If a photo is worth a thousand words, a video demo must be worth a million. In this case though I don't understand the underlying technology (that's ok I'm not the target audience) this video demo drive home the benefits of this new tool in a way that I wouldn't have understood through text alone. Seeing is believing. Thanks Jackson Cavenaugh for bringing it to life.

Snowflake’s Cortex Code CLI (generally available since February 2026) is an AI‑powered command‑line assistant that lets data engineers, analysts, and developers work with Snowflake—and tools like dbt and Apache Airflow—using plain English, straight from the terminal. I just came across this tool, and it’s already changing how our data team works—had to share! Here are 5 use cases that blew me away: 1️⃣ Revenue queries in plain English — no more manual SQL. 2️⃣ Auto-generate dbt CLV models from a single sentence. 3️⃣ Build production Airflow DAGs in seconds. 4️⃣ Root-cause pipeline failures in 30 seconds flat. 5️⃣ Discover PII data with role-aware governance scanning. Watch the short demo video below to see it in action ↑ #Snowflake #DataEngineering #AITools #dbt #Airflow #CortexAI #DataCloud #MLOps #DataTeams #snowflake Shubhangi Singh Snowflake VCreaTek Consulting

80% of new databases on Databricks are now being created by AI agents — not engineers. That number is striking enough to stop and think about what's actually changing. Data engineers spent years building expertise around schema design, partition strategies, and storage optimization. That judgment is now being delegated to agents operating at scale. Databricks ships Genie Code, Snowflake has Cortex Code, dbt just released agent skills so tools like Claude and Cursor can follow analytics engineering best practices end-to-end. But here's the nuance worth sitting with: this is a role shift, not a disappearing act. The engineers who navigate this well won't be the ones who write the best CREATE TABLE statement — they'll be the ones who define the right guardrails, audit what agents produce, and know when to override. The mental model is moving from "I build data infrastructure" to "I govern systems that build data infrastructure." Most data teams haven't started redesigning their workflows around this reality. The question isn't whether agents will take over more of the build — it's whether you're shaping how that happens, or just watching it happen to you. How is your team thinking about the human-agent split in your data stack? #DataEngineering #AIAgents #DataPlatform

We recently explored a new feature in Databricks that could fundamentally change how data pipelines are built - not in theory, but in a real client use case. The Lakeflow Designer introduces a fully visual, no-code approach to building data pipelines; with live data previews and tight integration into the Lakehouse. But how far is it really from replacing legacy ETL tools? We put it to the test in a real-world scenario and documented what works, what’s missing, and where it’s heading. Curious to see if visual data engineering is ready for enterprise scale? Read the full blog post, in English or German language, through the links in the first comment. #Databricks #LakeflowDesigner #DataPipelines #DataPlatform #AI #MachineLearning

We recently explored a new feature in Databricks that could fundamentally change how data pipelines are built - not in theory, but in a real client use case. The Lakeflow Designer introduces a fully visual, no-code approach to building data pipelines; with live data previews and tight integration into the Lakehouse. But how far is it really from replacing legacy ETL tools? We put it to the test in a real-world scenario and documented what works, what’s missing, and where it’s heading. Curious to see if visual data engineering is ready for enterprise scale? Read the full blog post, in English or German language, through the links in the first comment. #Databricks #LakeflowDesigner #DataPipelines #DataPlatform #AI #MachineLearning

Hevo Data invited me to do a live walkthrough of the Snowflake cortex code. 115,000 people follow them. That was the most nervous I've been in a long time. Here's what we covered — and why it matters for every data engineer right now 👇 The topic: Cortex Code inside Snowflake. Most people are sleeping on this. You can now write in plain English inside Snowflake. And get production-ready SQL and workflows built for you. In minutes. Without leaving Snowflake. No switching tools. No copy-pasting between ChatGPT and your IDE. No "let me clean this up before it goes to prod." Just describe what you want. Cortex Code builds it. But here's the thing I kept saying during the session — Cortex Code is only as powerful as the data beneath it. An AI that sits on top of bad pipelines, unmodelled data, and zero governance — is just a faster way to get wrong answers. The stack has to be right first. What the full stack looks like when it's right: → Clean ingestion (Snowpipe, dynamic tables) → Modelled data (dbt or native Snowflake transformations) → Governed access (masking policies, row-level security) → THEN Cortex Code on top In that order. Not the other way around. That's what we walked through live. The moment that surprised the audience most: I typed a plain English question about revenue trends. Cortex Code wrote the SQL, ran it, and explained the output. No query writing. No debugging. No Stack Overflow. The data engineer's job is not disappearing. It's moving up the stack. Less time writing SQL. More time making sure the foundation Cortex sits on is bulletproof. That's the shift. That's what I'm spending 2026 focused on. Watch the full session — link in the comments. If this hit different — repost it to someone who's still debating whether AI inside Snowflake is real. Elsa Mayer Amilee Alesna Ivanna Lorena Hernandez Santos It's real. I did it live. 🎯 #Snowflake #CortexCode #DataEngineering #SnowflakeSquad #HevoData #DataPlatform #AIData

