AI Administration
LiveLower the model ceiling, switch AI features off, cap daily spend, and read the AI audit history and security events.
Plan availability. The administration pages are on every plan. Custom training appears only where the Custom Model Training feature is enabled, and this is not switched on in any of our four published plans. Starter, Pro, Business and Enterprise all resolve to the same feature set and this one is off in all of them, so a subscription will not unlock it — it has to be enabled for your organisation by arrangement. Ask us before you plan around it.
- ✓Model ceiling that only lowers the tier
- ✓Switch off Deep Audit's OpenAI and Gemini readers
- ✓Per-feature and per-request-type switches
- ✓Daily spend cap in the organisation's billing currency
What it is.
An organisation can only lower what the platform would spend: set a model ceiling, switch off the second-opinion vendors used by Deep Audit, switch off individual AI features or request types, and set a daily spend cap in your billing currency. Each is checked before any model is called; a blocked request is refused with a reason, and a policy that fails to load does not stop work.
AI Audit History lists every AI request — type, user, model, time — with no content. AI Security lists injection detections and blocked requests, storing a fingerprint, never the prompt. The page states which settings are enforced and which are only recorded: the client-data rule, redaction of personal identifiers, the human-review requirement and the policy statement are stored but not enforced. Custom training reads your documents into staged house-style preferences an admin activates; it is not fine-tuning, and no model is trained. The prompt library stores prompts, which are not yet offered in Vio or drafting.
Three steps.
End to end.
More › AI Administration › Model Configuration and AI Policies.
Set a daily cap; calls past it are refused until the next UTC day.
Read AI Audit History and AI Security.
What you get.
- ✓Model ceiling that only lowers the tier
- ✓Switch off Deep Audit's OpenAI and Gemini readers
- ✓Per-feature and per-request-type switches
- ✓Daily spend cap in the organisation's billing currency
- ✓AI audit history without content
- ✓AI security events stored as fingerprints
- ✓A stated table of what is enforced and what is only recorded
- ✓Prompt library (not yet used by Vio or drafting)
Quick answers.
No. The policy can only lower the ceiling.
No. Speech goes to a different vendor path and is not covered.
More in AI Controls & Guardrails.
We do not train models on customer content. There is no opt-in, because there is no training programme.
User-pasted text is sanitised before reaching the LLM.
A monthly AI token quota per plan, an 80% warning by email and in-app, and a hard stop at 100%.
Stable prompt prefixes are cached. Faster responses, lower bill, passed to you.