chore: README, e2e grader, share-permission fix, sprint + quality results

- README: chat_interface.py is the full app (App tab primary), app.py entry point; updated structure/config
- tests/e2e_quality.py: date-aware grader, COMPLETENESS rubric rewrite, --ts-environment override, hard-fail on missing dropdown label
- thoughtspot_deployer.py: share-permission 400 fix — notify_on_share=false + empty message to satisfy the GraphQL $message contract
- sprint_2026_04.md: progress updates
- tests/quality_results: run artifacts incl. 2026-07-28 82.9/B run

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

README.md

@@ -11,48 +11,55 @@ license: mit
 python_version: "3.11"
 ---
 
-# Demo Wire - AI-Powered Demo Builder
+# DemoPrep — AI-Powered ThoughtSpot Demo Builder
 
-A powerful Gradio-based application that automatically generates and deploys complete demo environments for ThoughtSpot, including Snowflake schemas, data population, and semantic model creation.
+A Gradio application that generates and deploys complete ThoughtSpot demo environments end to end — company research, an LLM-authored data blueprint, deterministic data generation, a Snowflake schema + data load, and a ThoughtSpot semantic model and liveboard.
 
-> **Deployed on Hugging Face Spaces**: This app can be deployed to HF Spaces for easy access. See [DEPLOYMENT.md](DEPLOYMENT.md) for setup instructions.
+> **Deployed on Hugging Face Spaces.** `app.py` is the entry point. See [DEPLOYMENT.md](DEPLOYMENT.md) for setup.
+
+## 🖥️ The interface
+
+The app is a **tabbed Gradio UI**, and the **App tab is the primary way to build a demo**: fill in a short form and press **GO**, and the full pipeline runs automatically with live progress.
+
+> **Naming note:** the whole application lives in **`chat_interface.py`** — that name is *historical*. It is **not** just a chat interface; that single file hosts every tab (App, Chat, Admin, Settings, Run History, …), and the **App tab is the main path**. A **Chat tab** offers an alternative conversational flow that runs the same pipeline. `app.py` simply configures and launches this app.
 
 ## 🚀 Features
 
-- **AI-Powered Research**: Automatically researches companies and industries using LLM
-- **Schema Generation**: Creates optimized Snowflake DDL based on business context
-- **Data Population**: Generates realistic demo data with strategic outliers
-- **ThoughtSpot Integration**: Deploys connections, tables, and semantic models
-- **Interactive UI**: Clean, intuitive Gradio interface with real-time progress updates
+- **AI-powered research** — researches the company and industry with an LLM
+- **Single blueprint pipeline** — one path, no keyword routing or fallback: research → LLM-authored `DemoBlueprint` → deterministic engine → validation → derived DDL → load
+- **Realistic data** — bounded/related measures (a part never exceeds its whole, rates stay ≤ 1, prices are stable per entity) plus planted, discoverable "Demo to Win" insights the validator proves are visible
+- **ThoughtSpot integration** — deploys the connection, tables, semantic model, and an enhanced liveboard
+- **Interactive UI** — tabbed Gradio app with real-time progress
 
 ## 🛠️ Tech Stack
 
 - **Frontend**: Gradio (Python web UI)
-- **Backend**: Python 3.12+
-- **Database**: Snowflake
-- **Analytics**: ThoughtSpot
-- **AI**: OpenAI GPT models
-- **Data Generation**: Faker library
+- **Backend**: Python 3.11
+- **Database**: Snowflake (keypair auth)
+- **Analytics**: ThoughtSpot (per-environment trusted auth)
+- **AI**: multiple LLM providers (OpenAI and Anthropic/Claude), routed via `llm_config.py`
+- **Settings**: Supabase (admin + per-user settings)
 
 ## 📋 Prerequisites
 
-- Python 3.12+
-- Snowflake account with appropriate permissions
+- Python 3.11
+- Snowflake account (keypair configured)
 - ThoughtSpot Cloud account
-- OpenAI API key
+- An LLM API key (OpenAI and/or Anthropic)
+- Supabase project (settings storage)
 
 ## 🚀 Quick Start
 
 1. **Clone the repository**
    ```bash
-   git clone https://github.com/yourusername/demo-wire.git
-   cd demo-wire
+   git clone <repo-url>
+   cd demoprep
    ```
 
-2. **Set up virtual environment**
+2. **Set up a virtual environment**
    ```bash
-   python -m venv demo_wire
-   source demo_wire/bin/activate  # On Windows: demo_wire\Scripts\activate
+   python -m venv .venv
+   source .venv/bin/activate   # Windows: .venv\Scripts\activate
    ```
 
 3. **Install dependencies**
@@ -60,82 +67,80 @@ A powerful Gradio-based application that automatically generates and deploys com
    pip install -r requirements.txt
    ```
 
-4. **Configure environment variables**
+4. **Configure bootstrap environment variables** (see Configuration)
    ```bash
    cp .env.example .env
-   # Edit .env with your credentials
+   # edit .env
    ```
 
 5. **Run the application**
    ```bash
-   python demo_prep.py
+   python app.py
    ```
+   `app.py` launches the full Gradio app defined in `chat_interface.py`.
 
 6. **Open your browser**
-   Navigate to `http://localhost:7860`
+   Navigate to `http://localhost:7860` and use the **App** tab.
 
