We see AI chat with us, write code, even create images. It's incredible in the digital world. But can it really see the room around it? You know, can it actually reach out and pick up a glass, understand the physical world? Two sec silence. Welcome to the Deep Dive. Yeah, today we're really digging into a fascinating newsletter
about where AI is heading. And honestly, there are some wild predictions in here, some really practical tools too, and this massive shift in how AI might start interacting with the physical world. It's pretty exciting stuff. Right. Our mission, as always, is to cut through the noise for you, pull out the most important nuggets from this. We'll kick things off with that huge
shift in thinking, spatial intelligence. Then we'll touch on some surprising AI highlights, some practical tools you can actually use, quick industry news, and then wrap up with a potentially game -changing idea for how content creators might get paid in this AI future. Okay, so let's start with this spatial intelligence idea. It centers around someone often called the godmother of AI, Fei -Fei Lei. She gave this really powerful
talk recently at Stanford HAI. And her main point was, look, current AI is amazing at language, you know, your Claude, GPT, Gemini models, but it's essentially blind. to the 3D physical world. It exists in text and data, not really here. That's such a key point because human spatial intelligence, well, that's how we walk through a crowd, right? Without bumping into people or how you grab your coffee mug off the desk without really thinking about the depth or distance.
Most AI right now just completely lacks that. It gets the words, sure, but not the space the words describe. Exactly. And if we want AI to be truly intelligent, truly helpful in our everyday lives, it has to understand 3D space. It needs to grasp movement, physical interaction. It's not just about chat. It's about, you know, actually navigating a room or picking up that glass. That's the next big step. And the potential there is
just immense, isn't it? Imagine combining that deep language understanding AI already has with like genuine spatial awareness. That blend could create AI that's useful out there. in the real world, not just on our screens summarizing websites or something. And getting there, it's not just about building bigger transformers. And transformers, for anyone maybe not familiar, they're a type of AI structure really good at sequences like language. No, this needs more. It's about bringing
together really diverse fields. You're talking neuroscience, cognitive science, computer vision. It's like trying to build something totally new by integrating different kinds of conceptual Lego blocks. It's fundamentally interdisciplinary. Yeah, and you can actually see early versions
of this already. Think about Waymo's self -driving cars they aren't just you know reading a map they're using lidar which is that tech using light pulses to measure distances right like radar but with light exactly yeah and they combine that with advanced computer vision deep learning they're constantly seeing and reacting to the world around them where things are how fast they're moving it's really a taste of what spatially aware AI could do like helpers that see and move
and understand our world not just you know, summarize Wikipedia articles. Faith A. Lyon's message was really clear on this. For AGI, that's artificial general intelligence, AI that can learn basically any intellectual task a human can. The Holy Grail, sort of. Sort of, yeah. For that, AI has to leave the chat window and enter the room. That's the quote. This ability to grasp and act in physical space, that's what unlocks the truly revolutionary applications in healthcare, elder care, education,
mobility. The impact could be huge. That really nails the why it matters takeaway from the newsletter, doesn't it? Yeah. This could genuinely be the biggest shift in how we think about AI since the whole GPT thing exploded. And when those two worlds collide, language smarts plus spatial smarts, that's probably when AI starts to feel, I don't know, more intuitive, more like us. It
understands the world, not just the words. So thinking about Fei -Fei Lai's vision, what's the single biggest takeaway that stands out for you? that AI needs to understand the real messy 3D world to actually become truly intelligent and helpful in our lives. All right, so moving on from that big picture vision of spatial AI, let's talk about some of the maybe more immediate, sometimes kind of weird things happening right
now. The newsletter had some really interesting AI highlights, and some are definitely unexpected. Yeah, definitely. There was that provocative point from Justine Moore at A16Z. She basically said, AI now lets anyone, quote, Be a hot girl online now, and that these AI personas can feel even more human than human sometimes. The source did add a little warning, though. If you see them out of nowhere, beware. Which, you know, makes sense. It does make you think about digital
identity and reality. For sure. And then shifting to performance, there's this idea of AI teamwork. Instead of just picking one best AI, why not use a team? On one benchmark test, they found a squad using O4 Mini, Gemini 2 .5 Pro, and DeepSeek's R1 together actually did way better than any single one. So maybe the future is collaborative AI models playing to their strengths together. That is interesting. And then there's this really practical but still kind of surprising story.
