Google is killing off Gemini’s Gems in favor of ‘skills’

Google's decision to kill Gems is more than a product roadmap tweak—it's an admission that the "agent store" model, where users assemble a toolkit of specialized bots, may have been a transitional phase rather than the endgame. When Gems launched, the pitch was empowerment: let users craft narrow-purpose agents for everything from resume writing to meal planning. But the rise of Meta's Muse and Instinct suggests the market is consolidating around a different philosophy: one ambient, context-aware assistant that handles the whole spectrum of tasks, rather than a dashboard full of siloed helpers. This is a classic platform tension. Task-specific agents create a rich ecosystem but fragment the user experience and dilute the brand's core assistant. By folding those capabilities into "skills," Google is betting that users want depth inside a single conversational surface, not a Swiss Army knife of discrete tools. It mirrors the broader industry shift from "AI apps" to "AI as the interface"—where the model itself, not a collection of widgets, becomes the product. The competitive subtext is hard to ignore. Meta's aggressive push into all-in-one consumer agents threatens to define the category's default UX. Google, with Gemini already embedded across Search, Android, and Workspace, has distribution muscle but has struggled to articulate a cohesive agent vision. Killing Gems may be an attempt to avoid the perception that Gemini is a patchwork of experiments rather than a coherent assistant. It's a high-stakes simplification: if "skills" feel like re-skinned Gems, users will see through it; if they feel genuinely native, Google could leapfrog the fragmented competition. There's also a developer angle worth watching. Gems had begun to cultivate a creator ecosystem, and sunsetting them risks alienating power users who invested in building custom agents. Google's challenge is to migrate that enthusiasm into the skills framework without losing the community's trust—or handing Meta an opening to court displaced builders. The source report (https://techcrunch.com/2026/09/28/google-is-killing-off-geminis-gems-in-favor-of-skills/) doesn't detail migration paths, but the execution here will matter as much as the strategy. Ultimately, this is the right instinct at the right time. The AI market is maturing from novelty to utility, and users are showing they'd rather have one great companion than a closet full of single-purpose gadgets. Google's willingness to kill a prominent feature—one it hyped heavily—signals discipline and clarity. The risk is that "skills" becomes a rebranding exercise rather than a genuine architectural upgrade. If Google can make skills feel emergent and effortless, it will have made a prescient bet. If not, it will have handed Meta a gift-wrapped narrative about who truly understands the all-in-one future.