 ## ⚙️ Configuration
 
-Create a `.env` file with the following variables:
+`.env` holds only the **bootstrap** secrets needed to start the app and reach Supabase and the LLM providers. All other credentials — the Snowflake keypair and the ThoughtSpot **per-environment trusted-auth keys** — live in **Supabase admin settings** and are loaded at runtime.
 
 ```env
-# OpenAI
-OPENAI_API_KEY=your_openai_api_key
-
-# Snowflake
-SNOWFLAKE_USER=your_username
-SNOWFLAKE_PASSWORD=your_password
-SNOWFLAKE_ACCOUNT=your_account
-SNOWFLAKE_WAREHOUSE=your_warehouse
-SNOWFLAKE_DATABASE=your_database
-SNOWFLAKE_SCHEMA=your_schema
-
-# ThoughtSpot
-THOUGHTSPOT_URL=your_thoughtspot_url
-THOUGHTSPOT_USERNAME=your_username
-THOUGHTSPOT_PASSWORD=your_password
+# Supabase (settings storage)
+SUPABASE_URL=...
+SUPABASE_ANON_KEY=...
+
+# LLM providers
+OPENAI_API_KEY=...
+GOOGLE_API_KEY=...            # optional (Gemini)
 
 # Slack deployment notifications (optional, outbound-only)
-SLACK_BOT_TOKEN=xoxb-your_bot_token
+SLACK_BOT_TOKEN=xoxb-...
 SLACK_DEPLOYMENT_CHANNEL_ID=C0123456789
 ```
 
-Slack notifications use the Slack Web API to post deployment status messages from
-DemoPrep into one approved channel. This path is outbound-only: no Socket Mode,
-event subscriptions, slash commands, or public Slack request URL are required.
-The Slack app only needs the `chat:write` bot scope, and the bot must be invited
-to the target channel.
+Slack notifications use the Slack Web API to post deployment status into one approved channel. This path is outbound-only: no Socket Mode, event subscriptions, slash commands, or public request URL. The Slack app needs only the `chat:write` bot scope and must be invited to the target channel.
 
 ## 🎯 Usage
 
-1. **Enter Company Information**: Company name, URL, and industry
-2. **Start Research**: AI analyzes the company and industry
-3. **Create DDL**: Generate optimized Snowflake schema
-4. **Generate Data**: Create realistic demo data with outliers
-5. **Deploy**: Deploy to Snowflake and ThoughtSpot
+The **App tab** is the primary interface — fill the form, press **GO**, and the pipeline runs end to end:
+
+1. **Fill the form** — vertical / line of business / function (or a custom use case), the company URL, and the target **ThoughtSpot environment**
+2. **Press GO** — research → blueprint → data generation → validation → Snowflake load → ThoughtSpot model + liveboard
+3. **Review** — model and liveboard links, a Demo Pack, and a Spotter Viz story appear on completion
 
 ## 📁 Project Structure
 
 ```
-demo-wire/
-├── demo_prep.py              # Main Gradio application
-├── schema_utils.py           # Schema parsing and generation utilities
-├── thoughtspot_deployer.py   # ThoughtSpot deployment logic
-├── snowflake_auth.py         # Snowflake authentication
-├── demo_personas.py          # Demo persona configurations
-├── prompts.py                # LLM prompt templates
+demoprep/
+├── app.py                    # Entry point — launches the Gradio app on :7860
+├── chat_interface.py         # The full Gradio app (App/Chat/Admin/Settings/…) — historical name
+├── thoughtspot_deployer.py   # ThoughtSpot deploy: connection, tables, model, liveboard
+├── liveboard_creator.py      # MCP liveboard creation + TML post-processing (enhance_mcp_liveboard)
+├── snowflake_auth.py         # Snowflake keypair authentication
+├── llm_config.py             # LLM provider/model routing (single source of truth)
+├── demo_personas.py          # Vertical × Function use-case configs
+├── demoprep_app/             # The demo-generation pipeline package
+│   ├── pipeline/build_demo.py    #   the single build entry point
+│   ├── scenario/                 #   blueprint contract + LLM authoring + directives
+│   ├── dataset/                  #   deterministic engine + validator
+│   ├── ddl/                      #   Snowflake DDL derived from the dataset
+│   └── integrations/snowflake/   #   Snowflake row loader
 ├── requirements.txt          # Python dependencies
 ├── docs/                     # Documentation
-├── tests/                    # Test files
+├── tests/                    # Unit tests + the e2e quality harness (e2e_quality.py)
 └── results/                  # Generated demo results
 ```
 
 ## 🧪 Testing
 
-Run the test suite:
-
 ```bash
+# Unit tests (pytest collects test_*.py)
 python -m pytest tests/
+
+# End-to-end quality harness — drives a running app via the browser and grades the output
+python tests/e2e_quality.py --env-name test
 ```
 
 ## 🤝 Contributing
@@ -154,16 +159,16 @@ This project is licensed under the MIT License - see the [LICENSE](LICENSE) file
 
 - ThoughtSpot for the analytics platform
 - Snowflake for the data warehouse
-- OpenAI for the AI capabilities
+- OpenAI and Anthropic for the AI capabilities
 - Gradio for the web interface
 
 ## 📚 Development Notes
 
-Development notes and sprint planning are stored in `dev_notes/` (not committed to version control).
+Sprint planning and working notes live in `dev_notes/` (gitignored); architecture and handoff docs are in `docs/`.
 
