Google Gemini gets Daily Brief, Gemini Spark, and a new look — here's what's actually changed
Google I/O 2026 brought a redesigned Gemini app with a new UI, a personalised daily summary, and an agentic assistant that works while you sleep.
Google used I/O 2026 to announce a significant update to the Gemini app. The short version: it looks different, it now gives you a personalised morning briefing, and a new agent called Gemini Spark can handle tasks in the background while you get on with your day. Here is what each of those actually means.
A new look: Neural Expressive
The first thing you will notice is that the app looks and feels different. Google has introduced a new design language called Neural Expressive, which replaces the previous text-heavy response format with something more dynamic. Important information now sits at the top of a response in bold. Where relevant, you will see inline images, timelines, narrated videos, or interactive visualisations rather than dense paragraphs.
Gemini Live, previously a separate mode for voice conversations, has been folded into the main app. You can now move between typing and speaking without switching context or navigating to a different screen.
This redesign is rolling out now on Android, iOS, and the web, globally.
For most users, this is the change they will experience first. Responses should be easier to scan, and the app should feel more cohesive now that voice is no longer tucked away.
Daily Brief: your morning summary, done properly
Daily Brief is a personalised overview that pulls from your connected apps, such as Gmail and Google Calendar, along with your recent Gemini conversations, and organises everything into a single view each morning.
It splits into two sections. “Top of Mind” surfaces your immediate priorities with suggested next steps. “Look Ahead” focuses on longer-term goals you have been working towards. You can refine it over time with thumbs up and thumbs down feedback, and it will adjust accordingly.
This is not Google’s first attempt at something like this. An earlier version called Daily Hub was planned for the Pixel 10 series but was quietly pulled after poor feedback during testing. This relaunch takes a different approach, embedding the feature directly into the Gemini app rather than tying it to specific hardware, and building it around your existing Workspace data rather than treating it as a standalone product.
Daily Brief is rolling out now to Google AI Plus, Pro, and Ultra subscribers in the United States. You will need to have connected your Google apps for it to work.
If you currently start your morning by checking Gmail, then your calendar, then maybe a to-do list, Daily Brief is trying to collapse that into one place. Whether it actually saves time will depend on how well it reads your patterns, but the intent is straightforward.
Gemini Spark: the agent that keeps going when you stop
Gemini Spark is the most significant announcement in this update. It is a cloud-based agent, running on Gemini 3.5 with something Google calls the Antigravity harness, and it is designed to carry out multi-step tasks on your behalf, even after you have closed the app.
The examples Google has shared give a good sense of the scope. Spark can parse your monthly credit card statements and flag recurring subscription charges you may have forgotten about. It can monitor your inbox for school communications, extract important dates, and send you a consolidated digest. It can take scattered meeting notes from across your emails and chats, synthesise them into a polished Google Doc, and draft a follow-up email to go with it.
Because it runs on Google Cloud rather than your device, it continues working in the background. On Android, you will see live progress updates through a new interface called Android Halo. On desktop, Google is bringing Spark to the macOS app this summer, where it will also be able to work with local files.
Security is worth addressing directly here, because handing an AI agent access to your inbox and documents is a meaningful decision. Google says Spark runs in isolated, ephemeral virtual machines, with each task starting fresh so data does not carry over between sessions. A secure Agent Gateway enforces data loss prevention policies, and user credentials are never directly exposed to the agent. For high-stakes actions, such as sending an email or making a purchase, Spark is designed to ask for your confirmation first.
Spark is rolling out to trusted testers this week, with a Beta for Google AI Ultra subscribers in the US following shortly after.
The macOS app and what is coming next
The macOS app is available to download today, though Spark and the new voice features will arrive later this summer. New MCP connections to Canva, OpenTable, and Instacart are also live now, with Spark able to use those connections to take action on your behalf in the coming weeks.
Further down the roadmap, Google is planning to add the ability to text or email Spark directly, create custom sub-agents for specific workflows, and let Spark operate your local browser.
One thing worth knowing about usage limits
Google is also changing how it manages usage. Rather than a flat daily prompt limit, it is moving to a compute-based model that accounts for the complexity of what you ask. A simple text question uses far less compute than generating a video or running a long coding task, so limits will reflect that. They refresh every five hours until you hit your weekly allowance.
This is a sensible change, but it is worth being aware of if you are a heavy user. Tasks run through Spark are likely to consume more compute than standard prompts, so Ultra subscribers who rely on agentic workflows will want to keep an eye on how the new limits play out in practice.
The bigger picture
Gemini now has more than 900 million monthly users across 230 countries, up from 400 million a year ago. These updates reflect where Google is pushing the product: away from a question-and-answer interface and towards something that monitors, organises, and acts on your behalf over time.
Spark puts Google squarely alongside OpenAI’s Operator and Anthropic’s agent offerings. The MCP integrations are the more telling detail here. The ability for Spark to connect to third-party services like Canva and Instacart, and eventually operate a local browser, suggests Google is building for a world where Gemini is less of an app you open and more of a layer running underneath everything else.
For now, the Neural Expressive redesign and Daily Brief are the changes most users will encounter first. Spark, for most people, is still a few weeks away. But it is clearly where the team’s attention is focused.