SuperJoin Deep Dive: AI Data Analysis Inside Google Sheets - Connectors, Prompt-Driven Reports, and In-Cell SuperGPT Formulas

Prompt (as read): "act as a CFO, use sales, profit, discount, region and segment and answer: are we growing profitably, are discounts hurting margins, which region or segment looks risky; keep it short and executive; make the report in a new sheet named CFO." It reported growth with a profitability dip, discounts loosely linked to lower margins, APJ (2.77% margin) and SMB as risks. A competitor comparison ("compare sales, number of customers, number of countries, profitability %, average discount %") built formula-driven tables (SUM, COUNT, SUMIFS) in under a minute. Free plan: five queries a day; lowest paid: 500 AI messages a month. Runs can take up to ten minutes; everything is also doable by hand with pivots - the value is speed. The data are synthetic.

Demos: tag fruits and vegetables against a list; extract first names from emails; "create a motivational quote about teamwork"; list the five most populous countries; rewrite a sentence as pirate speak. SuperJoin recommends temperature 0.5. He states lower temperature is more creative - that is backwards: higher temperature means more varied output.