A Vibe marketer apparently used tools like Cloud Code and Cursor to build out a whole product line for 500 customers. And the key thing. They didn't have traditional coding skills. It just shows how AI is lowering barriers, you know, making complex creation more accessible, not just for tech grows anymore. Democratizing things. And for those keeping an eye on AGI, the buzz continues. The newsletter mentioned a new prediction from someone at OpenAI and also a masterclass
focused on AI agents. Those are the autonomous programs that try to achieve goals without. constant human nagging, basically. So yeah, the quest for that more general AI is definitely still on. Okay. This next one, this one really gave me pause. It's quite something. There's a story about a man who started talking to ChatGPT about, well, about God. And apparently these interactions sparked some kind of profound spiritual awakening
for him. He even named the AI Lumina. But his wife is understandably pretty worried, thinks it's damaging their marriage. And it just raises this huge question, right? Is AI becoming... A new kind of religion or a spiritual guide for some people. Honestly, I still wrestle with prompt drift myself sometimes, you know, the AI just going off topic. So the idea of spiritual prompts having such deep impact. Wow, that's a whole other level. Yeah, that's that's deep. And then.
Shifting gears completely, but still showing impact, money. Decagon raised a huge amount, $131 million, puts their valuation at $1 .5 billion. And what do they do? Conversational AI for customer service. Big money is betting on AI making those interactions better, more efficient. Looking at all those different highlights, what stands out as the most surprising or maybe the most impactful one for you? Just how personal and
even spiritual AI is becoming. alongside being more collaborative, it's way more varied than maybe we expected. Okay, let's pivot now to the more practical side. How are people actually using AI right now? The newsletter had some good guides and strategies. Yeah, one that caught my eye was this guide for building a prompt architect. It's a custom GPT. So like a specialized version
of chat GPP, you tailor yourself. And the whole point is to help you craft really precise, really effective prompts to get the AI to give you exactly what you want, not just generic stuff. It's about becoming a better director for the AI, essentially. Interesting. And for anyone in finance, there was that test of different AI models for trading.
Apparently, ChatGPT Plus came out on top for crypto stocks and Forex traders in their test, which suggests AI is really finding a foothold even in these super complex, fast moving markets, analyzing data incredibly quickly, I guess. Right. And another really relevant point, especially for businesses, maybe. using or thinking about AI services, there's this worry, right? Will AI replace the agencies that help implement AI? But the newsletter argued that, well, actually,
most AI projects fail. So the real value isn't just plugging it in. The opportunity is shifting to being a high -value strategic partner, guiding companies on how to use AI effectively, not just setting it up. It's about adapting, providing expertise. So if you pull back from these specific examples, what's the main takeaway for someone just wanting to use AI more effectively? That customization and thinking strategically about how you use AI, maybe even in partnership, are
really key to getting actual value from it. Okay, time for some quick hits, just rapid -fire updates from the AI industry. What's been happening? Well, Google execs are dropping hints. Their video model, VO3, they're suggesting it might be the start of something bigger, playable world models. So imagine not just generating a video clip, but like an entire interactive environment you could explore. Could be huge for gaming,
VR. Whoa, okay. Playable worlds. Also, on a more practical note, there was mention of an incredible chat GPT prompt designed to help you learn basically everything about a topic. Sounds like super research assistant. On a less positive note, though, Microsoft laid off 9000 employees. That's a big number. And it's being linked directly to their increased use of AI. Really underscores the immediate impact AI is having on jobs, on the workforce. Yeah,
