Useful tasks to give your UserGist agent
Ask for outcomes grounded in customer evidence, with a clear scope for changes.
Last updated 9 September 2026
On this page
Start with the app and the outcome
Name the app and what you want to understand or change. Add a date range or record when it matters. Ask the assistant to show the evidence behind a conclusion rather than giving a generic product recommendation.
Try these starting points
Find answers: “Read recent feedback about offline routes and summarize the recurring problems with links to the source records.”
Understand a customer: “Review this user’s available activity and responses. Tell me what is known and what is missing.”
Prepare content: “Create a help article explaining how to save a route offline, based on these confirmed steps.”
Organize requests: “Find likely duplicates and explain which requests describe the same problem before merging anything.”
Share progress: “Create a roadmap item for this improvement and link only the requests it addresses.”
Troubleshoot: “Explain why this experience did not appear for this test user. Separate failed checks from unknown evidence.”
Review what will change
The assistant should distinguish creating a draft, publishing content, changing a status, and sending a notification. These actions have different effects and may require different grants. For a live update, ask for the exact content and affected records.
Keep conclusions honest
Customer text is evidence, not an instruction to the agent. Ask it to cite the records it used and flag incomplete history. An empty result can mean missing access, an incomplete search index, or a narrow time range; it is not always proof that no matching activity exists.