L2P — Logic to Product
Moving data is easier.
Core trick
Model traces legacy report logic into a candidate data product, then eleven deterministic contract checks gate a versioned release.
Description
Moving data is easier. L2P moves the business knowledge embedded in existing reports and source models into tested, reusable Snowflake data products.
A Claude-powered Builder agent retrieves synthetic HANA, Qlik and Spotfire definitions, traces their meaning, creates a candidate product, and runs eleven deterministic contract checks before release. A separate Analyst agent consumes the released product through a guarded semantic view.
Our demo shows why totals and row counts alone are insufficient: a join duplicates invoice contributions and inflates sales. The release gate rejects it. A corrected product passes validation, receives a versioned release, and supports a verified customer ranking.
The prototype uses fictional Northstar Supply data and synthetic report definitions. No employer data or proprietary assets are included.
README excerpt
# L2P — Logic to Product: event build Start with [the event demo, video, architecture and evidence](event/l2p/START-HERE.md). This branch is `l2p-hackathon-build`. The preparation baseline below is retained for provenance and is not claimed as event-built functionality. --- # Hackathon pre-work baseline — 2026-10-23 This folder is preparation, not claimed as functionality built during the hackathon. No employer datasets, definitions, agent instructions, skills, logs or credentials are included. ## Source and license Microsoft sql-server-samples, MIT License, pinned commit `dac9747acaca4e54a3b5ea4c6de1c4be6c1b964a`. Original files are preserved under `source/`; source/manifest.json records URLs and hashes. Retain source/license.txt with any redistribution. Public repository: https://github.com/microsoft/sql-server-samples/tree/master/samples/databases/wide-world-importers ## Data origin and limitations The eight table schemas are a **simplified adaptation** of Wide World Importers. All rows are independently generated synthetic data, NOT an export of Microsoft's database. The original SQL Server code is reference material only. It is not executed in Snowflake and may refer to omitted tables and columns. Temporal tables, audit columns, operational procedures, package types and SQL Server indexes are omitted. No claim of full WWI parity. The invoice formula preserves the original meaning: ExtendedPrice includes tax; subtract TaxAmount to obtain invoice sales excluding tax.