# AI Usage Policies

SaaS admin console where team leads define maximum inference budgets and assign models based on user seniority.

- id: `ai-usage-policies`
- Page: https://agentic-ui-kit.com/components/ai-usage-policies
- Live preview: https://agentic-ui-kit.com/live/component/ai-usage-policies

## Install

```bash
agentic-ui add ai-usage-policies
npx shadcn@4.17.0 add @agentic-ui/ai-usage-policies
```

## Metadata

- **Level:** block
- **Category:** block
- **Installable in:** Radix UI (registry variant new-york)
- **Capabilities:** spend-limit-quotas, token-quota-allocation, model-whitelisting, role-based-access, live-usage-tracking, rate-limiting, enforcement-actions, pii-redaction-guardrail, human-approval-threshold, tool-capability-controls, policy-creation-and-duplication, read-only-audit-mode, keyboard-navigation, live-status-announcements, empty-state, error-recovery
- **Intents:** configure, manage-collection, track-status, audit-trail, create-edit
- **Suits domains:** settings, saas, ai-agents, billing, operations, ai-workspace
- **Interaction:** configure-limits, toggle-model-whitelist, switch-tabs, inline-edit, filter-by-scope, create-policy, duplicate-policy, delete-policy, save-discard, keyboard-navigation
- **Screen role:** management
- **Density:** medium
- **Dark mode:** yes
- **Use it for:** SaaS admin console where team leads define maximum inference budgets and assign models based on user seniority, enterprise governance panel that restricts expensive reasoning models to Data Science and downgrades to eco models when the limit is reached, security audit view to certify that all policies enforce mandatory PII anonymization, API limit manager for autonomous agents with a human approval threshold before running expensive tasks
- **Avoid when:** management of provider API keys and BYOK secret rotation (use ai-provider-keys), global catalog of models available in the organization without association to users or limits (use ai-model-registry), detailed audit of individual agent execution traces (use ai-audit-actions or ai-agent-run-trace)
- **Preserve when adapting:** usage-limits-and-quota-contract, model-whitelist-selection-logic, live-consumption-indicators, enterprise-guardrails-toggles, live-accessibility-announcements
- **Replace when adapting:** domain-copy, model-catalog, currency-and-rate-metrics
- **Adaptation effort:** low
- **Coupling:** low
- **Quality score:** 9.2/10
- **Quality breakdown:** visual 9/10, responsive 9/10, accessibility 9/10, code 9/10
- **Uses primitives:** badge, button, card, checkbox, dialog, input, label, progress, select, separator, switch, tabs, alert
- **Motion:** profile subtle, library css, reduced-motion fallback yes
- **Registry dependencies:** lib-primitives