 ## 📞 Support
 
-For support, email support@demo-wire.com or create an issue in this repository.
+For support, create an issue in this repository.
 
 ---
 

sprint_2026_04.md

@@ -267,6 +267,7 @@ should tell. KPI targets, growth trends, and anomaly patterns live in the matrix
 - [x] **AI Viz Titles** — `_humanize_viz_titles()` in `liveboard_creator.py`; one LLM call renames all raw TS column-name titles to business-readable labels; runs as Step 6.5 in `enhance_mcp_liveboard()` before TML re-import ✅
 - [x] **KPI conversion fix** — Step 3.5 no longer promotes "by X" dimensional breakdowns to KPIs (was putting "Averagesellingprice by Category Weekly" in the Key Metrics group) ✅
 - [x] **Chart variety** — multi-dim breakdowns ("by X and Y") → STACKED_COLUMN; single-dim categorical breakdowns now catches LINE charts (MCP often generates LINE for categoricals, blocking donut conversion) ✅
+- [x] **Share fix** — `share_objects()` in `thoughtspot_deployer.py` was failing with `400 "Variable \"$message\" of required type \"String!\" was not provided"` on both model + liveboard auto-share (surfaced on Nike run, se-thoughtspot staging). The REST 2.0 `/security/metadata/share` endpoint proxies to a GraphQL mutation that requires a non-null `message`; payload now sends `message: ""` + `notify_on_share: False` ✅
 
 ### Shipped at end of Sprint 3 / mini sprint (Apr 28-29)
 
@@ -286,4 +287,4 @@ should tell. KPI targets, growth trends, and anomaly patterns live in the matrix
 
 ---
 
-*Last updated: April 29, 2026 (late evening — tag v2 fix, claude-opus-4-6, viz titles sprint item)*
+*Last updated: July 23, 2026 — share fix (`$message` required-variable 400 on model + liveboard auto-share)*

tests/e2e_quality.py

@@ -72,7 +72,7 @@ RUN_SETTINGS = {
     "share_with":         "mike.boone@thoughtspot.com",
     "geo_scope":          "USA Only",
     "ai_model":           "claude-sonnet-4-6",          # model used for this test run
-    "ts_environment":     "secloud - primary",          # change to test other TS instances
+    "ts_environment":     "sebe - se",                   # se-cloud having issues; run on sebe
 }
 
 
@@ -949,11 +949,14 @@ def grade_data_quality(company: str, vertical: str, line: str, function: str,
             "failure at this volume, but it should reduce realism/story quality.\n"
         )
 
+    today = datetime.now().strftime("%Y-%m-%d")
     prompt = f"""You are grading a ThoughtSpot demo dataset.
 
 Company: {company}
 Vertical: {vertical} / {line}
 Analytics function: {function}
+Today's date is {today}. Treat any date on or before today as HISTORICAL — do NOT
+penalize current-year or recent dates as "future-dated"; the demo is built to run today.
 
 The goal is a compelling demo with realistic data, outliers that drive a narrative,
 and a schema that supports the key KPIs for this use case.
@@ -972,11 +975,17 @@ Score based on what you can observe.
 1. REALISM (20 pts): Values look like real {company} data at realistic scale and ranges.
 2. STORY POTENTIAL (30 pts): Outliers, trends, or anomalies exist that anchor a demo narrative.
 3. TIME COVERAGE (20 pts): 12–24 months of history with meaningful trends over time.
+   Judge coverage relative to today's date above — recent/current-year data is historical,
+   not "future"; only genuinely implausible far-future dates should count against this.
 4. SCHEMA FITNESS (15 pts): Star schema design supports the key KPIs for {line} {function}.
-5. COMPLETENESS (15 pts): Tables are populated. Dimensions have 20+ distinct members.
+5. COMPLETENESS (15 pts): Fact tables are well-populated (thousands of rows) with variation
+   across dimensions. Dimensions have REALISTIC cardinality for what they represent — a
+   handful of values for a naturally-small dimension (channel, region, tier, segment) is
+   CORRECT and must NOT be penalized; entity dimensions (products, customers, accounts,
+   stores) may have many. Do NOT require any fixed member count.
    RULE: If the row counts above show any key table at 0 rows, score COMPLETENESS = 0 for
    that criteria. If the fact table is 0 rows, also deduct heavily from STORY POTENTIAL.
-   PENALTY: If there are a few generated-looking dimension labels with numeric suffixes
+   PENALTY: If there are generated-looking dimension labels with numeric suffixes
    such as "North Corridor Route 31", "Customer 17", or "Product 42", penalize realism.
    If this pattern is repeated or widespread, the data should fail.
 

