The MWS Data Scout service has become part of the MWS Data platform and is an AI agent based on a large language model (LLM). It scans all of the company's databases and generates a brief description of what they contain and how they are related to each other.
MTS emphasized that this is the first such service in Russia. The AI agent can connect to the company's IT systems both from the cloud and from the client's secure environment. The solution integrates with popular data catalogs such as DataHub, OpenMetadata, as well as the data catalog from MWS. The service can analyze how tables are related to each other, what data is stored in them, and determine their other characteristics.
The service can also identify which databases store critical information, such as passport data (number, series, date of issue), personal data (full name, place of residence, telephone), and banking data (pin, cvv, cardholder name).
At the first stage, the AI agent receives meta-data (general information about the names of tables and columns in them), and also connects to the company's Confluence, where additional information about the database may also be stored. This allows the service to get a more complete picture of the structure and purpose of the data, improve the accuracy of descriptions, and take into account the business context recorded in the documentation. Next, the AI agent describes the tables and columns themselves and finds critical data. After the analysis is completed, the AI agent provides a structured report describing the detected tables, the relationships between them, and uploads the results to the data catalog.
In the future, the service will be able to build data pipelines from finding the right source (for example, with master data) to enriching the data and delivering it to BI systems or ML models with mandatory data quality checks. It will also be able to detect anomalies, helping to recognize sharp deviations in the data that may signal problems or suspicious events.
