"Deploying AI Agents at Scale: 3 Patterns for Enterprises"

🎥 Missed this week’s Data Drift? Here’s the video edition of: “From Query to Action: How AI Agents Are Rewiring Enterprise Decision-Making.” Most enterprise AI still stops at insights. The next frontier is AI that can reason, plan, and act — turning data into decisions and decisions into execution. In this week’s breakdown, I walk through the architecture behind a Snowflake-native AI agent that connects analytics to real action: 🧩 How agentic AI moves beyond answering questions to executing workflows 🧠 How Cortex + Snowpark fuse structured and unstructured data into a reasoning engine 🔍 Why governance, traceability, and full audit logs are non-negotiable 🚀 What this shift means for the future of enterprise automation If you’re building AI systems — or leading teams that use them — this evolution from insight → action will define the next decade of enterprise intelligence. 🎬 Watch the full video below 📬 Read the full LinkedIn article: https://lnkd.in/gZkUyJzA 📝 Read the full Medium deep dive: https://lnkd.in/gn7zTv_b #AI #AgenticAI #EnterpriseAI #Snowflake #CortexAI #AIStrategy #Automation #DataDrift

The Databricks Data+AI World Tour stop in NYC delivered exactly what I hoped for: meaningful reconnections and substantive conversations about where AI is really heading. Spent the day in dialogue with leaders navigating their organizations' AI transformations—not just talking about the technology, but wrestling with the real challenges of implementation and ROI. One line from the sessions captured it perfectly: "Data as the foundation, AI being the multiplier." That framing reinforced something I've been thinking about a lot lately. You can't multiply by zero. Without solid data infrastructure and governance, even the most sophisticated AI becomes an expensive science project. Many of the insights I heard echoed themes from my recent post on AI ROI and practical implementation (link below). It's encouraging to see these principles validated across different organizations and industries. https://lnkd.in/eM_6yWHm #databricks #dataaiworldtour

Every day, enterprises pour massive amounts of data into Splunk… but how much of it is actually doing something for you? 𝐓𝐡𝐚𝐭’𝐬 𝐭𝐡𝐞 𝐪𝐮𝐞𝐬𝐭𝐢𝐨𝐧 𝐝𝐚𝐭𝐚𝐬𝐞𝐧𝐬𝐀𝐈 𝐟𝐨𝐫 𝐒𝐩𝐥𝐮𝐧𝐤 𝐰𝐚𝐬 𝐛𝐮𝐢𝐥𝐭 𝐭𝐨 𝐚𝐧𝐬𝐰𝐞𝐫. By using AI to score data efficiency, reveal underused sources, and connect costs to real value, teams finally get a clear picture of what’s powering insights, and what’s just adding weight. Here are 5 ways datasensAI helps teams fall back in love with their data: 1️⃣ Finds hidden value in overlooked data 2️⃣ Simplifies management with automated visibility 3️⃣ Cuts license waste through smarter data scoring 4️⃣ Speeds up troubleshooting with real usage insights 5️⃣ Maximizes ROI by aligning data with business impact 𝐁𝐞𝐜𝐚𝐮𝐬𝐞 𝐨𝐛𝐬𝐞𝐫𝐯𝐚𝐛𝐢𝐥𝐢𝐭𝐲 𝐢𝐬𝐧’𝐭 𝐣𝐮𝐬𝐭 𝐚𝐛𝐨𝐮𝐭 𝐬𝐞𝐞𝐢𝐧𝐠 𝐞𝐯𝐞𝐫𝐲𝐭𝐡𝐢𝐧𝐠  𝐢𝐭’𝐬 𝐚𝐛𝐨𝐮𝐭 𝐮𝐧𝐝𝐞𝐫𝐬𝐭𝐚𝐧𝐝𝐢𝐧𝐠 𝐰𝐡𝐚𝐭 𝐦𝐚𝐭𝐭𝐞𝐫𝐬. 🔗 𝐑𝐞𝐚𝐝 𝐭𝐡𝐞 𝐟𝐮𝐥𝐥 𝐛𝐥𝐨𝐠 𝐭𝐨 𝐝𝐢𝐯𝐞 𝐝𝐞𝐞𝐩𝐞𝐫 𝐢𝐧𝐭𝐨 𝐡𝐨𝐰 𝐝𝐚𝐭𝐚𝐬𝐞𝐧𝐬𝐀𝐈 𝐭𝐫𝐚𝐧𝐬𝐟𝐨𝐫𝐦𝐬 𝐒𝐩𝐥𝐮𝐧𝐤 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞 𝐚𝐧𝐝 𝐑𝐎𝐈. 👉 https://lnkd.in/gU6TR4s7 bitsIO Inc. #Splunk #AI #DataAnalytics #Observability #bitsIO #datasensAI #DataStrategy