tests/quality_results/2026-07-23_21-14-12_quality_run.md

@@ -0,0 +1,42 @@
+# DemoPrep Quality Run — 2026-07-23T21:14
+**Target:** https://thoughtspot-dp-test-demoprep.hf.space  |  **Avg:** 56.4/100  Grade: D
+
+| Company | Use Case | Data | LB | Total | Note |
+|---------|----------|------|----|-------|------|
+| Gap | Retail & Consumer Goods / Fashio | n/a | n/a | 16.0/F — | ⏰ timeout |
+| Accenture | Custom | 62 | 72 | 74.0/C [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/50ebee01-e68c-47ed-bf7c-485a575ffe2d) | ✅ solid |
+| Dynatrace | Technology / Software as a Servi | 62 | 72 | 74.0/C [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/43ea010a-ea4a-4ae1-8924-119dab5e451f) | ✅ solid |
+| Chipotle | Travel & Hospitality / Restauran | 58 | 72 | 72.0/C [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/1feb215c-a613-4fce-920a-e5fcccd0e7a3) | ✅ solid |
+| Wells Fargo | Financial Services / Banking / M | 52 | 72 | 69.0/C [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/46247799-e75e-41de-a1cb-de4dd370337d) | ✅ solid |
+| Johnson & Johnson | Healthcare & Life Sciences / Lif | 42 | 72 | 64.0/C [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/180815ea-9124-41f0-a267-aa0db787c710) | ✅ solid |
+| Nike | Retail & Consumer Goods / Fashio | 62 | 72 | 74.0/C [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/742a52c0-a92e-4975-8354-069840d928ba) | ✅ solid |
+| PwC | Custom | n/a | n/a | 8.0/F — | ⏰ timeout |
+
+## Issues
+
+- **Gap**: TIMEOUT — last event: unhandled pipeline exception
+- **PwC**: TIMEOUT — last event: run failed
+
+## Data Quality Weaknesses
+
+**Accenture** (data=62/100):
+  - Critical data integrity failure: BIDS_WON exceeds BIDS_SUBMITTED in multiple rows, making win rates mathematically impos
+  - TOTAL_VALUE_WON_USD exceeds TOTAL_VALUE_SUBMITTED_USD in some rows, which is logically impossible
+**Dynatrace** (data=62/100):
+  - Critical data integrity flaw: OPTIMIZED_SPEND_USD frequently exceeds CLOUD_SPEND_USD by large multiples (up to 12x), whi
+  - REALIZED_SAVINGS_USD is deeply negative in many rows (e.g., -$206,079, -$49,927), which contradicts the metric's name an
+**Chipotle** (data=58/100):
+  - BOUNCE_RATE exceeds 1.0 in multiple rows (1.03, 1.75, 1.50) — a mathematically impossible value that would immediately d
+  - CLICK_TO_ORDER_RATE of 1.83 in row 4 is impossible (orders cannot exceed clicks); DIGITAL_ORDERS_INITIATED=1,650 vs CLIC
+**Wells Fargo** (data=52/100):
+  - Critical data integrity errors: FILL_RATE >1.0, APPROVAL_RATE >1.0, and BOUNCE_RATE >1.0 are mathematically impossible a
+  - BOUNCED_SESSIONS exceeding LANDING_PAGE_SESSIONS in at least one row is an impossible combination
+**Johnson & Johnson** (data=42/100):
+  - Critical completeness failure: DIM_ACCOUNT, DIM_PRODUCT, and DIM_SALES_REP each have only 10 rows — well below the 20+ m
+  - With only 10 reps, 10 products, and 10 accounts, the 3,333 fact rows are highly repetitive combinations that undermine a
+**Nike** (data=62/100):
+  - All dimension tables fall well below the 20-member threshold required for completeness scoring: DIM_PRODUCT=12, DIM_CHAN
+  - COGS_PER_UNIT values are inconsistent and sometimes implausibly low (e.g., $13.61–$21.18 for what should be a $124–$156 
+
+---
+*JSON: 2026-07-23_21-14-12_quality_run.json*

tests/quality_results/2026-07-24_01-51-18_quality_run.md

@@ -0,0 +1,27 @@
+# DemoPrep Quality Run — 2026-07-24T01:51
+**Target:** https://thoughtspot-dp-test-demoprep.hf.space  |  **Avg:** 12.5/100  Grade: F
+
+| Company | Use Case | Data | LB | Total | Note |
+|---------|----------|------|----|-------|------|
+| PNC Financial | Financial Services / Banking / H | n/a | n/a | 8.0/F — | ⏰ timeout |
+| Deloitte | Custom | n/a | n/a | 8.0/F — | ⏰ timeout |
+| Nike | Retail & Consumer Goods / Fashio | n/a | n/a | 20.0/F — | ⏰ timeout |
+| FedEx | Transportation & Logistics / Shi | n/a | n/a | 8.0/F — | ⏰ timeout |
+| General Mills | Retail & Consumer Goods / Grocer | n/a | n/a | 20.0/F — | ⏰ timeout |
+| Wells Fargo | Financial Services / Banking / M | n/a | n/a | 20.0/F — | ⏰ timeout |
+| J.B. Hunt | Transportation & Logistics / Tru | n/a | n/a | 8.0/F — | ⏰ timeout |
+| EY | Custom | n/a | n/a | 8.0/F — | ⏰ timeout |
+
+## Issues
+
+- **PNC Financial**: TIMEOUT — last event: run failed
+- **Deloitte**: TIMEOUT — last event: run failed
+- **Nike**: TIMEOUT — last event: run failed
+- **FedEx**: TIMEOUT — last event: run failed
+- **General Mills**: TIMEOUT — last event: run failed
+- **Wells Fargo**: TIMEOUT — last event: run failed
+- **J.B. Hunt**: TIMEOUT — last event: run failed
+- **EY**: TIMEOUT — last event: run failed
+
+---
+*JSON: 2026-07-24_01-51-18_quality_run.json*

