AivexaNewsSearch
AI news for builders and product teamsChecked every hour

As AI agents reshape the web, Google opens Gemini and Search to more apps

Collected Oct 28, 2026

Google has introduced a developer platform, the Connected Apps Platform, that lets businesses submit their own services to run inside Gemini and Google Search. The company announced the offering as a public preview for developers in the United States who join a waitlist; Google has not said when it will reach general availability. Connected apps first appeared in Gemini and Search earlier this year, and until now the roster has been assembled largely through direct partnerships.

The underlying idea is that users connect a third-party service to Google's AI surfaces and then act through them, rather than switching to a standalone app. Google's examples include linking Canva to design a flyer or connecting YouTube Music to assemble a playlist. The premise, as the company frames it, is that integrations of this kind could gradually move consumer behavior away from opening individual apps toward reaching the same services through whichever AI assistant they already use. Rivals are chasing the same ground: ChatGPT and Meta's Muse have both been experimenting with comparable connected-service models.

For developers, the onboarding path has three parts. First, integrate with a cloud-hosted MCP server, or with Android MCP through the AppFunctions APIs. Second, use the Google Connected Console to register the app, test it and submit it. Third, choose distribution: an app can be exposed through Search, through the Gemini app, or through both. Once live, developers get dashboards tracking traffic, latency and error rates, and they keep control over brand presentation, including verified logos and authenticated accounts.

MCP, the Model Context Protocol, is an open specification for exposing tools and data to AI models in a standard way, so a model host can call a service's functions without a bespoke integration for every client. A cloud-hosted MCP server means the integration lives on the developer's infrastructure and Google's systems call into it. The Android MCP route with AppFunctions APIs points at on-device capabilities, where an app declares functions the assistant can invoke locally — a meaningfully different trust and latency profile from a remote server. The practical implication is that teams with existing MCP servers may have a shorter path to submission than those starting from scratch, while Android developers face decisions about which functions are safe to expose and how to scope permissions. That reading is inference from the described architecture rather than something Google has detailed.

The catalog Google is building on is already sizeable. More than 50 connected apps are live in Gemini, spanning partners such as Adobe, Experian, Localiza and Peloton alongside Google's own services, which run from Workplace and Photos to YouTube Music and Google Home. Search currently carries a thinner set, including Instacart, Canva, YouTube Music and Yahoo Fantasy Football. Google says the Gemini apps will arrive in Search over the coming weeks. Calm was named as a further new partner, with its meditation app slated to launch on both Gemini and Search soon.

Google's framing of the platform as a route to visibility echoes how search engine optimization once became a discipline in its own right. If assistants become a primary discovery surface, submitting and tuning a connected app could take on the role that optimizing a web page did — and the performance metrics Google exposes, particularly latency and error rates, suggest the quality of an integration will affect whether users keep invoking it. A likely trade-off is that developers cede some direct relationship with the user in exchange for reach inside surfaces they do not control, and that the ranking or surfacing logic for connected apps remains opaque. Oversight of brand assets and authenticated accounts partly mitigates the first concern; the second is unresolved in what Google has shared. Another open question is how much of the value accrues to smaller developers versus the large partners already in the catalog, since a crowded assistant interface tends to favor services users already recognize.

Why it matters: Teams building consumer services now have a defined, if waitlisted, channel into Gemini and Search, with MCP integration as the technical entry point and per-app analytics as the feedback loop. For product managers, the near-term work is deciding which actions are worth exposing to an assistant and whether to target Search, Gemini or both; for engineers, it is standing up an MCP server or Android AppFunctions surface and instrumenting it for the latency and error tracking Google surfaces. If assistant-mediated discovery follows the trajectory of search, early submissions may compound in visibility — an inference, not a guarantee, given the beta's U.S.-only scope and the absence of a general availability date.

Read at TechCrunch · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

A new developer platform will let businesses bring their services to Google's AI experiences, potentially changing how consumers discover and use apps.