After this lesson you can add a Use AI column from a template, read its cost a row, and run it on 10 rows.
1:34 video with captions, written steps and transcript. Checked against Duckrow on October 3, 2026.
AI Columns and Jev · Lesson 1
90 seconds
Narration: preview voiceA SaaS team writes an opening line for each account. The names and companies are example data.
In a table, click Add column and choose Use AI. Keep Flexible mode, pick a model and read its cost a row. Under Prompt, open Templates and choose one, such as Opening line: your columns drop in as chips, and you replace the bracketed text. Then Save and run 10 rows, open a finished cell to read the answer, and run the rest once it reads well.
About five minutes, in your own workspace.
You're done when ten rows have an opening line you'd be happy to send.
Three quick questions. Nothing is saved.
Writing a good first line for every lead takes hours. A Use AI column writes one for each row, from your own columns.
Click add column. Under Enrich, Use AI shows its price: from one credit a row. Choose it, and the Use AI panel opens.
Keep Flexible mode, which answers whatever your prompt asks. The model is GPT four o mini, and the panel says about two credits a row. That's an estimate: each row is charged what it really uses.
Next to Prompt, open templates and pick opening line. The prompt fills in, with your name, title, company and website columns as chips. Replace the bracketed text with what you sell. To add another column, type a slash.
Open save, and choose save and run ten rows. The cells show processing, then process complete. Click one to read the opening line it wrote.
Read a few. If they sound right, run the rest of the table. Each table has one Use AI column, and it's named Correlation.
To recap. Add column, Use AI, and read the cost a row. Start from a template, and type a slash to add columns. Run ten rows, read them, then run the rest. Next, AI answers you can filter.