tests/quality_results/2026-07-24_05-49-30_quality_run.md

@@ -0,0 +1,43 @@
+# DemoPrep Quality Run — 2026-07-24T05:49
+**Target:** https://thoughtspot-dp-test-demoprep.hf.space  |  **Avg:** 81.9/100  Grade: B
+
+| Company | Use Case | Data | LB | Total | Note |
+|---------|----------|------|----|-------|------|
+| Landstar | Transportation & Logistics / Tru | 72 | 74 | 79.5/B [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/0c4997a8-693e-4b77-953c-c74c3c331fab) | 🏆 great |
+| Lear | Manufacturing / Automotive / Sal | 72 | 72 | 79.0/B [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/00b625ea-be9c-457f-a1b0-880a5a3ba4d4) | 🏆 great |
+| Wells Fargo | Financial Services / Banking / M | 78 | 72 | 82.0/B [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/b10e50e0-a4ec-4fbd-aac2-cae593db3c66) | 🏆 great |
+| Nike | Retail & Consumer Goods / Fashio | 82 | 72 | 84.0/B [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/833be44d-b9ca-43ae-822b-fb96155cfe48) | 🏆 great |
+| Deloitte | Custom | 82 | 72 | 84.0/B [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/5338db4d-2b4a-415c-a4d2-10c92ee36c25) | 🏆 great |
+| Hyatt | Travel & Hospitality / Hotels /  | 82 | 72 | 84.0/B [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/01365018-1def-4e1a-a8c3-c46c2581bec2) | 🏆 great |
+| PwC | Custom | 82 | 72 | 84.0/B [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/cde36441-ff38-49e4-b13f-a45d046d917e) | 🏆 great |
+| Edward Jones | Financial Services / Asset & Wea | 72 | 72 | 79.0/B [lb](https://se-thoughtspot-cloud.thoughtspot.cloud/#/pinboard/9b17aa08-27cc-4c0b-87fe-4ff2f30cc12f) | 🏆 great |
+
+## Data Quality Weaknesses
+
+**Landstar** (data=72/100):
+  - Critical data integrity failure: SPOT_LOADS + CONTRACT_LOADS frequently do not sum to TOTAL_LOADS_AGENT (e.g., 203+480=6
+  - CAPACITY_COVERAGE_RATE exceeds 1.0 in multiple rows (1.14, 1.26, 1.18, 1.29) without clear explanation — this metric sho
+**Lear** (data=72/100):
+  - All three fact tables have exactly 3,333 rows — identical counts strongly suggest templated/synthetic generation rather 
+  - FACT_DELIVERY_QUALITY model joins only include manufacturing region, missing OEM customer and vehicle platform joins tha
+**Wells Fargo** (data=78/100):
+  - FULL_FUNNEL_CONVERSION_RATE is 0.00 across all sampled rows — this appears to be a broken or miscalculated metric, which
+  - Time coverage extends only to late 2025 (roughly 15 months from earliest visible date), falling short of the preferred 2
+**Nike** (data=82/100):
+  - Only 24 products in the PRODUCT dimension limits assortment breadth stories and makes product-level drill-downs feel con
+  - Only 6 promotional events in the PROMO table is thin for demonstrating promotional effectiveness analysis across a full 
+**Deloitte** (data=82/100):
+  - All three fact tables have exactly 3,333 rows — the identical count across independent fact tables looks artificially ge
+  - Sample data only shows dimension keys (not label values), so actual quality of dimension member names (e.g., whether pra
+**Hyatt** (data=82/100):
+  - Only 18 properties in the PROPERTY dimension is thin for a global hotel company demo — limits regional/brand/tier breakd
+  - Only 12 corporate accounts constrains the AR Collections narrative — cannot demonstrate meaningful segmentation across l
+**PwC** (data=82/100):
+  - STANDARD_FEE_USD values repeat across engagements (e.g., $1,021,325.29 appears multiple times), suggesting fees are set 
+  - CLIENT_ACCOUNT dimension at only 15 rows is sparse for a firm-wide PwC advisory demo; limits client-level storytelling
+**Edward Jones** (data=72/100):
+  - All three fact tables have exactly 3,333 rows — identical counts are an obvious generation artifact that undermines real
+  - FACT_ADVISOR_PRODUCTIVITY joining to DIM_REVENUE_TYPE_KEY is semantically questionable; advisor productivity is typicall
+
+---
+*JSON: 2026-07-24_05-49-30_quality_run.json*

