wtf is data-generation-sdk?

ai-data-platform, also referred to in its README as ADP, is a local-first tool for generating realistic, fake data for testing, demos, and analytics work, built to avoid the wait for real production data access and the privacy risk of using real records. Starting from a schema definition or a small sample of real data, it produces synthetic rows that respect foreign key relationships, so tables stay consistent with each other rather than containing orphaned or mismatched references. The tool includes automated data quality scoring, a semantic layer that detects facts, dimensions, and measures automatically instead of requiring them to be hand written, and support for running SQL analytics against the generated data. According to its README, a typical use case like generating fifty thousand rows takes under a minute, compared to production data access processes that can take weeks of waiting for compliance approval. The README states that the same tool works across different domains including retail and e-commerce, healthcare, finance and banking, SaaS and CRM systems, and manufacturing, without needing domain specific code, since the underlying data in each case is made up of the same basic building blocks: tables, keys, categories, amounts, and dates. Example projects are included for domains like retail and healthcare, showing multi-table datasets with realistic relationships. The general workflow described in the README involves three steps: initializing a project, applying a schema specification, and generating the data, after which the resulting tables can be loaded into a database or used directly. Because everything runs locally, no real data or generated data needs to leave the user's own machine, which the README highlights as a way to avoid exposing personal information while still building and testing with realistic-looking datasets.