DemoPrep — Quick Start Guide

Overview

DemoPrep builds a complete ThoughtSpot demo from a company URL and a use case selection. It researches the company, generates a Snowflake database with realistic data, creates a ThoughtSpot semantic model, and deploys a finished liveboard — end to end in roughly 10–30 minutes depending on data size.

How to Run a Demo

Log In

Go to thoughtspot-dp-demoprep.hf.space and log in with your credentials. Your settings are saved per user and restored automatically.

Important: Always use the direct URL above. The HuggingFace wrapper URL blocks auth cookies and will prevent login.

Configure the Right Panel

The right panel has two always-visible dropdowns at the top, plus a collapsible ⚙️ Settings section for run options.

Always Visible

Control	What it does
TS Environment	The ThoughtSpot cluster to deploy to. Required — GO will not run without a selection.
AI Model	The LLM used for research, DDL generation, and liveboard questions. Defaults to the system default (GPT or Claude).

⚙️ Settings — click to expand run options

Field	What it does
Liveboard Name	Name for the created liveboard. Leave blank to auto-derive from the company URL (e.g. `Nike - Sales Analytics`).
Data Size	Controls how many rows are generated. Small — ~1,000 fact rows. Fast, good for testing or quick demos. Medium — ~10,000 fact rows. Richer data, better chart distributions. Takes longer.
Geo Scope	USA Only — US cities, states, regions in dimensional data. International — global locations. Use for global-brand prospects.
Tag Name	Optional ThoughtSpot tag applied to all created objects (model + liveboard). Leave blank to skip tagging.
Column Naming Style	How Snowflake column names are formatted: `Regular Case`, `snake_case`, `camelCase`, `PascalCase`, `UPPER_CASE`, or `original`.
Object Naming Prefix	Short prefix added to all TS object names (e.g. `acme` → `acme_CompanyModel`). Useful for multi-tenant or shared environments.
Share With	ThoughtSpot username or email to share the created liveboard with after deployment. Leave blank to skip.

All Settings values take effect immediately for the current run — no save or page reload needed.

Choose Your Demo Type

Defined — recommended for most demos

Use the 3-level cascade to pick a curated use case from the matrix.

Industry	e.g. Retail & Consumer Goods, Financial Services, Technology
Line of Business	e.g. Specialty Retail, Wealth Management, Software as a Service
Function	e.g. Sales, Supply Chain, Marketing, Finance

When the combination is in the matrix, you get curated KPIs, story-driven visualizations, and a target persona. If the combination is not fully defined, the AI adapts from the closest match.

Custom — for demos outside the matrix

Enter the company URL and describe the use case in plain language. The more context you provide, the better the liveboard story.

Hit GO

Press GO. The pipeline runs automatically through six stages:

Research AI reads the company website and builds a profile — industry, products, business context, and the metrics that matter for the use case.
DDL Schema generated for Snowflake: fact and dimension tables with realistic column names and types for the use case.
Data Snowflake database created and populated with AI-generated data. Small runs ~1k rows; Medium runs ~10k rows.
ThoughtSpot Semantic model built from the schema. Column descriptions, synonyms, and Sage search index added automatically.
Liveboard MCP creates the initial liveboard. Post-processing adds AI-humanized chart titles, KPI groups, donut charts, brand colors, and layout.
Complete Links to the model and liveboard appear in the Chat tab. Total time: Small ~10–15 min, Medium ~20–30 min.

Watch the Pipeline Status panel on the right for live progress. Detailed logs are in the AI Feedback tab.

Review Your Results

When the pipeline finishes, the Chat tab shows direct links to the model and liveboard in ThoughtSpot. The Demo Assets tab has:

SpotterViz TS	Matrix-grounded Spotter prompts you can paste into Spotter to walk through the demo story step by step.
SpotterViz AI	AI-generated alternative story flow based on the actual data and liveboard created.
Demo Pack	Talking points, insight bullets, and KPI context for the presentation.

Settings Tab

The Settings tab is for persistent configuration — credentials and defaults that carry across all runs. Per-run options (data size, geo scope, liveboard name, etc.) live in the ⚙️ Settings accordion on the main screen.

Section	What's here
TS Environment credentials	Trusted auth URL and secret key for each ThoughtSpot environment you have access to.
Default Run Inputs	Optional. Enable to have the app pre-fill Industry, Line of Business, Function, and Company URL at page load. Useful if you always demo the same use case.
Admin defaults	Admin-only. Set default values for AI model, TS environment, data size, etc. — these become the starting values everyone sees when they log in.

Run History Tab

Shows all your previous pipeline runs — company, use case, model, data size, stage reached, and outcome. Clicking a row shows the full stage-by-stage log for that run. Admin users see runs across all users with filtering by email.

Tips

Liveboard Name blank — auto-derives from the company URL (strips TLD, capitalizes). You can always rename in ThoughtSpot after the run.
Small vs Medium — start with Small for testing or when speed matters. Use Medium when you want richer data: 10k rows produce better trend lines, more meaningful KPIs, and more varied chart distributions.
Geo Scope — use International when the prospect is a global company; USA Only for US-focused demos or when you want cleaner geographic data.
Tag Name — great for keeping demo objects organized. All created objects (model + liveboard) get the tag automatically. Useful when running multiple demos on the same cluster.
Data Adjuster — paste any existing liveboard URL into the Chat tab prompt to skip the build pipeline and go straight to adjusting data values for an existing demo.
Default Run Inputs — in Settings, enable this to have the app pre-fill your favorite vertical/use case combination at every page load. Saves clicks when you run the same demo type repeatedly.

Troubleshooting

Problem	What to try
Pipeline hangs at ThoughtSpot	Check the Deploy Log in the AI Feedback tab. MCP has a 5-minute timeout and will fail gracefully — the pipeline will report the error and stop.
Liveboard has no data or wrong data	Verify the TS Environment dropdown matches where you expected the data to land. Each environment is its own Snowflake + ThoughtSpot pair.
"Auth key not set" error	Select a TS environment from the dropdown before hitting GO. The pipeline requires an environment selection — it will not run without one.
Login issues	Use the direct URL `https://thoughtspot-dp-demoprep.hf.space` — the HuggingFace wrapper URL blocks auth cookies.
Data generation timed out	Try switching to Small data size. Medium runs (~10k rows) can take 20–30 minutes for data generation.
Insert errors in Data stage	Usually a schema type mismatch. Try running again — the AI regenerates the DDL on retry and typically resolves the conflict.