tests/quality_results/2026-07-24_16-37-19_quality_run.md

@@ -0,0 +1,27 @@
+# DemoPrep Quality Run — 2026-07-24T16:37
+**Target:** https://thoughtspot-dp-test-demoprep.hf.space  |  **Avg:** 4.2/100  Grade: F
+
+| Company | Use Case | Data | LB | Total | Note |
+|---------|----------|------|----|-------|------|
+| Wipro | Technology / IT Services / Marke | n/a | n/a | 2.0/F — | ⏰ timeout |
+| Shake Shack | Travel & Hospitality / Restauran | n/a | n/a | 2.0/F — | ⏰ timeout |
+| Nike | Retail & Consumer Goods / Fashio | n/a | n/a | 8.0/F — | ⏰ timeout |
+| Wells Fargo | Financial Services / Banking / M | n/a | n/a | 8.0/F — | ⏰ timeout |
+| Landstar | Transportation & Logistics / Tru | n/a | n/a | 8.0/F — | ⏰ timeout |
+| UPS | Transportation & Logistics / Shi | n/a | n/a | 2.0/F — | ⏰ timeout |
+| McKinsey | Custom | n/a | n/a | 2.0/F — | ⏰ timeout |
+| Accenture | Custom | n/a | n/a | 2.0/F — | ⏰ timeout |
+
+## Issues
+
+- **Wipro**: TIMEOUT — last event: run failed
+- **Shake Shack**: TIMEOUT — last event: run failed
+- **Nike**: TIMEOUT — last event: run failed
+- **Wells Fargo**: TIMEOUT — last event: run failed
+- **Landstar**: TIMEOUT — last event: run failed
+- **UPS**: TIMEOUT — last event: run failed
+- **McKinsey**: TIMEOUT — last event: run failed
+- **Accenture**: TIMEOUT — last event: run failed
+
+---
+*JSON: 2026-07-24_16-37-19_quality_run.json*

tests/quality_results/2026-07-27_12-47-19_quality_run.md

@@ -0,0 +1,27 @@
+# DemoPrep Quality Run — 2026-07-27T12:47
+**Target:** https://thoughtspot-dp-test-demoprep.hf.space  |  **Avg:** 20.0/100  Grade: F
+
+| Company | Use Case | Data | LB | Total | Note |
+|---------|----------|------|----|-------|------|
+| NetSuite | Technology / Software as a Servi | n/a | n/a | 20.0/F — | ⏰ timeout |
+| General Mills | Retail & Consumer Goods / Grocer | n/a | n/a | 20.0/F — | ⏰ timeout |
+| PwC | Custom | n/a | n/a | 20.0/F — | ⏰ timeout |
+| Zendesk | Technology / Software as a Servi | n/a | n/a | 20.0/F — | ⏰ timeout |
+| Yum Brands | Travel & Hospitality / Restauran | n/a | n/a | 20.0/F — | ⏰ timeout |
+| Nike | Retail & Consumer Goods / Fashio | n/a | n/a | 20.0/F — | ⏰ timeout |
+| Wells Fargo | Financial Services / Banking / M | n/a | n/a | 20.0/F — | ⏰ timeout |
+| Accenture | Custom | n/a | n/a | 20.0/F — | ⏰ timeout |
+
+## Issues
+
+- **NetSuite**: TIMEOUT — last event: run failed
+- **General Mills**: TIMEOUT — last event: run failed
+- **PwC**: TIMEOUT — last event: run failed
+- **Zendesk**: TIMEOUT — last event: run failed
+- **Yum Brands**: TIMEOUT — last event: run failed
+- **Nike**: TIMEOUT — last event: run failed
+- **Wells Fargo**: TIMEOUT — last event: run failed
+- **Accenture**: TIMEOUT — last event: run failed
+
+---
+*JSON: 2026-07-27_12-47-19_quality_run.json*

tests/quality_results/2026-07-27_19-23-28_quality_run.md

@@ -0,0 +1,27 @@
+# DemoPrep Quality Run — 2026-07-27T19:23
+**Target:** https://thoughtspot-dp-test-demoprep.hf.space  |  **Avg:** 18.5/100  Grade: F
+
+| Company | Use Case | Data | LB | Total | Note |
+|---------|----------|------|----|-------|------|
+| Nike | Retail & Consumer Goods / Fashio | n/a | n/a | 20.0/F — | ⏰ timeout |
+| Darden | Travel & Hospitality / Restauran | n/a | n/a | 20.0/F — | ⏰ timeout |
+| XPO Logistics | Transportation & Logistics / Tru | n/a | n/a | 20.0/F — | ⏰ timeout |
+| FedEx | Transportation & Logistics / Shi | n/a | n/a | 20.0/F — | ⏰ timeout |
+| McKinsey | Custom | n/a | n/a | 20.0/F — | ⏰ timeout |
+| EY | Custom | n/a | n/a | 20.0/F — | ⏰ timeout |
+| Wells Fargo | Financial Services / Banking / M | n/a | n/a | 20.0/F — | ⏰ timeout |
+| Splunk | Technology / Software as a Servi | n/a | n/a | 8.0/F — | ⏰ timeout |
+
+## Issues
+
+- **Nike**: TIMEOUT — last event: run failed
+- **Darden**: TIMEOUT — last event: run failed
+- **XPO Logistics**: TIMEOUT — last event: run failed
+- **FedEx**: TIMEOUT — last event: run failed
+- **McKinsey**: TIMEOUT — last event: run failed
+- **EY**: TIMEOUT — last event: run failed
+- **Wells Fargo**: TIMEOUT — last event: run failed
+- **Splunk**: TIMEOUT — last event: run failed
+
+---
+*JSON: 2026-07-27_19-23-28_quality_run.json*

