Is it vibe analysis or vibe analyze? Neither quite flows off the tongue like vibe coding — but it’s just as powerful — especially with a memory graph that shows how you and the AI are vibing together to understand the data.
We initially built iterative SQL AutoCorrect (some might even call it an agent) to help with minor errors in syntax, misspellings of column names, and other common mistakes like forgetting an aggregate in a grouped query. Just click on the fix button in memories and we’ll take iterative passes to correct the query, fetching additional semantics and metadata as needed.
However, we quickly found out it could be used for far more than just fixing minor errors in formulas and case statements. Your SQL snippets (our intermediate ArcQL helps with composition) with the errors didn’t even need to be in SQL. Just describe the measure you want to see or segmentation you want to add and let our AutoCorrect powered by memories do its thing. The better the description — or even pseudo code to be more precise — the better the results.
Ask it for just “cost per square feet” of SF rentals or ask it to “use both min and max” calculate the average of the average of ranges to be a little more accurate. Or describe as segmentation of eastern vs western halves of SF and it’ll pick a longitude roughly splitting SF.
Pair it with our suggest, extend, and combine and now you’re really vibing with your data.
Check out the vibes in HyperArc with your own data.
Share this post