Unlimited charting & reference tags in prompts

Unlimited charting: no more 10,000 row limit

We've just shipped one of our most requested features: unlimited charting for large datasets!

For too long, our 10,000 row limit on charts has been a major frustration point. We know you've had to work around this limitation by pre-aggregating your data, applying filters, or sampling before creating visualizations. Those days are officially over.

You can now create charts directly from Pandas DataFrames with millions of rows without any preprocessing. Whether you're working with sales data, user analytics, or any large dataset, simply select your columns and chart away - no more workarounds needed.

Check out this quick video to see unlimited charting in action:

For datasets larger than 1 million records, we've implemented a smart preview mode to keep your charting experience responsive. You'll see an initial chart built from sampled data, allowing you to quickly experiment with different visualizations and configurations. Once you're satisfied with your chart setup, simply press Apply to run the aggregations on your complete dataset.

This enhancement works seamlessly with all our existing chart types, giving you the freedom to explore your data without constraints.

We're also actively working on bringing the same unlimited capabilities to Spark DataFrames, which will unlock charting for even larger datasets with hundreds of millions of records. More on that soon!

Reference tags in prompts

@References bring more precision to every Deepnote AI prompt in Workspace Home. Now, when you type @, you can add integrations like @bigquery - sales data or modules such as @sales_prediction directly into your request. Deepnote AI uses these references to anchor its answers in the exact context you choose, making insights more relevant and actionable.

Referencing integrations and modules in your prompt ensures that Deepnote AI draws from the most trusted and approved sources in your workspace. For example, referencing a module tells Deepnote AI to use your team's official logic and metric definitions, while referencing an integration ensures it bases answers on the right data source.

By guiding Deepnote AI with references, you get answers that follow your team's established rules and standards. No more generic responses - every insight is grounded in your specific context and trusted sources.

Soon, you'll be able to use references in the AI chat inside notebooks too, bringing this same level of accuracy and control wherever you're working with Deepnote AI.