Skip to content

ConfigurationGuide.md

Latest commit

 

History

History
128 lines (79 loc) · 4.47 KB

ConfigurationGuide.md

File metadata and controls

128 lines (79 loc) · 4.47 KB
Raw

Configuration Guide — Scenario Packs

Overview

Scenario packs are pre-built configurations that adapt the solution to a specific industry and use case. Each pack includes synthetic sample data, an ontology configuration, and curated sample questions so the solution is ready to evaluate immediately after deployment.

This guide walks you through selecting and applying a scenario pack, understanding what each pack contains, switching to a different scenario on an existing deployment, and bringing your own data.

Available Scenario Packs

Pack Industry Use Case Tables Documents
retail Retail Sales analysis & product performance 13 (account, customer, customeraccount, customerrelationshiptype, customertradename, invoice, location, orderline, orderpayment, orders, payment, product, productcategory) 7 PDFs
insurance Insurance Client meeting preparation 4 (customer, policy, claim, communicationshistory) 7 PDFs
default Retail Sales analysis & product performance Same as retail (used when no --scenario flag is provided) Same as retail

Note: The sample data associated with each scenario in this repository is synthetic and generated. The data is intended for use as sample data only.

Pre-requisites

Before configuring a scenario pack, complete the full solution setup described in the Deployment Guide.

Step 1: Choose a Scenario Pack

Use the --scenario flag when running the build script to select a pre-built scenario pack.

Retail Scenario

python infra/scripts/post-provision/00_build_solution.py --scenario retail

Insurance Scenario

python infra/scripts/post-provision/00_build_solution.py --scenario insurance

Default (no flag)

If you omit --scenario, the default scenario is used automatically (equivalent to retail):

python infra/scripts/post-provision/00_build_solution.py

Press Enter to start or Ctrl+C to cancel the process.

Replacing an existing scenario? If you already have a scenario loaded and want to switch to a different one, add --clean to clear and recreate existing artifacts (tables, ontology, search index) before loading the new scenario data:

python infra/scripts/post-provision/00_build_solution.py --scenario insurance --clean

Step 2: Validate the Scenario

After the build completes, test the agent to confirm the scenario data is loaded and queries are working:

python infra/scripts/post-provision/06_test_agent.py

Use the sample questions below to verify each scenario.

Scenario Pack Details

Retail

Industry: Retail & Consumer Goods
Use Case: Sales analysis and product performance

Sample Questions: See Retail sample questions.

Insurance

Industry: Financial Services / Insurance
Use Case: Claims processing and customer management

Sample Questions:

See Insurance sample questions.

Switching to a Different Scenario on an Existing Deployment

To switch from one scenario pack to another on an already-provisioned environment, re-run the build script with the new --scenario flag and --clean. This replaces the existing scenario data and ontology configuration:

# Switch to insurance
python infra/scripts/post-provision/00_build_solution.py --scenario insurance --clean

# Switch back to retail
python infra/scripts/post-provision/00_build_solution.py --scenario retail --clean

Note: Re-running the build from step 01 will overwrite the existing Fabric tables and agent configuration with the new scenario's data.

To re-run from a specific step without reloading data, use the --from flag:

# Re-run from step 03 onward (skips data upload)
python infra/scripts/post-provision/00_build_solution.py --scenario retail --from 03

Related Documentation

Document Description
Deployment Guide Step-by-step instructions for provisioning and deploying the solution
Technical Architecture System design and component overview
Troubleshooting Common issues and resolution steps
Customization Customize the solution with your date