Dashboard

Dashboard

The Dashboard is your account-level observability hub. It provides a unified view of agent runs, content source pulls, and credit usage over configurable time windows — giving you the insights you need to keep your AI workflows healthy and efficient.

Overview

The Dashboard is organized into three tabs:

TabIconWhat It Shows
Agents🤖Agent run metrics, failure analysis, and performance trends
Content Sources🔌Source pull metrics, success rates, and pull history
Credit Usage🪙Credit consumption charts and usage summaries

When you open the Dashboard, it defaults to the Agents tab. Each tab displays data scoped to the time frame you select.

Time Frame Selector

The Time Frame Selector at the top of the Dashboard controls the date range for all data across every tab. It supports both relative and absolute ranges.

Relative ranges (preset):

OptionRange
Last 24 hoursPast 24 hours
Last 7 daysPast 7 days
Last 30 daysPast 30 days
Last 90 daysPast 90 days
Last 6 monthsPast 6 months
Last yearPast 12 months

Absolute range: Pick a custom start and end date using the date picker.

The dashboard automatically determines the data granularity based on the selected range:

Range DurationGranularity
≤ 31 daysDay — data grouped by day
≤ 182 days (≈6 months)Week — data grouped by week
> 182 daysMonth — data grouped by month

Example: Selecting "Last 7 days" shows daily data points in the charts, while "Last year" shows monthly aggregated data.


Agents Tab

The Agents tab provides comprehensive monitoring for all agent runs across your account. It combines high-level summary metrics with detailed failure analysis and performance tracking.

Summary Cards

Three summary cards at the top provide at-a-glance metrics for the selected time frame:

CardDescriptionVisual Indicator
Total RunsTotal number of agent runs startedNeutral
CompletedRuns that finished successfullyGreen
Error RatePercentage of runs that failedRed when > 10%

Example: If 200 runs started, 180 completed, and 20 failed, you would see:

  • Total Runs: 200
  • Completed: 180
  • Error Rate: 10% (displayed in red)

Traces Chart

A bar chart visualizes completed vs. failed runs over time. Each bar is divided into two segments:

  • Green — Completed runs
  • Red — Failed runs

The x-axis labels adapt to the selected granularity (day, week, or month). Hover over any bar to see exact counts.

Example: A daily chart for "Last 7 days" might show:

Mon: 32 completed, 2 failed
Tue: 28 completed, 0 failed
Wed: 35 completed, 5 failed   ← spike visible as taller red segment
Thu: 30 completed, 1 failed
...

Recent Failures

Below the chart, a Recent Failures panel lists the most recent failed agent runs. Each entry shows:

  • Agent name — Clickable link that navigates to the specific run detail
  • Error message — Truncated preview of what went wrong
  • Timestamp — When the failure occurred

Use this panel to quickly identify and investigate problems without navigating through each agent individually.

Example entries:

AgentErrorWhen
Daily News SummaryRate limit exceeded for model gpt-4o2 hours ago
Customer FAQ BotKnowledge base "Support Docs" has no indexed content5 hours ago
Weekly Report GeneratorS3 bucket write permission denied1 day ago

Slowest Runs

The Slowest Runs panel highlights agent runs that significantly exceeded their historical p95 (95th percentile) duration. Each entry shows:

  • Agent name — Link to the run detail
  • Actual duration — How long the run took
  • P95 benchmark — The typical p95 duration for that agent
  • Multiplier badge — Shows how many times slower (e.g., "2.3× p95")

This helps you catch performance regressions before they become chronic issues.

Example: An agent that normally completes in under 12 seconds (p95) but took 45 seconds would display:

Content Analyzer   45s / p95: 12s   [3.8× p95]

Agents Slowing Down

The Agents Slowing Down section surfaces agents whose recent average duration has increased significantly compared to their historical baseline. Each card shows:

  • Agent name — Clickable link to the agent's runs page
  • Recent average vs. baseline average duration
  • Slowdown badge — Color-coded multiplier:
    • Yellow — 1.5× to 2× slower than baseline
    • Red — 2× or more slower than baseline

This proactive metric helps you detect gradual performance degradation across your agents.

Example:

Weekly Digest Generator          [2.1× slower]
Recent avg: 42s → Baseline avg: 20s

Product Update Monitor           [1.6× slower]
Recent avg: 8s → Baseline avg: 5s

Content Sources Tab

The Content Sources tab mirrors the Agents tab structure but focuses on source pull operations — the periodic fetches that keep your knowledge bases up to date.

Summary Cards

CardDescriptionVisual Indicator
Total PullsNumber of source pulls initiatedNeutral
CompletedPulls that finished successfullyGreen
Error RatePercentage of pulls that failedRed when > 10%

Pull History Chart

A bar chart identical in design to the agent run chart, showing completed vs. failed pulls over time. Use this to spot patterns — for example, a source that fails every weekend might indicate a maintenance window on the upstream server.

Note: If no content source data is available for the selected time frame, the tab displays: "No content source data available".


Credit Usage Tab

The Credit Usage tab shows how your credits are being consumed over time.

Usage Bar Chart

A bar chart displays credit consumption per time period, with bars sized to the selected granularity (day, week, or month).

Account Credit Summary

An overlay provides a snapshot of your current credit balance and usage:

  • Total credits used in the selected period
  • Current balance (subscription + purchased credits)
  • Subscription credits remaining this billing cycle

For more details on how credits work, see Credits & Usage.

Note: If no usage data exists for the selected period, the tab displays: "No credit usage data available yet".


Permissions

The Dashboard is available to users with editor, admin, or owner roles on the account. Users with the viewer role are redirected to the Agents list page instead.

RoleDashboard Access
Owner✅ Full access
Admin✅ Full access
Editor✅ Full access
Viewer❌ Redirected to Agents

Common Use Cases

Daily Health Check

Start your day by opening the Dashboard with "Last 24 hours" selected:

  1. Check the error rate on the Agents tab — anything above 10% needs attention
  2. Review Recent Failures to identify recurring issues
  3. Scan the Content Sources tab for pull failures that could affect data freshness
  4. Glance at Credit Usage to ensure consumption is within expected range

Weekly Performance Review

Select "Last 7 days" and compare to the previous week:

  1. Look for upward trends in the error rate
  2. Check the Agents Slowing Down section for performance regressions
  3. Review the Slowest Runs to investigate outlier executions
  4. Compare credit usage to projected budgets

Incident Investigation

When you receive an alert, use the Dashboard to understand the broader context:

  1. Set the time frame to cover the period around the incident
  2. Check if the issue is isolated (single agent) or systemic (multiple agents failing)
  3. Correlate agent failures with content source pull failures
  4. Use Recent Failures to trace the root cause

Next Steps

  • Alerts — Set up automated notifications for failures and performance issues
  • Agents — Learn about creating and managing agents
  • Content Sources — Configure the sources that feed your knowledge bases
  • Credits & Usage — Understand the credit system and pricing