Agent feature

An agent feature means the AI doesn't just answer your questions—it actually runs errands for you. It's like telling an assistant 'book this restaurant for me,' and the assistant makes the call and finishes the reservation. Ask the AI to do something, and it searches, sends emails, and organizes your schedule. That's a step beyond a plain chat window where you ask and get an answer.

The catch is that this feature is usually split off under its own name and its own screen. Even within the same app, the regular chat mode and the agent feature often live behind different icons and different menus, so users have to figure out every time which one their task belongs to. It's as if you had to first decide 'is this a job for my assistant, or for the front desk' before you could even ask for help—so even though the feature is useful, it ends up being a hassle to use.

The article explains, "Spark is genuinely useful thanks to its agent feature, which carries out tasks on the user's behalf... but the fact that Google split this off into a separate brand is called out as a problem." The point that's easy to misread here is that the agent feature itself isn't the issue—the problem is the screen structure that separates it into its own mode and forces users to choose.

Try making a request to an AI chatbot or assistant app that requires action, not just an answer. For example, ask something like "check this week's weather, and if there are any rainy days, put together a note with those dates and a reminder to bring an umbrella." See whether the app just replies with text or actually breaks the task into steps and carries them out—that's where you'll feel the difference an agent feature makes.