Core concepts

Understanding a few concepts will get you up to speed faster. How they fit together:

Attach datasets to a Space, start a conversation there to ask questions; save good results as a dashboard and share them, and turn frequent questions into a scheduled job pushed to an IM group.

Dataset

A data connection, or an uploaded file. On creation, the system reads its schema and sample data as the source for analysis. One dataset can be reused across multiple Spaces. See Datasets & connectors.

Space

The basic unit for analysis. A Space:

• attaches one or more datasets;
• stores business notes — table relations, metric definitions, field meanings — which the Agent consults;
• has its own conversation history and default tier.

Keeping related data, definitions, and conversations within one Space makes answers more consistent.

Conversation

A multi-turn Q&A within a Space. Each turn, you ask and the Agent returns its reasoning, SQL, a result table, and a chart. Conversations keep context, so you can follow up on the previous result, e.g. "now break it down by department". See Spaces & chat analysis.

Dashboard

A board that saves query results (charts, tables); you can generate a read-only share link. See Dashboards & sharing.

Scheduled monitoring job

A question that runs automatically on a set schedule. It runs on time, pushes the result to an IM group, and records the reason on failure. See Scheduled monitoring jobs.

Tier

The model tier used to answer, in three levels — Lite, Standard, Flagship — billed by usage. You can set a default tier per Space and switch temporarily in a single conversation. See Tiers & pricing.