How to extract your best AI prompts and turn them into one-click tools you can use in Chrome

Why Read This from an HCI Perspective

This article shows an attempt to reframe AI not as a ‘conversation feature,’ but as an ‘interaction tool’ that reduces repetitive work. Running saved prompts in one go is convenient from a UX perspective, but it also needs to make it clearer when users act, what permissions are involved, and what the system will do. If you’re an HCI practitioner or researcher, it’s worth reading for the balance between efficiency, trust, and a sense of control.

CIT's Commentary

The core of this feature isn’t that the model has become smarter—it’s that users’ frequent actions are packaged into ‘reusable units of interaction.’ In other words, it goes beyond saving prompts like notes and turns them into a workflow inside the browser. The key question here is whether, as convenience grows, users end up seeing less of the process. Especially for actions that affect the real world—like sending emails or adding calendar events—‘one click’ shouldn’t be enough; ‘one more confirmation’ becomes a safety mechanism. And once users start accumulating many prompts, a good tool becomes less about execution speed and more about a structure that helps people quickly find and understand what they want. In the context of domestic services as well, this feature will become truly practical only when the scope of user data access and the confirmation steps are designed with greater precision—not just when prompts are copied over.

Questions to Consider While Reading

• Q.As the number of saved Skills grows, how will users quickly find the task they want—and distinguish between Skills that are similar to each other?
• Q.When AI reads a page and suggests the next action, which tasks should be automated, and which must always go through human confirmation?
• Q.When prompts are transformed into reusable tools, what learning benefits and what kinds of confusion will this create for both novice and experienced users?