Natural-Language Chat

Natural-Language Chat

Chat lets you interrogate your data conversationally. Type a question in plain English and Kutta translates it into a query against your dataset, then answers with the most useful format. A table, a written summary, or a new chart. No SQL required.

Opening chat

Open chat from the dashboard's top navigation bar. Your conversation history for that dashboard loads automatically, so you can pick up where you left off.

Asking good questions

Be specific about the metric, the grouping, and the time frame. A few examples:

  • What was total revenue by region last quarter?
  • Which five products had the highest return rate?
  • Show me daily active users over the past 30 days.
  • Compare ad spend to conversions across channels.
TipIf a dataset was imported with a description, chat uses that context to better interpret column names and terminology specific to your data.

Response formats

Kutta picks the response format that best fits your question:

  • Inline table: row-level results rendered as a sortable, scrollable table.
  • Text summary: a written answer for questions that call for a number or explanation.
  • Artifact panel: larger result sets open in an expandable panel beside the conversation.

Generating charts from chat

When a question is best answered visually, you can turn the result into a chart that is added directly to your dashboard. A quick path from question to permanent visualization.

Conversation history

Each dashboard keeps its own chat history, persisted across sessions. You can scroll back through previous questions and answers, or clear the history to start fresh.