Your First Query
Greppilot allows you to query your connected data sources using natural language. The AI agent interprets your requests, generates SQL queries, and visualizes the results on your dashboard.
To perform your first query, ensure you have at least one integration connected and active. Supported database integration types include PostgreSQL, MySQL, MSSQL, and ClickHouse.
Querying your Data
Interact with Greppilot through the chat interface. Your input will be processed by the AI agent, which leverages available tools to fulfill your request.
Show me the total sales by month for the last year.
When providing a manual SQL query, the agent will attempt to execute it against your primary or most relevant connected database integration. If you have multiple database integrations, you can specify the target connector explicitly, e.g., (connector: "mysql") SELECT ....
How it Works
Upon receiving your query, the AI agent performs the following steps:
- Tool Selection: The agent evaluates your request and selects the most appropriate action. For data visualization, this typically means creating a new chart or table.
- SQL Generation: For natural language queries, the agent generates optimal SQL based on your connected database schemas. For manual SQL, it directly uses your provided query.
- Execution: The generated SQL is executed against your connected data source.
- Visualization: The results are rendered as a suitable chart (e.g., a
line-chartfor time series, abar-chartfor categorical data, or atablefor raw data).
When creating a chart, the agent uses the following key parameters:
| Parameter | Type | Description |
|---|---|---|
componentType | string | The type of visualization to create (e.g., "bar-chart", "line-chart", "table", "kpi-card", "markdown-card"). |
config | object | Configuration for the component, such as title, xAxis, yAxis, labelKey, valueKey, or syntax for diagrams. |
data | array | The array of result rows from the executed query. This field MUST NOT be empty for data components. |
query | object | Metadata about the query, including connector (e.g., "postgres", "mysql", "csv", "none"), sql, and naturalLanguage. |
The agent will respond with the generated chart directly on your dashboard, and a conversational summary in the chat interface. You can then further refine the chart, ask follow-up questions, or request additional analyses.