The Notebook Agent just got even better

Back in August we released the first version of the Notebook Agent — bringing agentic capabilities to where data teams were already working. Since then, thousands of teams use it every day.

Turns out, teams needed more than code generation; they needed analytics-specific tooling built on governed data context that can keep up when analysis gets complex.

Sure, the Notebook Agent is great at writing SQL queries — but that was never the hardest part of data work. The real challenge is building upon a query, slicing the data many ways, digging into visualizations, and asking why many, many times.

Today, we’re announcing new capabilities that make the agent a leap above the rest at tackling advanced analytics: new cell types, chart styling, better context awareness, and organization tools. But more than features, we want to show you new workflows the Notebook Agent has unlocked that are only possible in Hex.

Build beautiful, interactive analyses

You've written the queries and sliced the data every way possible. Now you need to turn it into something people can actually understand and use. Traditionally, this means hours of work: writing summaries and creating charts.

The Notebook Agent can do this in seconds. It references your entire project to generate summary metrics and create interactive charts with styling that matches your preferences.

Want the agent to always follow specific formatting? Just tell it in a prompt, or set it once in your User rules file.

Need to make your dashboard interactive? Normally you'd manually add Jinja parameters to every SQL cell. The Notebook Agent can create an input parameter and correctly filter all your downstream queries.

Whether you're exploring for yourself or building something for stakeholders, the agent handles the tedious transformation from "here's what the data says" to "here's what it means."

Organize and tidy up your analysis

Let’s be honest: analysis work is messy. You end up with hundreds of cells exploring dead ends, testing hypotheses, and debugging edge cases. At some point, your project may no longer feel usable, let alone shareable.

The Notebook Agent can help. Because it understands your project's DAG and cell context, it knows exactly what's in use, what's in your published app, and what's redundant.

It can organize and move cells into logical sections. And if you have inefficient queries or unnecessary logic, just ask the agent to clean them up.

Debugging your project is shockingly easy

A hundred cells deep into a project, and something looks... off. But where’s the problem? You could spend hours combing through upstream logic trying to find where things went sideways.

Instead of manually tracing through dozens of cells, just ask the Notebook Agent. It can traverse the entire graph, find the related upstream cells, isolate the issue, and come back with a fix.

For example, the agent can even check against your data descriptions to find an incorrect WHERE clause that referenced a similar but wrong column.

The most powerful platform for agentic analytics

Thousands of data teams are using the Notebook Agent every day, and the results speak for themselves.

Ricky at Ramp builds dashboards in minutes:

Molly at Figma cut her reporting time from days to 20 minutes:

Aks at Notion uses it as a true collaborator:

The Notebook Agent has transformed how tens of thousands of people work with data — we think it will for you, too!