The Great MLOps Pivot: Why Data is the real AI moat? For years, AI teams have focused too much on building fancy models - new architectures, better fine-tuning & endless benchmarks. But once these models go live, they often fail because real-world data keeps changing. The real game-changer isn't a new model - it's shifting from model-centric to data-centric AI. In a model-centric world, the model is everything and data is just fuel. In a data-centric world, data becomes the main asset - it's cleaned, improved, and updated to make models work better over time. Data-centric MLOps means building systems that keep learning and improving. It's powered by three key ideas: -> Feedback Loops: Learn from how users interact - what they like, ignore or correct. Detect when data or predictions start drifting off. -> Data Curation: Treat data like code. Version, track and validate it using tools like DVC or Pachyderm. -> Continuous Retraining: Automate retraining when performance drops or new data arrives - with safe rollouts. This turns static, fragile models into living systems that get smarter every day. As Andrew Ng says - improving data quality gives a far higher return than endlessly tweaking models. Start small: pick one feedback signal and start logging it. That's the first turn of your data flywheel. Stop building fancy models. Start building systems that learn. #MLOps #DataCentricAI #AI #MachineLearning #DataScience #AIML

🚀💡 Snowflake is simplifying governed AI deployment at scale. For the Snowflake Partner Network, this means massive opportunities to accelerate time-to-value and expand service offerings with cutting-edge solutions! These capabilities can be used to: - Provide every user with the ability to answer complex questions in natural language with Snowflake Intelligence - Develop data agents intuitively and flexibly on top of the enterprise context database, including all your structured and unstructured data, with enhanced reasoning and access to leading model providers out of the box - Accelerate multimodal AI pipelines using simple SQL, including embedded cost governance controls for easy TCO management - Accelerate production ML with advanced MLOps capabilities and low-latency serving of ML features for online predictions ➡️ Ready to accelerate your expertise? Learn more about these new capabilities: https://lnkd.in/gK_4FktN #EnterpriseAI #AIDataCloud #PartnerNetwork #SnowflakeIntelligence

When building large Data Vaults with Copilot, I work in bounded domains. Here's why: Large vaults have hundreds of interdependent objects. Copilot's context window can't track them all at once. Errors accumulate. The mechanism: context decay. Copilot prioritises recent instructions. Early standards get deprioritised. Hash keys drift. Naming becomes inconsistent. My approach: Work one domain at a time, Reference key standards in every prompt, Validate after each domain completes This keeps consistency across 100+ objects while maintaining AI speed. Quick note: This assumes you're already familiar with Data Vault 2.0. AI accelerates experienced architects, it doesn't replace foundational knowledge. What's your approach to managing context limits with AI? #DataArchitecture #DataVault #MicrosoftCopilot #AI #EnterpriseAI

The ability to customize AI models for specific domains is becoming increasingly accessible. Imagine a plug-and-play architecture where one can seamlessly integrate their own model. This advancement allows access to new data with minimal manual labeling, requiring only a few data points instead of extensive labeling efforts. This shift unlocks the potential of previously underutilized data. #AI #MachineLearning #DataScience #CustomizableAI #Innovation

60% productivity gains? Yes, it’s possible. Manufacturers are facing labour gaps, ageing workforces, and the risk of losing decades of expertise. But AI-powered knowledge systems are changing the game, delivering up to 60% productivity improvements across industries. Join Dr. Michael Gerstlauer Snowflake Field CTO for Manufacturing at Snowflake’s webinar to see how Conversational AI + Snowflake help manufacturers: - Capture expert knowledge - Train faster - Assist workers with instant answers - Future-proof operations Don’t miss out, Register now: https://lnkd.in/ef-HJk2G #Manufacturing #AI #DigitalTransformation #CatalystBI #Snowflake #Webinar

In many enterprises the biggest bottleneck isn’t AI modeling—it’s embedding intelligence into workflows. Consider the concept of “agentic AI”—systems that don’t just predict, they act across workflows. McKinsey & Company+1 Here’s a practical architecture I see working: Build a business semantics layer that links your KPIs, domain logic and data context. arXiv Deploy AI agents on that layer to automate multistep workflows—not just analytic tasks. Monitor both model performance and workflow outcomes (time saved, error reduced). In your next project: ask not only “Is the model accurate?” but “Is the workflow live and operational?” #AdvancedAnalytics #AgenticAI #DataArchitecture #charan #Oct24

Emilio Valdés highlights the real challenge for CDOs today — navigating messy, fragmented data environments to build clean, connected and governed data foundations. While boards push for quick AI wins, sustainable success comes from building scalable data infrastructure that can support #agenticAI, not just pilot projects. The key? Prioritize data visibility and quality first. AI value compounds over time, and fostering patience and understanding around foundational work is critical for lasting impact. #AI #DataDriven #CDO #DataGovernance #ArtificialIntelligence #DigitalTransformation