Introduction

This article explains how AI can test product activation by acting like a real user and finding friction teams often miss.

What's the problem it solves?

Most activation problems are hard to notice because teams are too close to the product. Traditional user tests miss small but critical friction. AI can catch these issues early by behaving like a confused first-time user.

Quick Summary

The article explains how an AI browser called Comet can be used to test product activation flows. The idea is simple. If AI struggles to complete a task, a real user probably will too.

The author tested several products by asking AI to complete one core action. In travel booking, the AI failed due to slow load times and confusing defaults. In a design tool, the AI got stuck on paywalls, icons, export formats, and hidden UI elements. These are the same issues many real users face.

Some tools performed much better. A scheduling tool worked almost perfectly, with only small issues around time zones and popups. A public railway site completely failed, where AI could not even start an action. The takeaway is that AI can surface real UX and activation problems quickly and cheaply.

Key Takeaways

• AI can act as a realistic first-time user
• Small UX issues block activation more than teams realize
• Icons, defaults, popups, and hidden controls confuse users
• Clean, simple UX leads to smooth activation
• AI testing can reveal problems humans miss

What to do

• Pick one core action and test it with AI
• Watch where AI gets stuck or confused
• Fix defaults, labels, and hidden UI elements
• Remove unnecessary popups during activation
• Use AI tests before running expensive user research

‍