Netwit · VZU flagship · 2026
Netwit.
Agentic AI automation for the enterprise. Multi-agent orchestration. MCP-native integrations. RBAC at the agent level. On-prem by default.
Netwit is the VZU flagship for agentic AI automation. Netwit is the runtime the other 11 brands in the VZU family ship on. Netwit is multi-agent orchestration, MCP-native integrations, RBAC enforced at the agent level, and deployment in cloud, hybrid, or air-gapped on-prem. Netwit ships with a SOC 2 / HIPAA / ISO 27001 posture, and the audit trail is on from day one.
The flagship process
How Netwit works, in five steps.
A real brief becomes a real product. The runtime handles the orchestration, the audit trail, and the enforcement. The operator owns the goal.
-
Step 01
Brief
A real brief. A real problem. A real operator. The brief sets the boundary of the engagement — what the system needs to do, what data it can touch, what regulator it answers to, and what success looks like at 90 days. Netwit ships on the brief. The brief is the contract between the operator and the runtime.
-
Step 02
Model the work
The work is decomposed into a goal, a toolset (implemented as MCP servers), an audit posture, and a memory schema. The Netwit runtime gets the goal, the tools, and the rules. The model of the work is the contract between the operator and the runtime.
-
Step 03
Orchestrate
The runtime plans the steps, dispatches the agents, observes the results, and replans. Multi-agent orchestration handles the cases where the work splits (parallel) or where a worker needs to be re-tasked (supervisor) or where the steps are strictly sequential (relay). Every call is logged in the audit trail.
-
Step 04
Enforce the perimeter
RBAC checks every tool call. The audit trail records who, what, when, with which model. Compliance posture (SOC 2, HIPAA, ISO 27001, PIPEDA) is enforced by the runtime, not bolted on. The data never leaves the operator's perimeter when on-prem is selected.
-
Step 05
Operate
A quarterly or annual contract. The on-call is 24/7. The audit trail is on from day one. The runtime gets updated, the agents get retrained on real outcomes, and the operator's posture improves over time. The work compounds.
The five steps are the same regardless of the brief. The complexity lives in the runtime, not in the operator's workflow.
How this works, in plain words. The animation is decorative; the text is the source of truth.
The runtime
Four primitives, one runtime.
01 · The goal
What the operator wants done.
A goal is what the system is trying to achieve. It is given to the runtime in natural language or in a structured form. The runtime plans the steps to achieve the goal. The goal is the contract between the operator and the runtime.
02 · The tools
What the agent can call.
A tool is what the agent can do to make progress on the goal. Every tool is an MCP server. The agent reads the tool list, decides which to call, calls it, gets the result, and continues. The tool does not need to know how the agent is implemented. The agent does not need to know how the tool is implemented.
03 · The memory
What the agent knows between calls.
A memory is what the agent retains across calls. The conversation history. The working state. The durable record. The memory is owned by the runtime, with the retention policy set by the operator. The audit trail is the long-term memory.
04 · The perimeter
What the agent is allowed to do.
A perimeter is the boundary of what the agent is allowed to do. The runtime checks every tool call against the perimeter. The perimeter is set by the operator. The perimeter is what makes the system auditable. The perimeter is what the regulator sees.
Where it runs
Three deployment modes, one runtime.
Mode 01
Cloud
Canadian region for PIPEDA. US region otherwise. EU region available. The fastest path to production. The default for new operators.
Mode 02
Hybrid
The runtime on the operator's hardware. The model API-called from the cloud. The data stays in the operator's perimeter. The runtime's audit trail records every cross-boundary call.
Mode 03
On-prem
The runtime, the model, and the tools all on the operator's infrastructure. Including air-gapped. The right answer for the regulator that requires everything inside the perimeter.
The compliance posture
SOC 2 · HIPAA · ISO 27001 · PIPEDA.
SOC 2
The runtime is in active pursuit of SOC 2 Type II. The audit window opens Q3 2026. The runtime is the compliance boundary. The audit trail is the auditor's form.
HIPAA
The runtime ships with HIPAA-aware infrastructure. On-prem deployment available for operators under HIPAA. The AI Scribe service (built on Netwit) is in production with a healthcare operator under PIPEDA + HIPAA.
ISO 27001
The runtime is ISO 27001-aligned. The operator's ISO 27001 program is layered on top of the runtime, with the runtime providing the access control, change management, and monitoring primitives.
PIPEDA
The default cloud region is Canadian, for PIPEDA compliance. The data residency is enforced by the runtime, not by convention. The audit trail is on from day one.
The brands on the runtime
What ships on Netwit.
- PostMTA — Self-hosted email delivery. The inbox that doesn't break.
- InboxGrove — SMTP infrastructure with IP warm-up and deliverability tooling.
- Voice Agently — AI voice agents, 24/7, in the operator's brand voice.
- TrackSphere — Real-time delivery tracking with proof of delivery.
- Agentic OS — The flagship runtime. This page. (See also: the legacy Agentic OS page.)
- AI Agent Marketplace — Twelve ready-to-deploy agents on the Netwit runtime.
The rest of the portfolio — Voice Agently, PostMTA, InboxGrove, TrackSphere, Plex Dubai, Wahix India, Eventek, and the AI Agent Marketplace — are listed on the about page.
Get in touch
Ship Netwit in two weeks.
Email the team. Reply in one business day. The pilot is priced, the architecture is real, the ship date is real. The runtime ships the same in cloud, hybrid, or on-prem.