (a supervisor agent dispatches work to workers and decides what to do next). The runtime picks the right pattern per blueprint.
MCP-native means every tool is an MCP server. The agent does not need to know how the tool is implemented. The tool does not need to know how the agent is implemented. The economics flip from N×M (every agent to every tool) to N+M (every agent speaks MCP, every tool speaks MCP).
RBAC at the agent level means permissions are checked on every tool call, against the agent's identity, not the user's. The agent sees what the user is allowed to see. The audit trail records every check. A regulator can ask "who decided this" and the answer is the agent ID, the prompt, the model, the tool calls, the result, and the human who approved.
Deployment modes are cloud (Canadian region for PIPEDA, US region otherwise, EU region available), hybrid (the runtime lives on the operator's hardware, the model is API-called), and on-prem (the entire stack — runtime, model, tools — lives inside the operator's data center). The same code, the same audit trail, the same RBAC, regardless of where it runs.
Disclosure
Dhiraj Chatpar is the founder of VZU. The VZU team wrote this essay. The runtime described is in production at a small number of operators. The benchmarks, where cited, are from internal testing or from operators who have given permission to publish.