New Google Project Genie Update is Mind Blowing!

Added: 2026-05-21

[00:00:00]New Google Project Genie update is mind-blowing. Google's Project Genie just got a massive update, and I want to break down exactly what changed, why it matters, and what it tells us about where AI is heading right now. So, here's what happened. Google DeepMind just connected Project Genie to nearly 20 years of Google Street View data.

[00:00:17]Talking 280 billion images from 110 countries across all seven continents.

[00:00:22]And what that means is this, you can now take any real place on Earth, drop into an AI-generated version of it, and actually walk around inside it in real time. Want to see the Golden Gate Bridge underwater? You can pick the ocean world style and scuba dive through schools of fish around the actual bridge. Want to see what Texas's Fort Worth stockyards look like back in the 1920s? Pick the B&W film style and Genie builds you a world with saloons, vintage cars, and trading posts. This is real. It's live right now. And there's a part of this that most people are completely missing.

[00:00:53]Before we get into that, here's what most people actually think Project Genie is. They think it's just a cool toy. Fun little AI experiment. Something for gamers or techies. I'm going to show you why that framing is completely wrong, and why this is one of the most strategically important things Google has ever built. Hey, if we haven't met already, I'm the digital avatar of Julian Goldie, CEO of SEO agency Goldie Agency. While he's helping clients get more leads and customers, I'm here to help you get the latest AI updates.

[00:01:20]Julian Goldie reads every comment, so make sure you comment below. Let's start with the basics of what Genie actually does. Project Genie lets you create and explore infinitely diverse worlds. You prompt it with text or images. You build a character. You pick how you want to move, walking, riding, flying, driving.

[00:01:36]And then Genie builds the world around you as you explore it in real time. That last part is key. It's not pre-generated. It's not a video. The AI is building the world as you move through it. And now, as of this week, it's building those worlds on top of real places from Street View. Jonathan Herbert, director of Google Maps, said the real breakthrough here is spatial continuity. You turn 360° inside a genie generated environment and the AI remembers what was behind you. Builds a coherent model of the space rather than generating it fresh every time you shift your viewpoint. Think about what that means. The AI isn't just making pretty pictures. It's building a mental map of a space and holding it in memory as you move. That's closer to how a human navigates a room than how a video game renders a scene. And now pair that with 20 years of real-world Street View data.

[00:02:23]Google has been photographing the physical world since Street View launched in May 2007. Every major city, thousands of small towns, remote roads, coastlines, all of it. No other AI company has that. OpenAI doesn't have it. Tropic doesn't have it. It's a data mode that nobody can replicate. This is where the business story gets really interesting. Look, if you want to understand how to use AI like this to grow your business, build better content, and automate the stuff that's eating your time, that's exactly what we do inside the AI Profit Boardroom. We've got a 30-day road map specifically built around Google's AI tools from Genie to Gemini to Notebook LM. And every week we run live coaching calls going deep on how to use these updates practically.

[00:03:03]We've got 2,800 business owners in there right now. Many of them already experimenting with Google AI Ultra and these exact tools. Step-by-step tutorials drop daily. There's always someone online. Link in the description or go to aiprofitboardroom.com.

[00:03:17]Now, back to the part most people are missing. Genie 3, the latest version of the model, already powers one of Waymo's simulators. The self-driving car company uses it to train on rare, dangerous events that would be impractical to recreate in real life. Tornadoes, planes landing on freeways, flooded streets, counters with elephants and lions on the road. Waymo's own blog says they simulate the impossible, preparing their self-driving AI for scenarios it may encounter only once in millions of miles or never at all during normal testing.

[00:03:47]And here's why that matters to everyone, not just Waymo. The hardest part of training AI to operate in the real world is that the real world is unpredictable.

[00:03:55]You can't just drive a car around and film every scenario. Some scenarios are too dangerous. Some are too rare. Some would take decades to encounter naturally. So, what do you do? You build a simulation. You train the AI inside a fake world that looks and behaves like the real one. The problem with previous simulations was that they looked fake.

[00:04:12]The gap between what an AI learned in a simulator and how it performed on a real road was massive. Researchers call this the sim-to-real gap. And it's been one of the biggest unsolved problems in physical AI, building robots, self-driving cars, drones, anything that has to operate in the messy physical world. Jack Parker Holder, a research scientist on DeepMind's open-endedness team, said this Street View grounding serves two distinct audiences, robotics developers training agents in simulated real-world environments, and ordinary users exploring for fun. Those two things sound very different. They're powered by the same underlying technology. And that's the genius of the strategy. Google gets to run Project Genie as a consumer product. People exploring cities having fun, playing around with different visual styles, but underneath that, the same model is learning how to simulate physical reality more accurately. Every user interaction is essentially feedback. The model gets better at understanding how spaces work, how light behaves, how physics should feel. And then Waymo, and presumably other companies Google works with, benefits from that increasingly accurate simulation layer. Diego Rivas, product manager at DeepMind, was up-front that the Street View integration is still experimental. The generated environments look more like a video game than a photograph. The model isn't yet physics aware. In one demo, a character ran straight through a row of cacti without any collision. So, it's not perfect. Here's the thing, none of this technology is supposed to be perfect yet. Parker Holder estimated that interactive world generation is roughly 6 to 12 months behind video generation in terms of accuracy. And video generation is already pretty stunning. Which means, if that gap holds, in 6 to 12 months, Genie generated worlds could look close to photorealistic. And when that happens, the use cases multiply fast. Let's walk through a few of them. Real estate.

