AI Commitment

Updated: 13 July 2026

Machine learning is a tool, not a tenant. Noiz uses it where it makes the work faster or sharper, and refuses it where it would cost the client privacy, ownership or a person to answer for the outcome. This is a working statement, kept honest as the technology and the obligations around it change.

Principles

• Client data stays the client’s. Nothing belonging to a client is used in any AI system without the client’s full, informed consent. Nothing is fed into third-party models for training, and nothing leaves the client’s boundary to make a vendor’s model better.
• A human is always accountable. AI assists triage, drafting and analysis. It does not sign off changes, approve access or close incidents. A named person does.
• Transparency over magic. Where AI shapes a decision that affects a client, Noiz can say so, and say why. Where AI has helped create content, Noiz clearly labels it as such.
• Open and portable by preference. Self-hosted and open-source models are chosen first, so capability does not become lock-in and inference does not become surveillance.

What this looks like in practice

Noiz applies machine learning to the unglamorous parts of managed infrastructure: surfacing anomalies in logs, summarising alerts, accelerating routine review. The judgement stays human. Configuration, remediation and anything that touches a client’s environment is reviewed and authorised by an engineer before it takes effect.

Models that process client material run on infrastructure Noiz controls, or not at all. Convenience does not override that.

Why it matters

The fastest way to deliver hosting and security today is to hand data, context and decisions to whatever model is largest this quarter. Noiz declines that trade. The point of managed infrastructure is that the client keeps control of their systems and their information while someone else carries the load. AI that quietly erodes either of those is not a feature; it is a liability wearing a feature’s clothes.

Questions about how Noiz uses AI on a specific engagement are welcome. There is a person to ask.