The Future of Agentic Data Science
AI agents have revolutionized software development. Engineers vibe code prototypes in an afternoon, refactor entire codebases in a single chat, and ship faster than they ever have.
But what have agents actually done for data science? Not nearly as much… Not yet.
That's about to change, and the path forward looks nothing like the one software took.
Join Thomas Wiecki (PyMC Labs, Decision.AI) and Hugo Bowne-Anderson (VG) for a live conversation about what it actually takes to do data science with agents, and why this might be the moment data science finally delivers on its original promise of making better decisions.
They'll dig into why the agent playbook from software development doesn't transfer to data work, how the data scientist's job is changing in real time, what an agent in your team chat actually does for you, and why the most valuable skill is now the one nobody trained you for.
What We'll Cover
Why data science isn't software: Ten ways to write a function all work. Ten ways to analyze the same dataset don't. That asymmetry breaks every assumption the Cursor era was built on.
Correlation isn't causation, and most data tools blur the line. Business questions are usually causal: what changes if we do this? Most data tools quietly answer a different question: what's correlated with what? The gap between those two answers is where bad decisions live.
Why running an analysis isn't the same as having an answer: Every analysis involves dozens of invisible choices that change the result. What it takes to make those choices visible, contestable, and correct.
An agent in your team chat: Slack and Discord agents that run scheduled reports, query the tools you already use, and ship insights without anyone opening a notebook. What it looks like in practice, and whether it replaces the people who used to do that job.
The new shape of the data scientist's job: When the agent runs the analysis, you stop being the person who runs the analysis. What's left, what matters more, and what's gone forever.
Judgment is the new intelligence: Today's hard judgment calls become tomorrow's automatically approved tool calls. Where the line is moving and what you should actually be skilling up on.
Thinking from the data generating process: The mode of thinking data scientists were never trained for, but increasingly the thing only you can do. Why agents make this skill more valuable, not less.
Register to join live or get the recording afterwards.