MOSTLY AI AI features, pricing and GDPR (October 2026)

MOSTLY AI generates synthetic versions of sensitive tabular data, so banks, insurers and other companies can share, test and analyze data without exposing real personal records.

MOSTLY AI was founded in Vienna in 2017, and its about page still names Vienna. Since June 2026 the brand belongs to the Dutch vendor Syntho and appears as MOSTLY AI powered by Syntho. The offer includes the MOSTLY AI Data Intelligence Platform for data access, synthetic data generation and analysis, an open-source Synthetic Data SDK under Apache 2.0 with its TabularARGN model architecture, and an AI assistant that writes Python code from natural language. With the SDK, data can be synthesized locally without leaving the company's environment. The website names Swiss Post and Erste Group as users and states ISO/IEC 27001.

• Creates privacy-safe copies of tablesGenerative models learn the statistical patterns of a table and produce new records that keep correlations but contain no real person. Data teams can pass data to testers or partners with less privacy risk.
• Adds differential privacy to trainingDifferential privacy can be switched on during generator training to limit what the model memorizes about individuals. Data protection officers get a mathematical guarantee in addition to empirical checks.
• Analyzes data from plain languageThe AI assistant writes and runs Python code for questions typed in natural language. Business analysts explore synthetic data without writing code themselves.

• Apache 2.0 open-source SDK that runs fully inside your own environment, next to the commercial platform.
• Focus on tabular and relational data, with named users in banking and postal services.

MOSTLY AI was founded in Vienna; since June 2026 the brand belongs to the Dutch vendor Syntho and operates as MOSTLY AI powered by Syntho.

Since Syntho acquired the MOSTLY AI brand in June 2026, ask which company is your contracting partner and how support and the platform roadmap continue. Decide whether the open-source SDK in your own environment is enough or whether you need the platform, whose prices are only available on request. Have your data protection officer assess whether the synthetic output counts as anonymous for your use case. When real employee data is used as input, check co-determination and data processing terms.

No reviews yet. Be the first to share how your team uses it.

Every review shows its verification level. We read each one before it goes live. How we check reviews