BigQuery has launched AI functions — AI.IF, AI.CLASSIFY, and AI.SCORE — designed to enhance SQL capabilities with generative AI directly embedded into queries. These functions simplify the process of integrating large language models (LLMs) by eliminating the need for complex prompt tuning or separate tools. Users can leverage familiar SQL syntax to filter, categorize, and rank data, making sophisticated AI analysis more accessible to those accustomed to traditional SQL workflows. AI.IF enables semantic filtering and joining of data using natural language, allowing users to find relevant documents or identify sentiments in customer reviews. For instance, it can sort tech news articles based on their relevance to a specific company, even if the article uses an outdated name. AI.CLASSIFY focuses on categorizing unstructured data like text and images, which can streamline tasks such as routing support tickets or analyzing news article topics. AI.SCORE ranks results based on user-defined criteria, automatically refining prompts for better output quality. Behind the scenes, BigQuery optimizes the use of these functions to minimize costs and improve efficiency. It reduces the number of LLM calls by evaluating non-AI filters first and reorders queries to enhance performance. Users are encouraged to start with the managed AI functions for cost-effective, high-quality results, while the AI.GENERATE family is available for those wanting more control over prompts and model selection. Future improvements aim to further boost performance, with claims of potential speed increases up to 100 times.