AI-powered data engineering isn’t the future — it’s already here. And stepping back isn’t really an option. What used to take hours, days… even months — exploring systems, fixing pipelines, validating outputs — is now collapsing into minutes. With Altimate AI , the shift is real: From writing code → to executing outcomes From copilots → to operators It connects directly to your stack, runs real tasks, verifies outputs, and keeps going until the job is done. No tab switching. No copy-paste loops. Just intent → execution. Open-source agentic data engineering harness 100+ deterministic tools for: SQL • column-level lineage • dbt • FinOps • warehouse connectivity Works across your stack: Snowflake • Databricks • @BigQuery • PostgreSQL • AiRFLOW • GitHub • DuckDB • Microsoft Fabric more Bring your own LLM — no vendor lock-in. 👉 Get started in one command: npm install -g altimate-code #1 on ADE-Bench (74.4%) — real workflows, not demos. This isn’t a SQL copilot. It’s a data operator. ⭐ on Github If you find it useful: https://lnkd.in/e4PHNc-7 Captions are auto-generated and may contain mistakes. #OpenSource #AgenticAI #DataEngineering #AltimateCode #AI #dbt #Snowflake #Databricks #BigQuery #DataPipelines #AnalyticsEngineering #MLOps #DataStack

There’s a quiet shift happening in data engineering — from writing pipelines… to letting systems understand and execute them. In this video, I take a closer look at ¨Altimate AI — an agentic data engineering harness that recently ranked #1 on ADE-Bench, outperforming Cortex Code CLI, dbt Labs Copilot, and Claude Code across 43 real-world tasks. What makes it interesting isn’t just the benchmarks — it’s the approach. Altimate Code doesn’t stop at generating SQL. It connects to your actual data stack and works across it. Inside VS Code and Cursor (via the dbt Power User extension), it integrates directly with: Snowflake, BigQuery, Databricks, PostgreSQL, Redshift, Airflow, Dagster, GitHub, and more. And once connected, it behaves less like a copilot… and more like a system-level assistant. Here’s what stood out: • Agent harness with 100+ deterministic tools Not just suggestions — real execution across SQL, lineage, dbt workflows, and warehouse operations • Column-level lineage tracking Clear visibility from source → transformation → BI layer • Real-time cost analysis (FinOps signals) Understand query impact and optimize spend as you build • 10M free tokens (no credit card required) Easy to get started and experiment • BYOK (35+ providers supported) Flexibility to plug in your preferred models • Fully local execution No data leaves your environment — strong for GDPR/EU setups The bigger takeaway: We’re moving from “AI helps you write code” to “AI understands your data system and executes workflows” That shift reduces context switching, removes manual debugging loops, and brings us closer to intent-driven data engineering. If you’re working with dbt, Snowflake, or Databricks — this is worth exploring. 🔗 Install: npm install -g altimate-code 📖 Docs: https://lnkd.in/e_QrWk4v ⭐ GitHub: https://lnkd.in/e4PHNc-7 #dataengineering #aiagent #dbt #snowflake #databricks #analytics #llm #cursor #vscode

Wrapped up Snowflake's Data for Breakfast Virtual Event . Five days, 24 sessions, and honestly a lot more than I expected to take away. A few things that genuinely stuck with me: Snowflake Cortex Code is not just a coding assistant. It is an agentic AI tool that can build pipelines, generate SQL, and move ideas to production without leaving the Snowflake environment. Watching it in action was fascinating. The Northstar labs covering data engineering fundamentals. Ingesting, transforming, and delivering data through Snowflake pipelines. The Dynamic Tables session was a highlight as it showed how you can automate pipeline refreshes without manual scheduling, a concept that clicks immediately once you see it in practice. The line that stayed with me all week: there is no AI strategy without a data strategy. Simple but important regardless of which data role you are in. #Snowflake #DataEngineering #DataAnalytics #BusinessAnalysis #AI #DataForBreakfast

Migration isn't just about moving code—it’s about making it better.. 🚀 When migrating legacy Stored Procedures to Databricks, "lift and shift" is rarely the best answer. We recently took a more strategic approach to refactor and optimize. The Challenge: We started with a complex Stored Procedure that calculated opportunity counts and sub-vertical data. It relied on 4 separate SQL queries—all using the same heavy joins but with different filter conditions. A classic "lift and shift" would have just replicated this inefficiency in a notebook. The Evolution with Genie: 1️⃣ First Pass: Genie converted the Stored Procedure into functional notebook code. It worked, but it still followed the old 4-query pattern. 2️⃣ The "Pattern Approach": We prompted Genie to look for efficiencies. Since all 4 queries hit the same tables, we asked: "Can you consolidate these?" 3️⃣ The Result: Genie refactored the logic into a single, streamlined query. It converted those separate filtered queries into smart CASE WHEN statements within a single SELECT block. The Takeaway: Migration shouldn't just be about moving code from point A to point B. By using approach to tune AI Engine(Genie) to identify patterns and consolidate logic, we moved from a "Lift and Shift" model to an Optimized Approach. Better performance, cleaner code, and lower compute costs. That’s the power of the Lakehouse. 💡 #Hexaware #Databricks #DataEngineering #AI #SQL #Geniecode #DataMigration #Lakehouse