Turning Repeated Team Questions Into a Slack Agent That Knows Its Limits

Can an Agent Take Over the Repeat Questions?

Shuyi Chen's case study on UX Collective walks through how a remote design lead sorted two and a half years of team requests and found that most did not need her judgment at all. Roughly half of the 2,413 asks already had answers sitting in a doc, a Figma file or an old thread, so the real problem was findability. The author then built a small Slack agent called le goat, and the piece explains how it was tuned to show its steps, lead with the answer and say plainly when it is unsure.

For brand builders and design leaders, the article is a practical model for handling internal knowledge, not just customer-facing content. Its most useful lesson is that an agent is only as trustworthy as the documents behind it, and that conflicting specs can quietly undermine confidence. Anyone running a design system, a shared brand guideline or a remote team should read it before automating support work.