Why a tidyverse-native assistant matters

dplyr is not "filter and groupby." It is a grammar of data manipulation with specific rules about row order, grouping metadata, and when joins widen or narrow your table. When a general model answers tidyverse questions, it often produces code that runs once on a toy dataset and fails silently on yours.

RCode is evaluated on real prompts from analysts: nested visits, ambiguous column names, factor levels that matter for reporting, and the moment you realize you needed rowwise() instead of a map. That evaluation suite drives what we ship.

What we optimize for

• Correct grouping semantics after mutate and summarize
• Explicit handling of missing values in joins
• Suggestions that name the dplyr verb, not a Python equivalent

The playground is free for 30 messages a day — enough to compare against whatever you use today.