Live-demoing data science in the age of agents
Live-demoing data science in the age of agents
I deleted my own demo an hour before showtime, then rebuilt it live with an AI agent.
Hello fellow datanistas,
On October 2 I stood up in a lecture hall at UMass Chan Medical School and live-demoed an hour-long reanalysis of my published IRED-88 enzyme data, with a coding agent paired to a marimo notebook. The morning of the talk, marimo launched Lens and Studio, and I folded both into the demo the same day.
The agent flew through the charts, but stumbled three times on the epistasis transform I knew from hard-won experience, even after I let it read the answer. That gap between agent output and ground truth is exactly where the practice of science lives.
In this post, I tell the story of the hour through the prompts I dictated, verbatim, with screenshots of every artifact the run produced.
Happy coding,
Eric
P.S. Tools change; the ability to learn any field on demand does not. Daniel Chen and I are organizing a retreat on how to learn anything with AI, February 2027. Applications open to the waitlist first: https://learn-anything.nonlinearlabs.ai/
If you liked this post from Data Science Programming Newsletter, share it around!
© 2026 Eric J. Ma
548 Market Street PMB 72296, San Francisco, CA 94104