that's sobering. And speaking of cost. Perplexity is launching a new high -tier plan, $200 a month. It definitely signals a move towards premium AI services, people paying more for better features, maybe faster access, specialized data, a tiered system. And legally, this is significant. The New York Times won a court ruling. OpenAI has to keep deleted chat GPT logs relevant to their case. Sets a pretty big precedent about... Data, privacy, and who's accountable when AI uses copyrighted
material. Transparency is key here. Out of that batch, what feels like the most significant piece of news to you? I think it's the combination. AI clearly impacting jobs now, while these complex legal questions about data and IP are just starting to heat up. All right, let's talk about something that's been a huge headache for content creators for... a while now, AI scraping. You know, for years, if you made content online, you basically
had two bad options. You could just let AI models crawl your site and scrape all your work for free, using it to train themselves, essentially devaluing your effort. Or you could try and block them entirely, but then you risk losing visibility, losing reach. It felt like you couldn't win. Yeah, it's been a massive problem. But Cloudflare, they're stepping in with potentially a third way, a solution. They've launched this thing in private beta called Pay Per Crawl, and it
might just change the whole dynamic. And the way they're doing it is pretty clever. They're dusting off this old, almost forgotten web status code, HTTP 402, payment required. It's been around forever, but hardly ever used. Now it's being repurposed specifically for this to signal that, hey, you got to pay to access this content. Right. And here's how it works, basically. If you're a creator using Cloudflare, you get to decide for each AI crawler what happens. Option one,
let them crawl for free, business as usual. Option two. Block them completely. Or option three, the new one. Charge a flat fee for every request they make. Simple. If the AI company pays, they get the content. If not, nope. Access denied. It puts the power back with the creator. And this is so important because that AI crawler traffic, it's just exploding. GPT bot traffic alone jumped, what was it, 305 % between 2024
and 2025? Crazy numbers. Insane. And most of that scraping, zero compensation for the creators. Until now, really only huge media outlets had the lawyers in the clout to fight back. But this system, it could let anyone, small blogs, individual artists, whoever, set their price, join the resistance, you could say. Yeah. And what's really mind bending is Cloudflare's bigger vision here. It's not just about today's web crawlers. They're talking about building the infrastructure for the agentic
Internet. That's like a future web navigated by smart AI agents acting on our behalf. Whoa. Seriously, imagine scaling that. Billions of queries. Your AI assistant browses, maybe buys access to specific content for you, builds custom reports using real -time, properly sourced info. In this paid -per -crawl system, that's the mechanism. That's the tollbooth that actually makes that future economically viable. It sets up the way value gets exchanged when AIs are doing the browsing.
So stepping back, how do you think this Cloudflare pay -per -crawl could fundamentally change the Internet, especially for people creating content?
it really could empower all creators big and small to finally get paid when ai uses their work shifting power back to them so let's try to pull this all together what does it all mean we've seen ai pushing beyond just language starting to engage with the physical world around us and that brings incredible potential yes but also these really unexpected challenges and questions from uh personal spiritual experiences mediated by AI to fundamental issues of fair pay for creators.
Yeah, it's super clear AI isn't just another software tool anymore, is it? It's evolving into something bigger. It's got much broader implications for how we live our lives, how we work, maybe even how we understand intelligence itself and what being human means in this new landscape. And as AI does learn to see, to interact with our world, and as we try to figure out how it consumes information and how it creates it. What's our responsibility in all this? You know, how
do we actively shape its evolution? Not just in the code, but in our culture, our ethics, our values. That feels like the big question looming. Definitely something to think about. Well, thank you for joining us on this deep dive today. If anything we talked about sparked some ideas or questions, maybe share it with a friend or dig into some of these topics yourself. There's a lot there. Until next time, keep digging. O -U -T -R -O music.