Google reveals Gemini Spark scheduler
- Google on May 12 introduced Gemini Intelligence for Android, while new reporting on May 14-16 pointed to a Gemini Spark scheduler inside the Google app. - Google said Gemini Intelligence will first reach the latest Samsung Galaxy and Google Pixel phones this summer, before expanding to other devices later. - Google I/O 2026 starts May 20, where Google executives are expected to detail Gemini app changes and Android AI features.
Google on May 12 unveiled Gemini Intelligence, a new Android feature set that can automate multi-step tasks across apps on selected devices, according to the company’s Android blog. Separate reports published on May 14 and May 16 pointed to a related product name — Gemini Spark — and to code in the Google app that appears to include tabs for tasks, schedules and skills. The reporting suggests Google is moving beyond one-shot assistant actions toward software that can queue jobs, run them later and draw on connected apps and personal context. Google has not publicly announced a product called Gemini Spark on its official blogs as of May 17, but its own materials now describe Android as an “intelligence system” built to anticipate needs and execute app actions on a user’s behalf. (blog.google) ### Where does the scheduler claim come from? Forbes reported on May 16 that code in Google App version 17.20 includes strings tied to schedules, skills and tasks inside what it described as Google’s agent system. The publication said those strings point to a task scheduler that would let the agent run actions on a timed basis rather than only when a user asks in the moment. (developer.android.com) 9to5Google reported on May 14 that Google app beta version 17.23 introduced the Gemini Spark name and showed a layout with “Chat” and “Agent” views, plus lists for active and scheduled tasks. 9to5Google said its findings came from an APK teardown and noted that such features may change or may never ship. ### What has Google itself confirmed? Mindy Brooks, Google’s vice president of product management, wrote on May 12 that Gemini Intelligence will bring “proactive” features to Android and automate multi-step tasks such as booking rides, shopping and filling out forms. (forbes.com) Google said the rollout will start this summer on “the latest Samsung Galaxy and Google Pixel phones,” with broader availability across watches, cars, glasses and laptops later in 2026. (9to5google.com) Matthew McCullough, vice president of product management for Android Developer, wrote the same day that Android is shifting “from an operating system to an intelligence system.” His post said Gemini can act across selected apps on behalf of the user and that developers can expose tools to the system through AppFunctions, an early-stage API now in private preview. (blog.google) ### What would a scheduled agent actually change? A scheduler would move Gemini from handling live requests to managing a queue of jobs with triggers, timing and retries. The difference matters because a system that runs in the background has to track whether a task already ran, whether a linked app can complete it, and what information was shared with outside services. That reading is an inference from the reported code structure and from Google’s own description of cross-app automation and developer-facing hooks. (developer.android.com) 9to5Google also reported that Gemini Spark may use data from connected apps, chats, tasks, websites a user is logged into, location and other sources. The publication said Google’s in-app text warns Spark is experimental and may share information with third parties or make purchases without asking in some cases, while being designed to request permission before sensitive actions. (forbes.com) ### Which devices are first in line? Google said on May 12 that Gemini Intelligence is coming first to the latest Samsung Galaxy and Google Pixel phones. Forbes reported that the full experience is currently tied to recent flagship hardware, citing requirements including Google’s Nano v3 model, at least 12GB of RAM and a flagship system-on-chip, though those hardware thresholds have not been confirmed in Google’s public posts. (9to5google.com) Google’s developer blog said the first task automation launches involved selected food and ridesharing partners and that the capability is expanding to more app categories and form factors. That gives developers a clearer signal about where Google is pushing next: deeper app integration, not just better chat responses. ### What should readers watch next week? (blog.google) May 20 is the opening day of Google I/O 2026, the company’s annual developer conference, and it is the most likely venue for Google to spell out whether Gemini Spark is a shipping product, a beta feature or an internal codename. Google has already used the run-up to I/O to announce Gemini Intelligence and new Android developer tools, and any public product pages, demos or device lists from the event will be the next hard evidence on scheduling, supported apps and hardware requirements. (developer.android.com) (blog.google)