Google tests a more personalized version of its virtual assistant, Google Now, with a new feature called Explore Interests that lets users pick specific topics
Context & Ripple Effects
Explore Interests extends a line Google started the year before, when it rolled out personalized storm-tracking information in Search results — content chosen for the user rather than queried by them. The new feature moves that from event-triggered cards to user-declared topics inside Google Now, its assistant surface.
The declared-interest model became a recurring Google build: it resurfaced in Assistant audio with Your News Update choosing stories by interests and location, then again when Area 120 launched Keen, a machine-learning app for tracking interests. Explore Interests is the earliest data point showing Google treating explicit topic selection as a signal worth collecting on its assistant.
First-order effects
- Users gain direct control over what their Google Now feed surfaces, shifting the assistant from passively inferred cards to explicitly subscribed topics.
- Google acquires a stream of first-party declared-interest data from users who opt in, sharpening the relevance ranking of everything the Now feed serves.
Second-order effects
- Publishers competing for assistant-feed placement now face a curation layer where user-declared topics, not just query intent, decide which stories reach the reader — the mechanism Your News Update later industrialized in audio.
- Interest-graph products outside search, like Pinterest's territory that Keen later targeted, become adjacent competition for the same attention Google is formalizing inside its own assistant.
Third-order effects
- If the pattern holds, Google's assistant surfaces converge on a single structure — declared interests plus inferred signals feeding automated curation — which is the same architecture behind its later generative experiments like SGE page summarization.
- Personalization becomes a distribution moat: whoever owns the interest graph controls which publishers' content reaches users without a query, concentrating referral power in the platform layer.
The trend: Google has spent a decade converting its assistant from a reactive answer box into a proactive, interest-driven curator, with each experiment — storm alerts, Explore Interests, Your News Update, Keen — adding another signal to the same graph.