How to do AI with SQL: Snowflake's ML functions and AISQL

💭 Why Data Cleaning Takes 60-80% of a Data Scientist’s Time Most people think data science is all about building fancy machine learning models. But the truth is, most of the work happens before the model even begins, in cleaning the data. You spend hours fixing missing values, standardizing text, aligning dates, and removing duplicates. None of it looks glamorous, yet it’s the difference between a model that performs well and one that fails quietly. Data cleaning is where you actually start to understand the story your data is telling. It’s not the boring part of data science, it’s the foundation. #DataScience #DataCleaning #MachineLearning #DataAnalysis #Analytics #DataPreparation #AI

In this article, Ian Pay our Head of Data Analytics and Tech, explores why data skills matter in an AI world and how our new Power BI Certificate can help you unlock insight and value from data: https://ow.ly/igYG50X2ILn #icaewAcademy #powerbi #data #ai

🚀 What is AI Query in Databricks? Let’s break it down simply! 🧠✨ Imagine you have a magical assistant who can instantly read heaps of information, understand it, and then give you quick answers or summaries—without you having to dig through all that data yourself. That’s exactly what AI Query in Databricks does! The ai_query function provides a simple way to apply AI directly on your data within Databricks. It supports querying powerful AI models from different sources: the Databricks foundation model endpoint, external model endpoints, and even your own custom model endpoints using Databricks Model Serving. How to use AI Query? Here’s the basic syntax: #sql ai_query(endpoint, request) endpoint: The name of the AI model endpoint you want to query. request: The question or command you want to ask the AI about your data. For example, to summarize customer reviews, you might write: #sql SELECT ai_query('databricks-meta-llama-3-3-70b-instruct', 'Summarize the key points of these reviews') AS summary FROM customer_reviews; With AI Query, you can: Summarize content Extract insights Detect fraud Forecast trends ... all with a simple query. And you don’t need to be a tech expert! Whether it’s summarizing feedback, translating text, or predicting sales, AI Query lets you unlock AI insights directly where your data lives—easily and efficiently. Imagine telling your data, "Give me a quick summary of these reviews," and getting an instant, clear answer – right inside Databricks. No jargon, no complexity, just actionable insights. This is a game changer for businesses wanting to benefit from AI without the tech headache. Ready to simplify your data with AI? 🔥 #AI #Databricks #DataScience #BusinessInsights #EasyAI #DataMagic

🚀 Gaining Insights from Unstructured Data with Snowflake Cortex AI Recently, I explored how Snowflake Cortex AI unlocks the power of unstructured data through an end-to-end analysis using the Tasty Bytes dataset — a global food truck network operating in 15 countries. Through Snowflake Notebooks, I worked on: 1.Translating multilingual customer reviews using CORTEX.TRANSLATE 2.Summarizing feedback efficiently with CORTEX.AI_SUMMARIZE_AGG 3.Classifying customer sentiment and recommendations via CORTEX.AI_CLASSIFY and CORTEX.AI_SENTIMENT 4.Generating actionable insights with CORTEX.AI_COMPLETE 5.Extracting meaning from images and transcribing audio using CORTEX.AI_TRANSCRIBE This hands-on experience showcased how Cortex AI simplifies complex AI/ML tasks—bringing translation, summarization, classification, and sentiment analysis directly into SQL workflows inside Snowflake. 💡 It’s exciting to see how Snowflake is transforming how we analyze unstructured data across languages, media types, and sources—all within one unified data platform! #Snowflake #CortexAI #AI #DataEngineering #UnstructuredData #MachineLearning #DataAnalytics #SnowflakeNotebooks

The data analytics market is projected to reach over $400 billion by 2032! But with this growth comes a critical question for enterprises: how do you manage the immense scale and complexity of your data? Enterprises can leverage Databricks for complex data science and then operationalize the results with Yellowbrick for fast, concurrent reporting. Read this blog for a deeper dive into this game-changing strategy: https://bit.ly/3Vcv5Tb #DataAnalytics #BigData #Databricks #Yellowbrick #AI #MachineLearning #EnterpriseTechnology

