Skip to content

Generate Custom Data

Run the AI Generator

Override .env settings directly from the command line: 

python scripts/00_build_solution.py --clean --industry "Insurance" --usecase "Property insurance with claims processing and policy management"

Flag Purpose
--clean Reset Fabric artifacts (required when switching scenarios)

Alternative: update .env configuration

With your .env configured, generate customer-specific data:

python scripts/00_build_solution.py --clean

Generation Process

============================================================
Generating Custom Data with AI
============================================================

Industry: Insurance
Use Case: Property insurance with claims processing and policy management

[1/4] Generating ontology configuration...
  → Analyzing use case for entities and relationships
  → Created: Policies, Claims, Customers, Agents
  → Defined: 6 relationships, 4 business rules
  ✓ ontology_config.json

[2/4] Generating structured data (CSV)...
  → policies.csv (16 rows) — Policy details, coverage, premiums
  → claims.csv (40 rows) — Claim status, amounts, dates
  → customers.csv (20 rows) — Policyholder information
  → agents.csv (8 rows) — Agent assignments
  ✓ 4 CSV files generated

[3/4] Generating documents (PDF)...
  → claims_process.pdf — How to file and process claims
  → coverage_guide.pdf — Policy coverage explanations
  → underwriting_policy.pdf — Risk assessment guidelines
  ✓ 3 documents generated

[4/4] Generating sample questions...
  → 15 sample questions spanning data, documents, and combined
  ✓ sample_questions.txt

============================================================
Data generated: data/20260202_143022_insurance/
============================================================

Review Generated Content

Check what was created:

# View the generated files
ls data/*/

# Read sample questions for testing
cat data/*/config/sample_questions.txt

Sample Questions (Insurance Example)

# Structured data questions (Fabric IQ)
How many open claims do we have?
What's the total value of claims filed this month?
Which agents have the most policies?

# Unstructured data questions (Foundry IQ)
What's our process for filing a property claim?
What does our standard homeowner policy cover?
What are our underwriting criteria for high-risk properties?

# Combined questions (Both)
Which open claims are approaching our SLA deadline based on our process guidelines?
Do any current claims involve coverage types not in our standard policy?

Expected Output

The complete build takes ~5 minutes:

============================================================
Building Solution: Insurance
============================================================

[01a/07] Generating AI data...
  ✓ Ontology, CSVs, PDFs, and questions generated

[02/07] Setting up Fabric workspace...
  ✓ Cleaned previous artifacts
  ✓ Lakehouse: iqworkshop_lakehouse
  ✓ Warehouse: iqworkshop_warehouse

[03/07] Loading data into Fabric...
  ✓ policies.csv → 16 rows
  ✓ claims.csv → 40 rows
  ✓ customers.csv → 20 rows
  ✓ agents.csv → 8 rows

[04/07] Generating NL2SQL prompt...
  ✓ Schema prompt created for insurance domain

[05/07] Creating Fabric Data Agent...
  ✓ Agent: fabric-agent-insurance

[06/07] Uploading documents to AI Search...
  ✓ 3 documents → 24 chunks indexed

[07/07] Creating Orchestrator Agent...
  ✓ Agent: insurance-multi-agent

============================================================
Build complete! Ready for customer PoC.
============================================================

Troubleshooting

AI generation times out

  • Check your internet connection
  • Try with --size small for faster generation
  • Verify Azure AI Services quota

Fabric errors on --clean

  • Ensure FABRIC_WORKSPACE_ID is correct
  • Check you have Contributor access to the workspace
  • Wait a minute and retry (Fabric operations can be slow)

← Overview | Test Your PoC →