Google’s latest Gemini voice update carries a promise that should define a useful consumer AI product: people should not have to determine which internal tool is required before asking for help. Yet the app’s growing collection of named features makes users do almost exactly that.
TechCrunch argues that Chat, Spark and Daily Brief occupy separate positions in Gemini’s navigation, each with its own identity. The structure exposes how Google has assembled the product instead of presenting one assistant capable of selecting the right workflow behind the scenes.
This is not just a naming complaint. Every additional mode asks users to learn what a feature does, remember where it lives and decide whether their request belongs there. The burden becomes especially visible when the product itself claims that AI should understand intent.
Daily Brief can turn personalization into unwanted interruption
Daily Brief creates proactive updates from information in Google services such as Gmail and Calendar. In concept, that could provide an agenda containing timely actions without requiring users to search across multiple apps.
TechCrunch’s experience was less focused. The feature reportedly struggled to separate urgent information from optional reminders. It could prompt a user to resume earlier chatbot research or surface subjects previously entered into Google Search.
That behavior illustrates the difference between access and judgment. A system may know that someone researched college scholarships or animal rescues without knowing whether the subject remains relevant, sensitive or welcome. Bringing old activity back into view can feel intrusive even when the underlying data connection was authorized.
Google describes the updates as proactive and personalized. Those are company characterizations rather than proof that every reminder will be useful. Personalization needs a reliable sense of context, priority and timing; otherwise, more data simply gives the assistant more ways to interrupt.
Spark’s useful capability is hidden behind another name
Spark presents a different problem. TechCrunch describes it as an AI agent that can take action for a user, potentially making it one of Gemini’s more valuable capabilities. But Google places it in a separate branded section rather than allowing any suitable request to activate agentic behavior automatically.
A mainstream user should be able to describe an outcome and let the application determine whether the task requires conversation, information retrieval or an agent. Naming the mode may help internal product teams distinguish their work, but it gives customers another concept to understand before they can benefit from it.
The article sees the same pattern across the industry. Claude separates Chat from Cowork, while ChatGPT distinguishes Chat and Work. These interfaces ask users to recognize architectural or product boundaries that the AI is supposed to make less important.
The result can make an advanced system feel less natural. Instead of beginning with intent, a person begins with a menu decision. If the wrong mode is chosen, uncertainty about the interface arrives before the model has a chance to help.
Simpler entry points may become the stronger product
TechCrunch contrasts this complexity with approaches that fit AI into interactions people already know. Apple adds intelligence to existing tools such as Spotlight, Photos, the camera and Siri without requiring a separate vocabulary for each underlying capability.
Messaging-based assistants follow the same logic from another direction. Services including Poke, Ollie, Lindy, Orchid, Lucas, Folk, Tomo and Instinct let users send a message to a contact-like interface. The assistant then decides how to handle the request. TechCrunch cites investor Justine Moore’s view that people increasingly want help through a contact they can text rather than another app they must open.
That approach does not guarantee better answers, privacy or task execution. It does reduce the mental work required to start. A familiar message box creates one path into multiple capabilities, while a collection of brands turns capabilities into navigation choices.
Google’s own statement recognizes the desired destination: users should not have to guess which Gemini feature is appropriate. Reaching it will require more than improving voice control. The app needs to hide more of its internal divisions, apply personalization with restraint and let intent—not product branding—decide what happens next.