tests/quality_results/2026-07-28_02-18-45_quality_run.md

@@ -0,0 +1,43 @@
+# DemoPrep Quality Run — 2026-07-28T02:18
+**Target:** https://thoughtspot-dp-test-demoprep.hf.space  |  **Avg:** 82.9/100  Grade: B
+
+| Company | Use Case | Data | LB | Total | Note |
+|---------|----------|------|----|-------|------|
+| Deloitte | Custom | 78 | 78 | 83.5/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/ec141083-908e-42db-902c-3e54a3126062) | 🏆 great |
+| Nike | Retail & Consumer Goods / Fashio | 78 | 74 | 82.5/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/a8d011b3-c221-4898-bd1c-e35018862598) | 🏆 great |
+| Wells Fargo | Financial Services / Banking / M | 72 | 72 | 79.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/2d211fc5-8008-4fe6-9dba-85d411656623) | 🏆 great |
+| Splunk | Technology / Software as a Servi | 78 | 72 | 82.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/0d329d55-758c-4d1d-9ac8-5e84fc4fb9c8) | 🏆 great |
+| McKinsey | Custom | 84 | 72 | 85.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/c709d276-d37e-4ff6-8743-6359f8e5f437) | 🏆 great |
+| Zendesk | Technology / Software as a Servi | 82 | 72 | 84.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/e3b2e2a4-6b87-4f2f-bd76-2be87c0814e6) | 🏆 great |
+| Wipro | Technology / IT Services / Marke | 84 | 72 | 85.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/cbab7edc-f92a-4f30-a4d6-a9eeff433ef6) | 🏆 great |
+| LG | Retail & Consumer Goods / Consum | 78 | 72 | 82.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/d6805e4f-c372-4660-a4c8-40750bf1fd1e) | 🏆 great |
+
+## Data Quality Weaknesses
+
+**Deloitte** (data=78/100):
+  - Only 12 engagement dimension records driving 3,333 fact rows creates excessive repetition per engagement (~278 rows per 
+  - FACT_PIPELINE_PROPOSALS sample data was not visible in the provided output, making it impossible to validate pipeline wi
+**Nike** (data=78/100):
+  - Only 23 products is thin for a Nike demo — limits cross-product comparison depth and makes the catalog feel narrow
+  - FACT_INVENTORY_SELLTHROUGH joins only to CHANNEL in the model despite having REGION_KEY in the table, blocking regional 
+**Wells Fargo** (data=72/100):
+  - Row 2 has a critical internal inconsistency: 3,921 accounts activated with only $5,575 total first-year revenue ($1.42/a
+  - Most recent sample date visible is March 2026, leaving a ~4-month gap to today (July 2026); unclear if current-quarter d
+**Splunk** (data=78/100):
+  - Only 12 customer accounts is a very thin customer dimension — limits cross-account segmentation and makes patterns feel 
+  - The extremely high DOWNTIME_MINUTES of 357.40 in one row (vs. typical 16–86) may read as a data error rather than a narr
+**McKinsey** (data=84/100):
+  - STANDARD_BILLING_RATE_USD appears to be a fixed per-partner attribute repeated in the fact table rather than varying by 
+  - All three fact tables have exactly 3,333 rows — the identical count across tables feels mechanical and slightly artifici
+**Zendesk** (data=82/100):
+  - PIPELINE_COVERAGE_RATIO is identical at exactly 3.23 across all visible rows — eliminates any meaningful trend or narrat
+  - FACT_SALES table does not appear to join to PRODUCT or REGION dimensions based on the schema shown, limiting product and
+**Wipro** (data=84/100):
+  - Only 10 campaigns for a global IT services firm feels thin and limits drill-down variety
+  - No account-level or named-account dimension despite ABM being a listed campaign type
+**LG** (data=78/100):
+  - Zero-value rows (LISTED_PRICE=0, CHANNEL_SELLING_PRICE=0, ASP_VS_COMPETITOR_INDEX=0) in FACT_CHANNEL_ASP_TRACKER appear 
+  - COMPETITOR_INDEX_PRICE of $0.00 for product 6 across multiple rows zeros out the ASP vs. competitor metric for that SKU,
+
+---
+*JSON: 2026-07-28_02-18-45_quality_run.json*

tests/quality_results/latest_test_summary.md

@@ -1,27 +1,43 @@
-# DemoPrep Quality Run — 2026-06-23T03:10
-**Target:** https://thoughtspot-dp-test-demoprep.hf.space  |  **Avg:** 20.2/100  Grade: F
+# DemoPrep Quality Run — 2026-07-28T02:18
+**Target:** https://thoughtspot-dp-test-demoprep.hf.space  |  **Avg:** 82.9/100  Grade: B
 
