DataOps.live

We’ll be at FICO World 26 next week! On Thursday's mainstage, Guy Adams joins FICO's Tara Bhat (SVP Product) and Nikhil Behl as they share how the FICO Platform is expanding its DataOps capabilities. Following the mainstage, join us in the Innovation Center for: - Demos at the booth with Guy and team - 1:1 consultations on operationalizing AI with trusted, governed data - Innovation stage presentation and demo If you are working to move AI from pilot to production, this is where you want to be. #FICOWorld26 #DataOps #AI #DataEngineering

Most conversations about AI-readiness start and end with data quality. Thinking about data quality isn’t wrong, but it's not enough. Treating AI-readiness as a data-quality checklist overlooks something important: quality is a point-in-time assessment. AI operates continuously, and so must your data operations. What AI demands is data that’s reliably complete, accurate, timely, and governed around the clock. That takes pipelines that are tested on every run, governance that's enforced by design rather than audited after the fact, and observability that catches drift before it reaches a model or an agent. That's a different problem than cleaning up a dataset. It's an operational problem. The organizations making real progress on AI have built the DataOps processes and automation to keep their data trustworthy at scale, under change, and across teams. That automated DataOps discipline is what separates a successful AI demo from a successful AI deployment. "Good enough" data has always had consequences. In the AI era, those consequences are faster, wider, and harder to contain. When your organization talks about AI-readiness, is the conversation focused on data quality, on data operations, or both? And where are you finding the biggest gap between the two? #dataops #AIreadiness #TrueDataOps #dataquality

It’s one of the most important questions in enterprise data right now: what does AI-ready data actually look like? It might seem like a general question, but there is a specific answer. #dataops #AIreadiness #TrueDataOps #dataops #automation

🚨 EMERGENCY 🚨 A pipeline breaks. A deployment fails. A “quick fix” makes it worse. Data teams that stitch together manual processes never know when the next disaster will strike. But it inevitably does, and suddenly the team’s not building anymore. They're reacting to what just broke. It’s easy to focus on improving incident response. The better move is to operationalize so that surprises stop happening in the first place. #deployment #deploy #dataops #datapipelines #deploymentissues

We left a Python library unpinned to test our AI agents. When a pipeline pulled a new version, the code broke. If you’ve been there, you know the routine: find the job that failed, grab the logs. ID the exception, patch it, move on. We put Snowflake’s Cortex Code (CoCo) and our own Metis agent to work together, asking them, “can you find the issue, fix it, deploy it, and test it?” Metis, the DataOps specialist, identified the failure precisely. CoCo diagnosed the issue, and explained why the new library version behaves differently and how that broke our code. Here’s where a human team, feeling intense pressure to get it back online, would have pinned it and shipped it. Instead, the AI surprised us. Metis asked CoCo for alternatives and trade-offs. CoCo laid out multiple options. Metis weighed governance, security, and technical debt to decide what to do next… And chose to refactor the code to properly support the new version. The agents understood the time pressure, but opted for long-term impact. As a result, they delivered the best possible solution. Have your AI agents surprised you with clear-headed decisions? #agenticAI #dataops #workflowautomation #TrueDataOps #dataops #automation

Where do data engineers fit now that specialized AI agents work right in Snowflake? #agenticAI #dataops #workflowautomation #TrueDataOps #dataops #automation

"AI compute is not a fixed cost. And our metrics mean nothing anymore." That's how Raymon Gompelman, SVP of engineering at DataOps.live, set up this 4th article in his AI engineering series when he posted it earlier today. It covers three cost categories most organizations don't track, five metrics that replace velocity and story points, and why the CFO's question — what percentage of our AI compute spend produced customer value? — is now an engineering responsibility. As head of an AI-first engineering organization, all of his articles come from first-hand observation. Check it out 👇 https://lnkd.in/gr9RSZcw

Who's headed to Data Summit in Boston next week? Make sure you add Keith Belanger's session to your calendar 📅 #DataOps #DataSummit #Boston https://lnkd.in/eH4ggAwg

Hey #Boston - Tomorrow, Wednesday, April 29, 2026 5:30 PM – 8:00 PM EDT. Location and more in Keith's post below. #Snowflake #SNUG #UserGroup #DataEngineering #DataArchitecture https://lnkd.in/gjtsHf9b

You try it. It works. Now what? DCM Projects makes it easy to deploy Snowflake objects. But data isn’t software. Deployment is just the start, and CI/CD can’t end there. #Snowflake #DCMProjects #CICD #DataOps #DataOpsAutomation