DataCebo Launches SDV 2.0 to Automate Generative Enterprise Modeling

For years, enterprises have struggled to leverage their operational intelligence without exposing sensitive production data. Previously, teams relied on labor-intensive, piecemeal methods to reconstruct database logic for testing or machine learning, often missing the critical relationships and rare events buried across deep relational schemas. SDV 2.0 shifts this paradigm by automating the discovery of primary keys, foreign keys, and complex business constraints, turning raw database structures into cohesive generative models.

Kalyan Veeramachaneni, CEO of DataCebo, describes this as a shift from constant reconstruction to singular intelligence capture. The platform integrates with major systems including Oracle, SQL Server, BigQuery, Spanner, and AlloyDB. Real-world applications are already yielding results: ING Belgium utilized the technology to generate 10,000 synthetic payments in two minutes, while Epiconcept leveraged synthetic databases to optimize query performance by over 100 times.

Built on over 15 years of research originating at MIT’s Data to AI Lab, the release builds upon the open-source Synthetic Data Vault ecosystem, which currently counts over 18 million downloads. By running inside a company’s own environment, SDV 2.0 addresses privacy and regulatory hurdles, allowing data scientists to simulate edge cases and train models without moving production records. The software is available immediately with consumption-based pricing starting at $500 per month.