How MySQL AI Brings Built‑In Machine Learning and GenAI to Your Database

Introduction

MySQL is the world’s most widely used open‑source database, trusted for decades by startups, SMBs, and large enterprises for its reliability and ease of use. To celebrate its 30th anniversary, Oracle announced MySQL AI, an optional feature of MySQL Enterprise that embeds predictive and generative AI capabilities, including auto‑ML, built‑in LLMs, and vector storage, all accessible through familiar MySQL tools, protocols, and the MySQL Studio development environment.

Generative Artificial Intelligence

Generative AI lets users extract accurate, context‑aware information from proprietary enterprise documents stored on local file systems without requiring AI expertise. It supports translation, summarization, and chatbot development, enabling concise, multilingual summaries while preserving privacy. The built‑in LLM uses patented quantization and caching techniques to accelerate CPU execution. MySQL AI also provides NL2SQL (text‑to‑SQL) so natural‑language queries can be converted into SQL statements.

Vector Engine

MySQL AI can create vectors from local documents and store them in InnoDB’s vector store, using an embedded embedding model optimized for CPUs. Automatic embedding generation and the built‑in LLM allow semantic search of proprietary documents via native SQL operators, delivering memory‑bandwidth‑level speed that scales with core count, eliminating the need for a separate vector database.

Predictive AI with AutoML

The AutoML feature in MySQL AI helps train, predict, and explain classification, regression, forecasting, recommendation, anomaly detection, and topic‑modeling models. It automates algorithm selection, data sampling, feature selection, and hyper‑parameter tuning, allowing users to apply machine learning directly on MySQL data.

Use cases span fraud detection in finance, sensor‑data‑driven predictive maintenance in manufacturing, demand forecasting in energy, and personalized product recommendations in retail. Combining AutoML with GenAI enables novel scenarios, such as generating anomaly reports from production logs and using a chat interface to diagnose root causes without manual log analysis.

JavaScript Stored Procedure Support for GenAI

MySQL natively supports JavaScript stored procedures, allowing developers to write JavaScript code that interacts directly with MySQL data via the GenAI API. This enables prompt preprocessing, post‑processing of LLM responses (e.g., sentiment analysis, summarization), and conversion of LLM output into actionable operations using JavaScript’s string and regex capabilities.

MySQL Studio

MySQL Studio is the new visual interface for MySQL AI, offering an integrated environment with SQL worksheets, a chat interface for querying vector‑stored documents, and an interactive notebook for developing ML and GenAI applications. The notebook is Jupyter‑compatible, allowing import, sharing, and collaborative development of AI projects.

Conclusion

MySQL AI empowers developers to build rich applications using built‑in machine learning, GenAI, LLM, and semantic search capabilities directly within MySQL. Vectors can be created from local documents, and AI workloads can run on‑premise or be migrated to MySQL HeatWave for lower cost, higher performance, and enhanced features without changing the application.