Skip to main content
This standalone walkthrough uses an imaginary company selling a café loyalty app: buy nine coffees and get the next one free. The product helps cafés bring customers back. 1:46 video, rendered September 2026. Recorded in the native product interface with fictional demo results.

​
What you’ll learn

  • Give a saved agent a useful research question.
  • Run it with the context in your table.
  • Review opportunities and supporting sources together.

​
Give the agent a useful question

The saved agent Café loyalty fit asks how the loyalty app could encourage another café visit, without assuming falling sales. Keep outputs short: Evidence, Opportunity, Source and Decision. Enable Web research when the answer needs public information. Ask for Needs review when evidence is missing.

​
Run it on the table

Choose Add to table and select the table. The example selects Use the whole row for names, companies and websites. Other agents may need Map columns to inputs. Check the output columns and run a small batch: here, five fictional owners.

​
Inspect the results

Dana’s weekday coffee-and-pastry menu supports another breakfast visit. Priya’s Saturday brunch suggests inviting guests back on a weekday. Missing evidence stays in Needs review; a wholesale equipment business is Not a fit. Open a row to inspect the finding and source together. Check the source before outreach, then focus the email on what the customer gets. The video reuses the same saved agent on a second table. Hannah’s lunch menu produces a different offer: reward the next lunch visit. Open the linked source to check the finding before using it. Enrich people data covers work-email lookups and verification. Send your first sequence shows how to introduce your product with a useful offer.