Govern AI the way you govern the rest of your estate.

Ajutant runs inside your own Azure tenant. Your content, logs and audit trail stay where your existing controls already reach, and every request passes through a policy gate that you set.

What your security team is actually asking.

These are reasonable questions, and they should have inspectable answers rather than assurances.

  • Where does the content go, and who can read it?
  • Who approved this assistant, and can we see what it was asked?
  • Does it respect the permissions we already set in SharePoint?
  • What happens on the day we need to switch it off?

A governance policy wrapped around someone else's SaaS is a document. Ajutant makes governance the runtime every assistant executes in.

Governance as the runtime, not a policy document.

The controls below are not settings you can opt into later. They are the path every request takes.

The data boundary is your tenant.

Deployed via Bicep into your own Azure subscription, using an identity you grant and can revoke. Content, logs and diagnostics stay inside it. Content leaves only to a model provider you have contracted with, or to web search if you enable it.

Every request is gated and logged.

An acceptable-use gate, guardrails with PII detection and redaction, role-based access with custom roles, and an immutable audit log written to your own Azure Log Analytics workspace.

The permissions you already set are honoured.

Knowledge from SharePoint answers each person according to their own permissions, and answers carry click-to-page citations, so a reviewer can check the source rather than trust the summary.

You can stop it.

Approval before anything is published, version history with one-click rollback, and a kill switch for a single assistant or the whole platform.

The questions that come up in the security review.

Q.01
Are you certified?
No, and we won't imply otherwise. Ajutant holds no ISO 27001, SOC 2 or Cyber Essentials Plus certification today. What we can do is show your team the architecture, the controls and the security documentation and let them assess it, in an environment that runs inside your own certified tenant, under your own controls.
Q.02
Where does your data go?
It stays in your Azure tenant. We never see your content. The only thing that comes back to us is summarised token usage, for billing, and statistical quality metrics. No copies in a shared vector database.
Q.03
What happens when something goes wrong?
You have a kill switch, an audit trail, role-scoped access, and per-team cost limits. When a user flags an ethical or correctness issue with a response, it reaches your IT operations team in real time, not the model provider. Every interaction is logged, and every assistant is governed by your acceptable use policy.

Bring your security team. We'll answer their questions.

Thirty minutes on the data boundary, identity, the audit trail and the incident path. Bring whoever needs to sign this off.

Book a 30-minute discovery call

Available this week · UK time zones · Calendar booking