AI Optimization

AI Optimization brings together everything your team needs to understand AI spend — which models and platforms you’re using, where savings opportunities are, and when something unusual happens — without jumping between tools.

​

Get started

1. Sign in to nOps.
2. In the sidebar, open Inform → AI Optimization, or go to clara.nops.io/ai-optimization.
3. Use the tabs at the top to switch between Model Overview, Token Allocation, Optimization Recommendations, and Anomalies.

​

Model Overview tab

Model Overview shows how your organization’s AI spend breaks down by model and platform.

• Model usage — daily spend and token volume trends over the last 30 days, plus a breakdown of token usage by platform.
• Spend and tokens by model — every model ranked by spend, with tabs to filter by platform (OpenAI, Anthropic, AWS Bedrock, and more). Click a model to open a detail pane with its usage trend.
• Export CSV to download the current model list, respecting your platform filter and search.
• Search models or platforms directly from the toolbar.

If a platform you use hasn’t been connected yet, an Add platform option appears alongside the platform tabs — it takes you straight to Settings to connect it.

​

Token Allocation tab

Token Allocation pins an Explorer dashboard focused on AI spend, so you can see how usage breaks down by team, project, or any other dimension you track.

• If your organization hasn’t set a default dashboard yet, choose one from your Shared dashboards.
• Organization admins can change the default dashboard for their org at any time.
• nOps super admins can set a platform-wide default that applies to any organization without its own override.

​

Optimization Recommendations tab

This tab shows Clara Feed results scoped to AI Services — any scheduled prompt you’ve set up to analyze AI cost and usage.

• If you haven’t created a schedule yet, you’ll see guidance on setting one up in Feed.
• Once schedules are running, results appear here automatically as they complete.

​

Anomalies tab

Anomalies surfaces unexpected changes in AI spend from two sources:

• AWS Detected Anomalies — anomalies from AWS Cost Anomaly Detection, filtered to AI services like Bedrock and SageMaker. Use the date range filter (Last 7 days, Last 30 days, Last 90 days, or All time) to adjust the window.
• AI Analysis — results from Clara scheduled prompts in the Anomaly Detection category, scoped to AI Services.

Summary cards at the top show the total number of anomalies, aggregate cost impact, and how many are Critical or High severity.

​

Connect your AI platforms

AI Optimization pulls usage from any AI platform you’ve connected. Head to Settings → Integrations to connect: AI spend from Google Cloud (Vertex / Gemini) and Azure shows up automatically once you’ve connected those cloud providers — see GCP setup or Azure setup.

​

Typical workflows

• Weekly AI spend review — start on Model Overview to see which models are driving spend, then switch to Token Allocation to see how it breaks down by team or project.
• Stay ahead of optimization opportunities — set up a Clara schedule for AI Services in Feed, then check back on the Optimization Recommendations tab.
• Investigate a spike — open Anomalies, filter to High severity, and drill into the AI Analysis results for more context.

​

FAQs

​

Related

• Explorer — analyze AI spend alongside the rest of your cloud costs
• Insights — broader cost recommendations and anomaly detection
• Feed — schedule the Clara prompts that power Optimization Recommendations