One of the main challenges our customers outline when talking about data validations and quality is how to define and create the validations. While creating rules and validations is a continuous process between the engineers and business users, we are always looking into solutions to help our users improve this process.

With the launch of SelectZero 2024.8 release, we introduced an AI-based assistant to help users create data validations. This is done within the platform’s interface to quickly generate test cases from the users’ business requirements, or suggest new validations based on metadata and other information available in the platform. The validations are generated from rules defined by the user in natural language, or when seeking inspiration, it is possible to let AI suggest validation rules in a single click.

Delivering instant value for the users

No need for SQL knowledge

Creating custom data validations generally requires proficiency in SQL and collaboration with business users to define expectations about the data. With our AI assistant the users can simply describe their validation rule in natural language, and the assistant automatically generates the necessary SQL, ready to be used as an executable data validation. As Large Language Models (LLMs) are becoming more advanced, it is also possible to define validation rules in multiple languages, enabling a broader range of users to ensure reliable data in their organization.

Create complex data validations in a matter of seconds

Defining data validations often ranges from trivial validations, such as checking for duplicates in a table, to highly complex business specific rules that require deep domain knowledge and complicated SQL queries. In most cases, creating these advanced validations is a time consuming process, requiring significant collaboration between business users and data engineers. With SelectZero’s AI assistant, this process can be significantly simplified while also saving time for the users.

Get suggestions for data validations

Creating data validations often requires domain knowledge, understanding the data model, and information about previous anomalies. Our AI assistant leverages metadata to help users generate tailored suggestions for data quality checks. This enables data engineers to define rules independently of analysts and stakeholders while enabling business users to explore new validations to further ensure data quality.

How does it work?

Recognizing the variety of AI solutions used by our users, we designed the AI assistant to be model-agnostic. Users can configure the assistant to work with any generative LLM their organization users, such as OpenAI, Copilot, Claude, or any other.

To learn more or see the assistant in action, feel free to contact us or book a demo with our team today.