Google’s annual developer conference is going on right now, and if you’re wondering what’s been announced so far, you’re in the right spot. From AI to, well, more AI, here’s everything Google has discussed on stage so far.
Google Photos is getting a Gemini-powered upgrade
The days of searching endlessly through your photo library are over. With Gemini coming to Photos this summer, you’ll be able to ask for your AI assistant to find specific images in your collection. Looking for your license plate? Trying to find memories of your daughter swimming? Gemini is just a question away.
Google Gemini 1.5 Pro is available globally today
And Gemini 1.5 Flash speeds things up
The latest version of Google’s LLM is coming to developers globally today, and you can access it through Gemini Advanced. It looks pretty impressive, and might even boost your productivity tenfold. For example, an hour-long Meet call can be summarized into a few handy bullet points. Say goodbye to daydreaming during your next all-hands meeting. Google also demoed a pretty impressive explanation of physics in basketball in NotebookLM using two friendly voice assistants.
Meanwhile, Gemini 1.5 Flash is all new. It’s faster and more efficient than its Pro-named sibling, and it’s now available in Google AI Studio and Vertex AI.
Get ready to shop with AI
Turns out our bingo predictions were right on the money. Google Shopping is getting some early AI tools that can help handle things like online returns and DoorDash
Project Astra is the next step in Gemini
Google describes this as a “universal AI agent helpful in everyday life,” but what it really looks like is a functional version of what Humane and Rabbit promised on their dedicated hardware. It’s an improved version of Gemini running on various devices — including smart glasses — that is constantly trying to help you, whether it’s through brainstorming or just by finding your lost items.
But, as you might expect by the name, this seems fairly far off. Google says some of the experiences it’s capable of are coming to Gemini later this year. We’ll be going hands-on with Project Astra later today, so stay tuned for more impressions coming this week.
New AI-powered creation tools are here to rival OpenAI
Google had a number of creatives talking about its latest Gemini-powered toolsets, designed for making everything from images to music to videos. From Mark Rebillet — who provided a pre-I/O set — to Donald Glover, it’s clear Google wants musicians and filmmakers to embrace tools like Imagen3, Music AI Sandbox, and its new video creator, Veo.
That said, Google is going slow on this front, keeping these tools for select creators through its Labs program. Whether any of this will hold up to scrutiny — and how YouTubers will react to Google pulling from its library to develop tools like Music AI Sandbox — remains to be seen.
Google Search is all about generative AI
“Google will do the Googling for you.” That’s how Google opened the section on search during today’s keynote, and it’s not kidding. AI Overviews, which have been in beta since last year, are the default now. It’s rolling out later today, and coming to one billion people by the end of the year. In my personal experience, these AI summaries continue to provide alternatingly incorrect and plagiarized answers, but I’m sure this will be fine, right? Right?
Google is using Gemini to embrace multi-step reasoning. It’ll break bigger questions down into specific elements to, in theory, give you all of the information you need about restaurant recommendations or gym memberships on a single page.
There are some pretty cool elements here, including meal planners using online recipes that push for you to view the recipe on the actual website, not within AI Overviews themselves. Meanwhile, video searches are coming to allow you to ask troubleshooting questions while delivering prom It seems like the company is, in some ways, doing its best to push for independent sites to, you know, not die out, all while hoping you spend as much time on search as possible. We’ll see how that works out in practice.
Workspace is getting even more AI
As if summarizing meetings wasn’t enough, Google is using Gemini to summarize emails, both individually and in threads. Beyond that, though, is the option for Google to look through your entire inbox to find related emails. Whether you’re putting together roof quotes or planning a trip, it seems like this could help wind down your need to rely on custom search filters in Gmail.
Gemini is also coming to apps like Tasks, Calendar, and Keep, allowing you to use the chatbot to ask questions and upload images.
Chip is an AI-powered Clippy
Google is working on a prototype “Gemini-powered teammate” that, to my ears, sounds a lot like Microsoft’s least-favorite assistant. Chip works within Workspace apps to track ongoing projects, and acts as its own individual being. This isn’t a prompt box — Chip has a Google account, lives in your chats, and works to assemble outlines on your behalf.
Gems allow you to personalize Gemini
Google can’t get enough of the Gem-branding, I suppose. With Gems, you can train specific chatbots to accomplish certain tasks or to focus on aspects of your daily life. Design a fitness trainer or a home chef and it’ll encourage you to work out every morning or try new recipes. It’s coming for Gemini Advanced users later this year.
Circle to Search can do homework for you
Circle to Search is a surprisingly useful tool, and with its latest update, you can show it things like physics problems to walk you through step by step. Rather than simply giving you the answer, the idea is that Circle to Search could act like a tutor for you — or, more likely, the students in your life.
Unlike many of the aspects on this list, the first round of homework-based tools are available today, with more STEM problems coming later this year. Meanwhile, Google confirmed Circle to Search is coming to even more Android devices later this year.
Gemini on Android knows what you’re looking at
More than an hour into the show, Google finally brought up everyone’s favorite mobile OS. Gemini on Android is getting contextual advancements, with suggestive chips for when you’re watching a YouTube video or looking at a PDF. It’s limited to just those two things for now, but being able to ask for specific sections in an 80-page pickleball rulebook does sound pretty cool. And — bonus — it’s exclusive to Android.
Gemini Nano will look for fraud risks
We’re finally seeing some new advancements to Google’s on-device LLM. The company is working on a system that monitors your phone calls on-device — totally locally — to watch for fraud risks. So the next time you get a suspicious phone call from your “bank,” Google will look to determine whether or not it’s the real deal. If it’s not, your phone will buzz and alert you to the suspected scam. It’s not ready for primetime, but we’re expected to hear more about it later this year.
We’re still going and going
This year’s I/O is a marathon, not a sprint, so keep it locked here for everything announced so far.
