Skip to content
Edit page
Copy page

Insights

AI Workspace includes a built-in Insights dashboard that gives you visibility into traffic, usage patterns, and performance across your LLM providers and proxies. Moesif, an AI-native API analytics platform, powers the dashboard, and it's embedded directly in AI Workspace. You don't need a separate Moesif account or setup.

Open Insights

To open the dashboard, follow these steps:

  1. Navigate to AI Workspace in your API Platform dashboard.
  2. Click Insights in the left navigation menu.

The page loads an analytics dashboard scoped to your AI Workspace activity.

Enable analytics on a gateway

The dashboard shows data only for gateways that publish telemetry. A gateway publishes telemetry when its runtime carries an analytics key in the MOESIF_KEY environment variable. The Get Started section on each gateway's page fills this value in for you when you copy the setup commands, so a gateway set up from those commands already publishes telemetry.

If you set up a gateway without the key, add it to configs/keys.env and restart the gateway:

MOESIF_KEY=<your-analytics-key>

For the full gateway setup procedure, see Set up an AI Gateway.

Note

If MOESIF_KEY isn't configured, the gateway publishes no telemetry, and the Insights dashboard shows no data for that gateway.

What you can see

Overview summary

The overview panel at the top shows key metrics at a glance:

Metric Description
Total Requests The total number of requests processed through your LLM providers and proxies
Avg Latency The average end-to-end response latency across all requests
Success Rate The percentage of requests that completed without an error response

Request trend

A time-series chart showing request volume over time. Use it to identify traffic spikes, usage patterns, and the effect of configuration changes on request volume.

Traffic split

A breakdown of traffic distribution across your App LLM proxies and providers. Use it to understand which proxies or providers handle the most load, and whether traffic is distributed as you expect.

Full analytics dashboard

The embedded dashboard provides deeper analytics, including:

  • Request logs: inspect individual API calls, their payloads, status codes, and latencies.
  • Filtering and segmentation: slice data by proxy, provider, endpoint, consumer, or time range.
  • Application activity: analyze usage for mapped GenAI applications.
  • Token usage trends: track prompt, completion, and total token consumption over time.
  • Estimated LLM cost: review cost estimates based on token consumption.
  • Error analysis: identify error patterns and guardrail interventions.
  • Consumer activity: monitor which applications or consumers generate the most traffic.

Use Insights for cost and rate limit management

Use the dashboard to tune your token-based rate limits:

  1. Use token usage trends to understand actual consumption patterns before you set rate limits.
  2. Check per-consumer traffic to determine appropriate per-consumer limits.
  3. Monitor error rates to detect rate limit violations (429 Too Many Requests responses), and adjust the limits accordingly.

Use Insights for guardrail monitoring

When guardrails are active, use the dashboard to:

  • Track how often guardrails intervene. Guardrail intervention responses appear as 422 errors.
  • Identify which endpoints or consumers trigger the most interventions.
  • Confirm that newly configured guardrails behave as you expect.

See Policies overview for more on configuring guardrails.