✨ From SQL Queries to Text-to-SQL — The Future of Analytics? ✨ When I first started with SQL, I remember spending hours perfecting queries to pull the right data. Today, AI is making that process faster — with Text-to-SQL, you can literally type: 👉 “Show me the top 5 products by sales in 2024” and get the SQL query written for you. This doesn’t mean SQL is going away. In fact, it highlights why knowing SQL is more important than ever: 🔹 To validate what the AI generates 🔹 To optimize queries for performance 🔹 To ensure accuracy when the stakes are high I see Text-to-SQL as an assistant, not a replacement — freeing analysts to focus on insights over syntax. 💬 Have you tried Text-to-SQL yet? Do you see it as a game-changer or just a trend? #SQL #DataAnalytics #TextToSQL #AI #FutureOfWork #DataScience

Relational Data meets Graph Foundation Models Most enterprise data lives in relational databases — think hundreds of interconnected tables powering everything from recommendations to fraud detection. Traditional ML methods often treat these tables in isolation, missing the context and relationships that actually drive value. Google Research recently introduced Graph Foundation Models (GFMs): Relational tables are transformed into graphs, where - each row = a node - foreign keys = edges - columns = features This lets a single model capture both structure and features — and even generalize to unseen graphs (different schemas, node types, features). The results? On internal tasks like spam detection in ads, GFMs brought 3x–40x improvements over tuned tabular baselines. Why? Because context matters — relationships between data points often carry more predictive power than isolated rows. Why it matters: - This bridges the gap between tabular ML and graph learning. - It opens up foundation-model-level generalization for enterprise data. - It could redefine how predictive services (from ads to traffic forecasts) are built. We’re moving toward a world where "generalist" models don’t just excel at language or vision — they’ll master relational data too. If your organization relies on complex relational databases, how much value do you think is currently “locked away” in the connections rather than the individual tables? https://lnkd.in/eN7QjZdb #AI #MachineLearning #GraphAI #FoundationModels #EnterpriseData #GoogleResearch

🚀 Master the Top 5 Clustering Techniques in Data Science Clustering is a powerful unsupervised learning technique that helps uncover hidden patterns in your data. Whether you’re a beginner or an experienced data scientist, understanding these algorithms can give you a competitive edge: 🔹 K-Means Clustering – Partition data into K clusters by minimizing variance within each cluster. 🔹 Hierarchical Clustering – Build a hierarchy of clusters by iteratively merging or splitting groups. 🔹 DBSCAN – Detect clusters based on density and automatically identify outliers. 🔹 Mean Shift – Locate and adapt to the centroids of data points for flexible clustering. 🔹 Spectral Clustering – Use eigenvalues of a similarity matrix to reduce dimensions and cluster complex shapes. 💡 Why this matters: Choosing the right clustering algorithm improves insights, reveals hidden structures, and powers better decision-making in fields like customer segmentation, anomaly detection, and pattern recognition. ✅ Save this post for quick reference ✅ Share it with your network to help others level up in Data Science #DataScience #MachineLearning #Clustering #AI #BigData #Analytics

🧠 What if creating a data dashboard took seconds, not hours? A project where dashboards build themselves! Just feed your dataset, and the AI handles everything — from data visualization to generating key insights. 💡 Behind the idea: As data scientists, we spend a lot of time designing dashboards manually. This project automates that flow — letting you focus on analysis, not formatting. 🎥 Here’s a short demo of it in action 👇 Follow Dharshini Karthikeyan for more AIML tips and insights! Grateful to share this with my network and byte partners Kiruthiga Ravi Kruthica T Jenefer Rexee George Anitha D Tejaswini Muralikrishnan #carrerbytecode CareerByteCode Sangeetha B #AI #ArtificialIntelligence #ML #MachineLearning #DATA #DataScience #selfbuilding #askdharshiniai

Just wrapped up an insightful workshop on **SQL AI tools** hosted by *B10x* 🚀 The session was a deep dive into how AI is transforming the way we interact with databases — from simplifying query generation to uncovering insights faster than ever before. What really stood out to me was how these tools bridge the gap between technical expertise and business decision-making. Key takeaways: 🔹 AI can make SQL more accessible to non-technical users 🔹 Faster data analysis = quicker decisions 🔹 The future of data is collaborative: humans + AI A big thank you to the B10x team for curating such a practical and forward-looking session! 🙌 Curious to hear from others — have you tried using AI with SQL yet? How has it changed your workflow? #AI #SQL #DataAnalytics #B10x #FutureOfWork