What do data modeling, artificial intelligence (AI) adoption, and frictionless CI/CD in Snowflake have in common? According to Keith Belanger—Senior Data Architect and Snowflake Data superhero at DataOps.live—they’re foundational elements often skipped or misunderstood in the race to build fast.
On a recent episode of our Behind the Data podcast, Keith walked us through what organizations are getting wrong (and right) when it comes to AI readiness, modeling strategies, and Snowflake adoption.
AI is accelerating development, but verification is non-negotiable
Keith is particularly interested in how AI plays a dual role in data workflows. Not just in how it powers business insights but also in how it speeds up the work of developers and architects.
“I like to say trust but verify, and I put a lot of emphasis on the verify at the moment,” Keith said. “What is the accuracy of the data coming through?” He continued, “Some AI can get 70%, some of it's 80%, maybe you can get up to 90%. You tell me if 70%, 80% or 90% accuracy is something you want to make business decisions off of.”
On the development side, Keith sees major time-saving potential: “I use AI as an accelerator for me to do my job. It remembers how I work. I don’t have to write the same five-page stored procedure again.”
But, he warns, the tool isn’t a substitute for experience. “It’s not going to overcome your experience. If you don’t know what you’re doing, you’re just going to accelerate your mistakes.”
Rediscovering the lost art of data modeling
Though less flashy than AI, data modeling is critical to getting trustworthy, scalable data systems in place, especially when preparing data for machine learning and analytics.
“Every organization used to do data modeling and data design. We would design it, then build it. Now we just build,” Keith said.
Why did it fall out of favor? “The cloud industry has made it very easy to just build things. We have unlimited storage, unlimited compute… just have at it.” But Keith argues this leads to technical debt and messy ‘popcorn architecture.’
Watch the full Behind the Data episode:
Embedding structure in agile teams
Keith emphasizes that modeling isn’t just about technical design. It’s also about creating a shared structure for communication and collaboration.
“If you’re finding out today that you need to build a data product today, you haven’t been communicating with the business,” he said. “There should be a roadmap, quarterly goals… That gives you the space to start conversations, design collaboratively, and build blueprints before the first line of code is written.”
Reflecting on his time leading architecture at a Fortune 100 insurance company, Keith described how layered roles helped coordinate across a large organization: “We had the big picture, then I had architects under me in different domains. And then each sprint team understood their perspective of the big picture. And it had to all come together.”
And he’s seen the limits of today’s data literacy firsthand. He says: “What was surprising to me is just the lack of resources that are available to people to teach and educate and learn how to do data modeling. There’s a huge gap.”
Making Snowflake truly frictionless
As self-proclaimed ‘frictionless czar’ at DataOps.live, Keith is working on simplifying CI/CD for Snowflake users by embedding DevOps tools natively.
“It usually takes weeks or months to buy infrastructure. Well, with Snowflake, it was like, here's your URL, and minutes later you have a full-blown enterprise data warehousing solution. Now we need six months to try and figure out how we're going to get stuff into it and manage it and roll it back. That’s the friction.”
“So our mission is to bring DataOps.live capabilities natively into Snowflake… so you could be a Snowflake customer and go to the marketplace and say I want to use CI/CD, click on a button. And there it is, right ready for you to go.”
Snowflake has invested directly in DataOps.live to bring these capabilities closer to the core platform.
Final thoughts: AI won’t replace you, but someone using AI might
Keith sees a future shaped not by hype, but by fundamentals. “I don’t think my job will be replaced by AI,” he said. “But if you’re not a data practitioner in some way, and you’re not leveraging AI, then you might get bumped by somebody who is.”
From accelerating code to shaping data for intelligent systems, the key is balancing speed with structure and keeping human experience at the heart of your stack.
For more from Keith on mentoring teams, building data culture, and designing smarter pipelines, listen to the full podcast episode.
Share
