AI Governance
AI Governance provides organisation-wide controls over AI usage, content safety, model access, and spending. Navigate to AI Services > AI Governance in the dashboard to configure these settings.
All governance policies apply across every AI surface in your organisation — the portal, QuantCode Gov, and all API access, which includes Slack bots, the Studio AI assistant, and workflows.
AI Access
Section titled “AI Access”The master toggle to enable or disable AI services for your entire organisation.
| Setting | Description |
|---|---|
| Enable AI Services | When disabled, all AI inference requests are blocked across all surfaces |
Content Safety
Section titled “Content Safety”Mandatory Guardrail Preset
Section titled “Mandatory Guardrail Preset”Enforce an organisation-wide guardrail that overrides agent-level and request-level settings. These presets are aligned to the Australian Protective Security Policy Framework (PSPF):
| Preset | Description |
|---|---|
| None | No organisation-wide guardrail enforced |
| OFFICIAL | Baseline content safety and Australian PII detection |
| OFFICIAL:Sensitive | Stricter filtering with word and topic policies |
| PROTECTED | Maximum filtering with grounding checks |
When a preset is set, it is enforced on every AI request regardless of individual agent configuration.
Mandatory Filter Policies
Section titled “Mandatory Filter Policies”Select filter policies that are enforced on every AI request organisation-wide, regardless of per-agent settings. This ensures that critical content filtering rules (such as PII redaction or prohibited content blocking) cannot be bypassed at the agent level.
Approved Models
Section titled “Approved Models”Control which AI models your organisation can use:
| Policy | Description |
|---|---|
| Unrestricted | All available models can be used |
| Allowlist | Only selected models can be used |
| Blocklist | Selected models are blocked; all others are available |
When using allowlist or blocklist mode, a grid of available models is displayed with their provider (and category, where available). Select or deselect models to build your policy.
This is useful for organisations that need to restrict AI usage to specific approved models for compliance or cost reasons.
Spend Limits
Section titled “Spend Limits”Set budget caps to control AI inference costs:
| Setting | Description |
|---|---|
| Monthly Budget (USD) | Organisation-wide monthly cap. Leave empty for no limit |
| Daily Budget (USD) | Organisation-wide daily cap |
| Per-User Monthly Budget (USD) | Monthly cap per individual user |
| Per-User Daily Budget (USD) | Daily cap per individual user |
| Warning Threshold (%) | Alert at this percentage of budget (e.g. 80%) |
You can also set per-interface caps — aggregate budgets for individual surfaces such as Slack, API gateway, streaming, workflows, orchestrations, and embeddings — and per-user overrides, which grant a named user a custom budget or mark them as unlimited.
When a budget is exceeded, AI requests are rejected (HTTP 429) until the budget window resets — the next day for daily caps, the next month for monthly caps. The warning threshold triggers notifications before the limit is reached.
Current Spend
Section titled “Current Spend”A read-only dashboard showing real-time spending data for the current month:
- Month to Date — total spend in USD
- Today — today’s spend
- Requests (Month) — total AI inference requests
- Tokens (Month) — total input and output tokens consumed
When a monthly or daily budget is configured, each gets its own progress bar showing spend against the budget (with spent, budget, and remaining amounts) using colour-coded thresholds:
- Green — under 60% of budget
- Amber — 60-80% of budget
- Red — over 80% of budget
Next steps
Section titled “Next steps”- Filter Policies — Create custom content filtering rules
- Agents — Configure individual agent settings
- API Reference — Manage governance settings via the API