The Distinction Between Production Data Pipelines and Experimentation

I think what I'm seeing a lot of recently, and we're seeing this paradigm shift where there's almost like, has been this dichotomy between production data pipelines and experimentation. What's been missing is kind of these workload tools that make sense on both ends. Like they are what you would be doing experimentation, but they're inherently like thinking about productionizing. These two worlds just completely talk and pass each other. And it's deeply, deeply frustrating for folks.