Pular para o conteúdo principal
The BI Dashboard is the financial and operational intelligence layer of Timely.ai. It transforms LLM usage data into actionable insights — from cost per conversation to consumption forecasts for the next 30 days.

​
Accessing the BI Dashboard

In the sidebar, click Analytics and select the BI Dashboard tab.
BI Dashboard with token usage charts and budget alerts

​
Token Usage

The token usage panel shows the total consumption of input and output tokens per period, broken down by agent and LLM model.
Pie chart and detailed table showing the token distribution by model (e.g., GPT-4o, Claude 3.5, Llama 3). Includes estimated cost per model based on current pricing rates from each provider.
Agents ranked by total cost in the period. Useful for identifying agents that consume more than expected and reviewing their context and instruction configurations.

​
Budget Alerts

Configure budget limits to receive alerts before exceeding your monthly budget.
1

Open budget settings

Click the Configure budget button in the Budget Progress card of the BI Dashboard.
2

Set the monthly limit

Enter the maximum value in tokens or estimated cost (USD) you want to use per month.
3

Configure alert thresholds

Set at what percentage of the budget you want to be alerted. The default is 80% (warning) and 95% (critical).
4

Save and activate

Save the settings. Alerts appear in the BI Dashboard and can be configured to send email notifications.
The Budget Progress card displays a real-time progress bar showing how much of the budget has been consumed in the current month.

​
Anomaly Detection

The anomaly detection system monitors consumption patterns and automatically flags when it detects significant deviations from the account’s history. Types of anomalies detected:
  • Sudden token spike in a single agent
  • Unusual increase in conversations at an atypical hour
  • Cost per conversation far above the historical average
  • Sharp drop in resolution rate combined with an increase in tokens
Each detected anomaly appears in the Anomalies card with a description of the deviation, the event timestamp, and a link to investigate related conversations.
Anomalies are not necessarily problems — they may reflect a legitimate demand spike. Use the historical context available in the card to interpret each alert before acting.

​
Token Forecast

The forecast projects consumption for the next 30 days based on recent trends. The chart shows:
  • Historical consumption line (solid)
  • Central projection (dashed)
  • Confidence interval (shaded area)
If the projection indicates budget overrun before the end of the month, a proactive alert is displayed with optimization suggestions (reduce context size, review high-cost agents, etc.).

​
Industry Benchmark

Compare your operation’s performance against averages in your industry. The benchmark is calculated based on anonymized data from customers in the same industry segment configured in Company Settings. Metrics compared:
MetricYour accountIndustry average
Cost per conversationcalculatedreference
Tokens per messagecalculatedreference
Resolution ratecalculatedreference
Average response timecalculatedreference
The benchmark requires the company’s industry to be configured in Settings > Company > Industry. Without this information, the comparison is not displayed.

​
Cost Recommendations

Based on usage data, the BI Dashboard generates automatic cost reduction recommendations, such as:
  • Migrate low-volume agents to more economical models
  • Reduce the context of conversations with long history
  • Deactivate agents unused in the last 30 days
Each recommendation includes an estimated monthly savings in tokens and cost.

​
Export

BI Dashboard data can be exported as CSV or XLSX. Click the Export report button in the upper-right corner of the panel. See more options at Report Export.

​
Required permissions

Exclusive access for owner and manager.