Feeling Behind
“What agent should own this responsibility?”
Andrei Kaparthy’s latest tweet resonated with a lot of people, and I’m seeing it pop up everywhere. I tend to view Andrei as 2-3 years ahead of most people, so when his tweets take off, there’s a good reason to read them.
Like most people who read it, I wondered what it means for me. And, in the spirit of his tweet, I asked Claude how I might incorporate his message into my own work, especially with respect to data engineering.
Here’s part of what Claude wrote back.
And as ChatGPT told me when asked the same question, “What agent should own this responsibility?”
I think the AI’s have a good sense of where things are going, per Andrei’s tweet. We have a rigid view of how we work with AI today, and if we learn to “become one” with the machine, it will be interesting to see where things go. Humans sit as the orchestrator and judge, AI goes the work, and we build much faster (to what end is a separate discussion).
But as I’ve discussed, to be an orchestra conductor or judge, you need to know what “good” looks like. If you want to move fast with agents, master the fundamentals! Or don’t, and find out what happens…
This new workflow will also mean setting up new ways to evaluate the effectiveness of the agents’ work, as traditional means won’t work at the scale we’re talking about in the near future.
We’re all going through a once-in-a-lifetime inflection in our field. You can choose to sit on the sidelines, or be a part of the change and build what’s next.




Can I have please Claude‘s complete answer for you? I have a gut feeling I will not get the same response.