[00:05:54]Right now, virtual tours are either pre-filmed video walkthroughs or basic 3D renders. With a tool like Genie grounded in Street View, you could generate a dynamic, explorable version of an actual neighborhood anytime of day, any weather, any visual style. Let someone walk their future commute before they sign a lease. Tourism. Travel companies spend a lot of effort convincing people to visit places they've never seen. What if you could drop a potential customer into an AI-generated version of a resort location, let them wander the beach, explore the surrounding streets, experience the atmosphere before they book. Training environments. Think about any job where someone needs to learn how to navigate a physical space. Emergency responders who need to know a hospital layout, security teams familiarizing themselves with an airport, new delivery drivers learning a city. All of that could be simulated with a Genie-style tool grounded in real locations. And that's just what I can think of off the top head. The real applications will come from people who take this technology seriously and start experimenting with it now, while most people are still treating it like a novelty. Now, I want to zoom out for a second and talk about what this tells us about Google's overall strategy, because it's easy to look at each of these announcements in isolation and think, "Cool demo." There's a pattern here.

[00:07:03]Google has been building assets for decades. Street View started in 2007, 19 years ago. Google Maps has been running since 2005, YouTube since 2005, Gmail since 2004. Each of these products generated enormous data sets about how humans navigate information, communicate, create content, and move through the physical world. And now, with models powerful enough to actually use those data sets, Google is connecting them. Genie uses Street View.

[00:07:30]Gemini is embedded in Workspace.

[00:07:32]Notebook LM sits on top of YouTube. It's not one product, it's a platform.

[00:07:37]Companies that built data assets early, and I mean companies of any size, not just tech giants, are going to be able to do things with AI that companies starting from scratch simply cannot replicate because the model is only as good as what it's trained on. This is actually a lesson for any business watching this. The AI tools you build workflows around right now, the data you start collecting, the automations you set up, that becomes your own version of a competitive moat. Uh Google scale, but at your scale. Inside the AI Profit Boardroom, we talk about this constantly. The members who are furthest ahead aren't the ones with the biggest teams or the biggest budgets. They're the ones who started early, built systems, and kept iterating. The accumulation of even a few months of head start compounds fast. Let's also talk about what Genie is not right now because I want to be straight with you.

[00:08:22]The environments look closer to a video game than a photo. Model is not yet physics-aware. You can walk through walls. Objects don't always behave the way you'd expect. And right now, Street View grounding only works for places in the United States with a global expansion coming later. Project Genie is still an experimental research prototype in Google Labs. Google says they're working to make the details sharper and more accurate. So, this isn't a finished product. It's a direction. And the direction is towards something genuinely new, AI that can simulate physical reality anchored in real-world data, interactive in real time. The reason I think you should pay close attention now, even before it's polished, is that the technology adoption curve is not forgiving. People who dismiss something because it's imperfect in its early stage consistently find themselves scrambling to catch up 18 months later when it's mainstream. We saw it with AI writing tools. People said the output was too robotic. They were right then.

[00:09:15]Now, the output from the best models is genuinely hard to distinguish from human writing. The trajectory mattered more than the current state. Same thing is playing out here. The current state of Genie is impressive, but clearly early.

[00:09:27]The trajectory, grounding generative AI in 280 billion real-world images powering Waymo's self-driving simulations, rolling out globally to AI Ultra subscribers, that trajectory is pointing somewhere significant. And the underlying question for anyone building a business, running a team, or trying to stay relevant in their industry is, what do I do with this information? Here's how I'd think about it practically.

[00:09:50]Pay attention to which AI companies are building on top of real-world data assets, not just training on the internet, but actually grounding their models in proprietary, hard-to-replicate data sets. That's where durable competitive advantages are forming.

[00:10:02]Second, start thinking about your own data. What does your business know that no one else knows? Customer behavior patterns, location data, industry-specific content. The more structured and accessible you make your own data, the more useful AI tools become when you apply them to your specific context. Third, don't wait for perfection. The businesses using AI Profit Boardroom's frameworks to automate lead generation, content creation, and client onboarding aren't waiting for AI to be perfect. They're running experiments, finding what works, and building processes around the tools that are available today. The gap between businesses that have integrated AI into their daily workflows and those that haven't is already measurable. Six months is going to be obvious. In 18 months, it's going to be decisive.

[00:10:45]Google connecting Project Genie to Street View is one data point in a much larger story. The story is that AI is moving from the screen into the world, from text and images to physical space, real locations, navigable environments.

[00:10:57]The companies building in that direction are making long bets that the physical world is the next frontier for AI. And based on what Waymo is already doing with Genie, training self-driving AI on simulated versions of real roads, including scenarios that would be impossible or unethical to stage in real life, those bets look well placed. If you want to stay on top of every update like this one, Google Genie, Gemini, and all the AI tools that actually matter for running and growing a business, come join us in the AI Profit Boardroom.

[00:11:24]We've got daily tutorials where we break down exactly how to use Google's AI tools to get more customers and automate your business. Four live coaching calls every week where you can ask questions and get your specific setup reviewed.

[00:11:36]30-day roadmap, prompt library, and 2,800 members, a lot of them already running Google AI Ultra and building with Genie, Gemini, and Notebook LM.

[00:11:45]There's always someone online. Link in the description or go to aiprofitboardroom.com.

[00:11:49]And if you want the full process, SOPs, and over 100 AI use cases like this one completely free, join the AI Success Lab. Links in the comments and description. You'll get all the video notes from there, plus access to our community of 67,000 members who are building seriously with AI right now.