 | Company | Use Case | Data | LB | Total | Note |
 |---------|----------|------|----|-------|------|
-| Accenture | Custom | n/a | n/a | 16.0/F — | ⏰ timeout |
-| Marriott | Travel & Hospitality / Hotels /  | n/a | n/a | 22.0/F — | ⏰ timeout |
-| Workday | Technology / Software as a Servi | n/a | n/a | 20.0/F — | ⏰ timeout |
-| FedEx | Transportation & Logistics / Shi | n/a | n/a | 22.0/F — | ⏰ timeout |
-| EY | Custom | n/a | n/a | 20.0/F — | ⏰ timeout |
-| Wells Fargo | Financial Services / Banking / M | n/a | n/a | 20.0/F — | ⏰ timeout |
-| Werner | Transportation & Logistics / Tru | n/a | n/a | 22.0/F — | ⏰ timeout |
-| Nike | Retail & Consumer Goods / Fashio | n/a | n/a | 20.0/F — | ⏰ timeout |
+| Deloitte | Custom | 78 | 78 | 83.5/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/ec141083-908e-42db-902c-3e54a3126062) | 🏆 great |
+| Nike | Retail & Consumer Goods / Fashio | 78 | 74 | 82.5/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/a8d011b3-c221-4898-bd1c-e35018862598) | 🏆 great |
+| Wells Fargo | Financial Services / Banking / M | 72 | 72 | 79.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/2d211fc5-8008-4fe6-9dba-85d411656623) | 🏆 great |
+| Splunk | Technology / Software as a Servi | 78 | 72 | 82.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/0d329d55-758c-4d1d-9ac8-5e84fc4fb9c8) | 🏆 great |
+| McKinsey | Custom | 84 | 72 | 85.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/c709d276-d37e-4ff6-8743-6359f8e5f437) | 🏆 great |
+| Zendesk | Technology / Software as a Servi | 82 | 72 | 84.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/e3b2e2a4-6b87-4f2f-bd76-2be87c0814e6) | 🏆 great |
+| Wipro | Technology / IT Services / Marke | 84 | 72 | 85.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/cbab7edc-f92a-4f30-a4d6-a9eeff433ef6) | 🏆 great |
+| LG | Retail & Consumer Goods / Consum | 78 | 72 | 82.0/B [lb](https://sebe.thoughtspotstaging.cloud/#/pinboard/d6805e4f-c372-4660-a4c8-40750bf1fd1e) | 🏆 great |
 
-## Issues
+## Data Quality Weaknesses
 
-- **Accenture**: TIMEOUT — last event: deploy started
-- **Marriott**: TIMEOUT — last event: tml import HTTP error
-- **Workday**: TIMEOUT — last event: run failed
-- **FedEx**: TIMEOUT — last event: tml import HTTP error
-- **EY**: TIMEOUT — last event: run failed
-- **Wells Fargo**: TIMEOUT — last event: run failed
-- **Werner**: TIMEOUT — last event: tml import HTTP error
-- **Nike**: TIMEOUT — last event: run failed
+**Deloitte** (data=78/100):
+  - Only 12 engagement dimension records driving 3,333 fact rows creates excessive repetition per engagement (~278 rows per 
+  - FACT_PIPELINE_PROPOSALS sample data was not visible in the provided output, making it impossible to validate pipeline wi
+**Nike** (data=78/100):
+  - Only 23 products is thin for a Nike demo — limits cross-product comparison depth and makes the catalog feel narrow
+  - FACT_INVENTORY_SELLTHROUGH joins only to CHANNEL in the model despite having REGION_KEY in the table, blocking regional 
+**Wells Fargo** (data=72/100):
+  - Row 2 has a critical internal inconsistency: 3,921 accounts activated with only $5,575 total first-year revenue ($1.42/a
+  - Most recent sample date visible is March 2026, leaving a ~4-month gap to today (July 2026); unclear if current-quarter d
+**Splunk** (data=78/100):
+  - Only 12 customer accounts is a very thin customer dimension — limits cross-account segmentation and makes patterns feel 
+  - The extremely high DOWNTIME_MINUTES of 357.40 in one row (vs. typical 16–86) may read as a data error rather than a narr
+**McKinsey** (data=84/100):
+  - STANDARD_BILLING_RATE_USD appears to be a fixed per-partner attribute repeated in the fact table rather than varying by 
+  - All three fact tables have exactly 3,333 rows — the identical count across tables feels mechanical and slightly artifici
+**Zendesk** (data=82/100):
+  - PIPELINE_COVERAGE_RATIO is identical at exactly 3.23 across all visible rows — eliminates any meaningful trend or narrat
+  - FACT_SALES table does not appear to join to PRODUCT or REGION dimensions based on the schema shown, limiting product and
+**Wipro** (data=84/100):
+  - Only 10 campaigns for a global IT services firm feels thin and limits drill-down variety
+  - No account-level or named-account dimension despite ABM being a listed campaign type
+**LG** (data=78/100):
+  - Zero-value rows (LISTED_PRICE=0, CHANNEL_SELLING_PRICE=0, ASP_VS_COMPETITOR_INDEX=0) in FACT_CHANNEL_ASP_TRACKER appear 
+  - COMPETITOR_INDEX_PRICE of $0.00 for product 6 across multiple rows zeros out the ASP vs. competitor metric for that SKU,
 
 ---
-*JSON: 2026-06-23_03-10-23_quality_run.json*
+*JSON: 2026-07-28_02-18-45_quality_run.json*

thoughtspot_deployer.py

@@ -2484,7 +2484,14 @@ class ThoughtSpotDeployer:
                     "metadata": [
                         {"identifier": guid, "type": object_type}
                         for guid in object_guids
-                    ]
+                    ],
+                    # The REST endpoint proxies to a GraphQL mutation that declares
+                    # $message as a non-null String! — omitting it makes the backend
+                    # reject the request ("Variable \"$message\" ... was not provided").
+                    # An empty string satisfies the contract; notify is off so nothing
+                    # is emailed to the recipient.
+                    "notify_on_share": False,
+                    "message": ""
                 }
             )
             if response.status_code in [200, 204]: