On May 19, 2026, Google used the opening of Google I/O 2026 to roll out a sweeping update to the Gemini app[1]. The centerpiece is Gemini Spark, a personal agent built to act on the user's behalf 24/7, paired with Daily Brief, Gemini Omni, Gemini 3.5 Flash, a new design language called Neural Expressive, and an upgraded macOS app. Together, they push Gemini from a "helper that answers questions" into a "partner that gets work done"[1][2].
Gemini Spark, a personal agent that keeps working around the clock
The headliner is Gemini Spark[1]. Spark runs on Gemini 3.5 with Google's Antigravity agent harness, and lives on dedicated virtual machines on Google Cloud[2][3]. Because it runs in the cloud, Spark keeps working in the background even after the user closes a laptop or locks a phone[1][2].
Google calls out three concrete use cases[1]. First, recurring tasks: Spark can parse a monthly credit card statement and flag new or hidden subscription fees. Second, "teach it new skills": users can tell Spark to monitor their inbox for ongoing updates from their kids' school, extract critical deadlines, and send a daily digest to themselves and a partner. Third, end-to-end workflows: Spark can synthesize raw meeting notes scattered across emails and chats, package them into a polished Google Docs file, and even draft a kickoff email for the project[1].
Spark is designed to operate under the user's direction, and Google emphasizes that it asks for confirmation before high-stakes actions like spending money or sending emails[1]. Spark rolls out to trusted testers this week, with a U.S. Beta for Google AI Ultra subscribers planned for next week[1][2].
Daily Brief and Gemini Omni, with 3.5 Flash reshaping the foundation
Launching alongside Spark is Daily Brief[1]. Connected to Gmail and Google Calendar, it compiles urgent updates and upcoming events into a quick morning digest and suggests immediate next steps[1]. Built on the Google Labs experiment "CC", it learns from simple thumbs-up / thumbs-down feedback over time[1]. Daily Brief begins rolling out the same day to Google AI Plus, Google AI Pro, and Google AI Ultra subscribers in the U.S.[1].
For creative work, Google introduced Gemini Omni, a new model that blends text, image, and video prompts to generate polished video[1]. Users can apply cinematic zooms or swap out backgrounds with a simple instruction, drop in clips from their camera roll, and finish with single-tap templates—no specialized gear required[1]. Omni also supports custom AI avatars that look and sound like the user[1]. Gemini Omni begins rolling out the same day to Google AI Plus, Pro, and Ultra subscribers worldwide[1].
Underneath, the model layer was refreshed too. Gemini 3.5 Flash is positioned as the first in Google's next generation of models, combining "frontier intelligence" with lightning-fast action, and it powers the long-horizon agentic workloads that Spark depends on[1][3].
Neural Expressive, macOS, and the road ahead for 900 million users
The UI got an overhaul as well. The new design language "Neural Expressive" introduces fluid animations, vibrant colors, refreshed typography, and haptic feedback[1]. Gemini Live's conversational mode is now woven into the main app, so users can flip between typing and speaking without missing a beat, and the mic has been re-engineered so users can think out loud without being cut off mid-thought[1]. Responses move beyond walls of text into a mix of imagery, interactive timelines, narrated videos, and dynamic graphics[1]. Neural Expressive begins rolling out globally the same day on web, Android, and iOS[1].
Desktop is moving forward too. Google's macOS app for Gemini ships the same day, with Spark integration and new voice features promised for later this summer—the voice work aims to clean up filler such as "um" and reshape free-flowing speech into a precise draft right where the cursor sits[1].
Third-party reach is expanding. Spark gains new MCP connections to Canva, OpenTable, and Instacart at launch, with more partners on the way[1]. Google also plans to add texting and emailing Spark, custom sub-agents, and local browser control in the coming weeks[1].
The user base keeps growing. Google says more than 900 million people across 230 countries and over 70 languages now turn to Gemini for help every month, up from 400 million a year ago[1].
Summary
Gemini Spark reframes Gemini as a partner that keeps running in the cloud and steps into real tasks under the user's direction, rather than a chatbot that simply answers questions. Paired with Daily Brief, Gemini Omni, Gemini 3.5 Flash, Neural Expressive, and the macOS app, the Gemini app is being rebuilt around acting across Workspace and a user's broader digital life. The U.S. Google AI Ultra Beta will be the first real-world test of how much Spark can genuinely take off users' plates.
Source [1]: https://blog.google/innovation-and-ai/products/gemini-app/next-evolution-gemini-app/
Source [3]: https://9to5google.com/2026/05/19/gemini-app-google-io-2026/
