DataCebo makes the case that enterprises need to build and own generative models of their proprietary relational data.
BOSTON, Sept. 15, 2026 /PRNewswire/ — DataCebo today released SDV 2.0, software that lets an organization build a generative relational model of its own databases.
DataCebo calls this new class of models generative relational models. Trained from scratch, the model learns the database as a connected whole, capturing the statistical patterns, data structure, relationships, context, and business rules embedded in the data and spanning its tables. SDV 2.0 enables organizations to build these models inside their secure environment and use them to generate synthetic data, train and evaluate AI agents, and more, without moving or exposing production data.
LLMs bring the world’s intelligence to the enterprise. Generative relational models bring the enterprise’s intelligence to AI.
Generative AI has dramatically accelerated how software is written and has changed how enterprise teams work with language, images, and code. But the operational intelligence unique to a company—its customers, transactions, relationships and business rules—lives inside that company’s own complex databases and data warehouses. This is the ‘data moat’ that companies must guard and sustain.
Enterprises Can Now Build Generative Relational Models of Their Own Data
Until now, enterprises reconstructed that operational intelligence one use case at a time. Software teams copied and masked production data slices for testing and development. Data engineers staged pre-production copies for distributed teams. Machine-learning engineers repeatedly engineered features and trained separate models on the same underlying records. Each workflow produced a narrow representation, often missing the relationships, rare events and behavioral patterns that are only present across the full database.
SDV 2.0, the next release of SDV Enterprise, allows companies to build a generative relational model using a representative subset of their data, typically in minutes to an hour. It can model schemas of any relational depth, with training taking place inside infrastructure the organization controls.
The resulting model can generate data for specific scenarios and edge cases, simulate behaviors, and provide enterprise context to analytics tools, models and agents, reducing reliance on production data while supporting privacy and regulatory requirements.
“Companies have spent decades building intelligence in their databases, but every team has had to reconstruct a small piece for each new use,” said Kalyan Veeramachaneni, CEO of DataCebo. “A generative relational model lets them capture that intelligence once and put it to work across the business.”
From Configuration to Automation
When DataCebo first launched the commercial edition of SDV in 2024, it proved that generative models could learn from complex relational databases and produce realistic, connected synthetic data.
Early deployments revealed a significant challenge: enterprise databases can contain hundreds of tables and thousands of columns, often with incomplete documentation of their keys, formats, relationships and business rules. To build a high-quality model, data and machine-learning teams had to supply much of that configuration manually.
SDV 2.0 automates much of this discovery and setup. When organizations connect the software to a database or collection of files, SDV 2.0 can:
- Detect schemas and relationships, including primary, foreign and composite keys
- Identify formats and context embedded in data values
- Detect and enforce business constraints
- Tune the generative model
- Generate task-specific datasets, including rare events and edge cases
SDV 2.0 connects directly to Oracle, SQL Server, BigQuery, Spanner and AlloyDB and runs within the customer’s environment.
Enterprises Are Replacing Production-Dependent Workflows
Organizations are already using DataCebo to generate realistic data with models built inside their own environments. With DataCebo, ING Belgium generated 10,000 synthetic payments in two minutes, achieving 100 times the test coverage in less than one-tenth of the time previously required. Epiconcept created a synthetic database in 55 minutes and used it to identify optimizations that improved query performance by 105 times.
Built on More Than 15 Years of Research
DataCebo’s technology originated at MIT’s Data to AI Lab. Co-founder and CEO Kalyan Veeramachaneni, a Principal Research Scientist at MIT, has spent more than 15 years pioneering how AI learns from relational and tabular data. Veeramachaneni and co-founder and CPO Neha Patki led the team that created the Synthetic Data Vault, the source-available platform for generating tabular synthetic data, as well as SDMetrics, a framework for evaluating synthetic data quality.
Today, SDV has more than 18 million downloads and has been cited in over 5,000 research papers and used by more than 30,000 data scientists. SDV is used across financial services, healthcare, life sciences, consumer goods and other highly governed industries.
Availability
SDV 2.0 is available now. Self-service, consumption-based pricing begins at $500 per month for unlimited tables. Organizations can get started at https://portal.datacebo.com/signup.
About DataCebo
DataCebo is the company behind the Synthetic Data Vault (SDV), the source-available ecosystem that pioneered generative modeling for tabular and enterprise relational data. Originating at MIT’s Data to AI Lab, DataCebo is backed by Link Ventures, Uncorrelated Ventures and Zetta Venture Partners.
SOURCE Datacebo