After this lesson you can turn each answer a Use AI column returns into its own column, then filter and sort on it.
1:13 video with captions, written steps and transcript. Checked against Duckrow on October 3, 2026.
AI Columns and Jev · Lesson 2
90 seconds
Narration: preview voiceA private equity team screens HVAC companies against its thesis. The companies are example data.
Ask your Use AI prompt for named answers, such as fit_score and fit. Click a finished cell to open Cell details: each answer is a card with Add to column. Name the column and press Add Column. Numbers become a Number column and the rest Text, filled for rows already run and every later run. Then filter, such as Fit score greater or equal to 7.
About five minutes, in your own workspace.
You're done when your table shows only the rows the AI scored 7 or higher.
Three quick questions. Nothing is saved.
An AI answer is only useful if you can sort and filter on it. Here's how to give each answer its own column.
This team's Use AI column ran the lead fit template, which asks for four named answers: a fit score, a fit of strong, maybe or weak, reasons, and missing information.
The cells say process complete. Click one, and Cell details shows a card for each answer.
On fit score, click add to column. Name it Fit score, and add column. It fills for every row already run, as a number. Do the same for fit.
Now filter like any column. Fit score, greater or equal to seven, and the screen narrows to the strong fits. You can also filter the AI column itself, for example rows that have an error.
To recap. Ask the prompt for named answers. Open Cell details and add each one to a column. Then filter and sort them like any other. Next, pick a model and know what it costs.