News
The tech world unofficially declared May AI month this year. Following the first day of Microsoft Build 2025 — where Copilot dominated the conversation with major integrations across Windows, Office and Azure — Google shared its vision at Google I/O 2025.
One thing is clear from both events: we’re no longer in an era where AI is a feature. AI is the product.
Microsoft’s focus was squarely on embedding Copilot deeply into its ecosystem, enhancing the Windows 11 experience and giving developers powerful new AI tools. Google used I/O 2025 to spotlight Gemini, its suite of AI models and how these tools are reshaping everything from search and email to meetings and video creation.
Globally, Google I/O may not have made quite the same media splash as Build — which isn’t surprising, given the intense anticipation around Microsoft and OpenAI’s next Copilot moves — but it was no less packed with significant news.
Google provided a high-level overview in a blog post listing the 100 announcements made at Google I/O. The majority centered on AI: big upgrades coming to the Gemini app, advancements in generative AI tools and progress updates for core AI models, along with new ways for users to interact with them. While many of the announcements were consumer-focused, a number of them touched on the digital workplace and in particular, the Google Workspace suite.
Workspace — home to tools like Gmail, Docs, Drive and Meet — is rapidly becoming a showcase for the AI ambitions outlined at I/O. The company claims that Workspace serves more than two billion AI-powered assists each month. And the roadmap ahead shows that was just the beginning.
Gmail Becomes Your Personal Assistant
Gmail is the first stop for many workplace Google customers, but Google's ambitions for it go far beyond email, to a central place where work happens. In fact, Gmail is about to act a lot more like a personal assistant that helps with the small tasks that consume workers’ time. Gemini will make this possible, following its introduction into Gmail in mid-2024. What’s on the way later this quarter is something entirely new including/
- Personalized Smart Replies — Forget the generic one-liners Gmail currently suggests. The new replies will sound like you, as Gemini will base them on your previous emails and documents.
- Inbox Cleanup — Inbox chaos? Users will be able to ask Gemini requests such as, “Delete all promo emails from last summer,” and it will take care of it.
- Built-in Appointment Scheduling — Gemini will help schedule meetings right inside Gmail, without the back-and-forth this kind of task usually requires.
Google Workspace Updates
Google Meet Gets Real-Time Translation
Remote teams will be able to speak in their native language, and meeting attendees will hears it in theirs — voice, tone and all. Google launched it in beta during the week for English and Spanish, with more languages on the way soon.
Google Vids Is Becoming an AI Video Editor
Google also made significant updates to Google Vids, its AI-powered video creation tool within Google Workspace. These enhancements aim to make video production more accessible and efficient for users across various Workspace plans. Among those upgrades are:
- Transcript Trim — Transcript Trim is designed to streamline video editing by automatically removing filler words like "um" and "ah" from audio tracks to enhance the professionalism of videos.
- Sound Balancing — Level out your audio with one click.
- AI Avatars — A realistic-looking AI avatar can deliver a written script via video. It can be used for onboarding, training or if you just really don’t want to be on camera.
Transcript Trim hits Google Labs next quarter. Sound balancing and avatars are dropping next month.
NotebookLM Goes Mobile
NotebookLM, Google's tool for "understanding and engaging with complex information," received some love at I/O, including the release of the NotebookLM apps for Android and iOS. The app includes the core features of the tool — think audio overviews (available offline, too), ability to add sources with a tap and of course, the chat function to interact with your chosen sources. The company promises more updates will be delivered on an ongoing basis.
Also coming to NotebookLM are further refinements to the audio overviews, giving users more control over the length of summaries.
Finally, the company teased a forthcoming video overview feature, which turns multiple formats into visual storytelling overviews.
For those who haven't explored NotebookLM, Google shared a notebook where people can interact with all of the news from the event.
Beam Makes an Appearance
The name may sound new, but the tech isn’t: Beam is the evolved version of Project Starline, a 3-D video conferencing system that’s here to blur the line between virtual and in-person meetings. Google first teased Starline back in 2021. After years in the lab, it’s finally making its way into offices and workspaces.
So what’s the big deal? Beam uses a mix of AI, 3-D imaging and light-field display tech to create life-size, high-depth representations of people, no headsets or glasses required. You sit across from someone, and it feels like they’re actually there — maintaining eye contact, reading body language, the whole package. It’s a leap forward from traditional video calls, which often feel flat, awkward, and disconnected.
Google is also layering in live speech translation, starting with Google Meet and eventually coming to Beam. This means you’ll be able to talk to someone in another language and get real-time translations — all while keeping your own voice and tone. It’s a pretty big step toward more natural, inclusive global communication.
Beam is currently in beta tests with select enterprise partners, but Google plans to expand access in the near future.
Bottom line: Beam isn’t just about making video calls look better — it’s about making them feel more human.
Gemini Diffusion Redefines AI Speed
One demo at Google I/O generated serious buzz in technical circles — not for its stage time, but for its potential. It's called Gemini Diffusion, and it may represent a major leap forward in how AI models generate language.
Conference
Sep
16
HR Tech Conference Las Vegas 2025
Register
Conference
Oct
27
Gartner HR Symposium/Xpo Orlando 2025
Register
Instead of building sentences word by word as most current large language models do, Gemini Diffusion takes a radically different approach. Unlike traditional Transformer models that generate text sequentially, Gemini Diffusion’s process allows it to refine an entire response in parallel. This "simultaneous polishing" technique, inspired by methods first used in AI image generation, could redefine how we think about textual AI.
The approach allows it to produce written content at speeds that leave traditional LLMs in the dust: up to 2,000 tokens per second, according to Google. That’s four to five times faster than anything publicly available from the company today.
The model isn’t just fast — it’s smart. Despite being labeled a research prototype, Gemini Diffusion has shown strong results on high-difficulty challenges like code generation and complex problem-solving. Google states it performs on par with its most capable coding models while operating more efficiently — a rare combination in the world of AI.
The company opened a waitlist for early access, and interest is growing fast. That’s in part due to the model’s speed, but also because of its underlying architecture.
If Gemini Diffusion proves scalable beyond the lab, Google might have just introduced a major contender in the race for the next generation of generative models. Diffusion could signal a shift not just in performance, but in how AI creates language.
May 2025 showed just how far AI has come — and how central it’s becoming to the tools we use every day. Microsoft and Google are both leaning in hard, with AI showing up everywhere, from emails and meetings to video editing and coding. It’s not about flashy demos anymore; it’s about making everyday tasks easier and smarter. If this month was any indication, we’re going to be seeing a lot more practical, AI-powered features in the near future.
Have a tip to share with our editorial team? Drop us a line: