Agent Usage & Actions

All pricing is based on actions, and every action corresponds to one tool call.

The Agent works by calling different internal tools to complete a user’s request. Each tool performs one action — like querying data, generating a visualization, or searching the web.

• 1 tool call = 1 action
• A typical question may trigger 3 to 5 tool calls, depending on complexity.

Type of tools

Examples

Example №1

Prompt: “Show me ROAS by campaign type over the last 90 days, and visualize it.”

This would typically include:

• Tool Call 1: Query data
• Tool Call 2: Transform data
• Tool Call 3: Generate a chart

Total: 3 actions

Example №2

Prompt: “Find average CPM for B2B SaaS in the US and compare to my campaigns.”

• Tool Call 1: Web search for CPM benchmark
• Tool Call 2: Query internal campaign data
• Tool Call 3: Merge and compare datasets
• Tool Call 4: Visualize comparison

Total: 4 actions

Example №3

Prompt: “Which Facebook creatives had the highest CTR last month?”

• Tool Call 1: Query ad performance data
• Tool Call 2: Filter by platform and date
• Tool Call 3: Rank by CTR
• Tool Call 4: Visualize Data

Total: 4 actions

Tracking Usage

You can track and audit your usage at any time in the Settings → Usage section of the platform. This dashboard shows how many actions have been used, broken down by date range, tool type, and workspace.

For each workspace, you’ll see visual charts that reflect how usage is trending over time. You can also identify which users are generating the most tool calls, and which types of prompts are driving usage.

The AI Agent’s usage model is simple: every tool call equals one action. Most real-world questions involve multiple steps — querying data, transforming it, visualizing it, or enriching it with external context — and typically use between 3 and 5 actions.