ThoughtSpot AI-Powered Search Analytics Cheat Sheet

Back to Business IntelligenceUpdated 2026-05-15

ThoughtSpot is a search-driven business intelligence platform that sits in the analytics layer of the modern data stack, enabling anyone to query live cloud warehouses by typing natural language questions instead of writing SQL. It solves the last-mile analytics problem — instead of waiting days for a new report, business users get instant self-service answers against live data in Snowflake, BigQuery, Databricks, and other cloud warehouses via its ThoughtSpot Embrace live-query engine. The key mental model is that ThoughtSpot separates data modeling from data consumption: analysts build governed semantic Models once, defining joins, column names, and row-level security rules, after which any user queries freely without ever seeing the underlying schema. The platform has evolved into a fully agentic analytics system — the Spotter AI analyst now integrates with Claude, ChatGPT, and any MCP-compatible client, making ThoughtSpot's semantic layer a trust anchor for enterprise AI pipelines.

What This Cheat Sheet Covers

This topic spans 12 focused tables and 109 indexed concepts. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.

Table 1: Search Bar FundamentalsTable 2: Search Keywords & OperatorsTable 3: Spotter AI AnalystTable 4: Liveboards & Dashboard ManagementTable 5: SpotIQ Automated Insight DiscoveryTable 6: Data Modeling — Models, Views & JoinsTable 7: Formulas & CalculationsTable 8: Security & Access ControlTable 9: TML Scripting & Version ControlTable 10: ThoughtSpot Everywhere — Embedded AnalyticsTable 11: REST API v2 Key EndpointsTable 12: Deployment, Administration & Platform Features

Table 1: Search Bar Fundamentals

Feature	Example	Description
Natural language search	`total revenue by region last year`	Type a question in plain English; the Relational Search Engine translates it to SQL and returns a chart or table automatically — no SQL knowledge required.
Search tokens	Each phrase appears as a bordered chip in the search bar	Search terms are wrapped in token boxes that visually separate each column, filter, or keyword, making it easy to click and edit individual terms.
Data source selection	Click Search data → choose a Model or Worksheet	Users pick which semantic Model to search before entering terms; every query is scoped to one data source.
Column types: measures vs attributes	`revenue` (measure) vs `region` (attribute)	Measures aggregate numerically; attributes group and filter results — misidentifying a column type is the most common cause of unexpected aggregations.

Table 1: Search Bar Fundamentals

Feature	Example	Description
Natural language search	`total revenue by region last year`	Type a question in plain English; the Relational Search Engine translates it to SQL and returns a chart or table automatically — no SQL knowledge required.
Search tokens	Each phrase appears as a bordered chip in the search bar	Search terms are wrapped in token boxes that visually separate each column, filter, or keyword, making it easy to click and edit individual terms.
Data source selection	Click Search data → choose a Model or Worksheet	Users pick which semantic Model to search before entering terms; every query is scoped to one data source.
Column types: measures vs attributes	`revenue` (measure) vs `region` (attribute)	Measures aggregate numerically; attributes group and filter results — misidentifying a column type is the most common cause of unexpected aggregations.