9 AI Breakthroughs That Shocked Us at TechCrunch Disrupt

TechCrunch Disrupt just hit San Francisco with its 20th anniversary. And for those who don't know what TechCrunch Disrupt is, we're completely so mo low. And we're molo so lo moso, bro. It's an annual gathering of startups, wild technologies and a few cornerstone tech firms, each bringing their a game to woo potential investors and achieve glory by facing off against their competitors on the battlefield. And I don't just mean a figurative battlefield, I mean a literal one. TechCrunch disrupts Startup Battlefield is a live competition on stage where founders pitch their ideas to some of Silicon Valley's top investors for a shot at 100 grand in equity, free funding, and more importantly, the kind of exposure that can launch a startup overnight. It's one of those events where you'll see the next wave of innovation before anyone else. The kind of place where companies like Dropbox, Fitbit and Trello all got their start. And this year, my good friends over at JetBrains who sponsored this video are brought me and my producer Dave out there. Not just to show off the impressive work they're doing for developers across the entire software life cycle, but to give me a chance to walk the exhibition halls in search of the coolest new startups bringing wild new tech to the world of AI. And there was a ton of wild technology on display. But in this video, I'm going to narrow it down to the nine that really grabbed our attention. All right, so without further ado, here's what I found. First up, we visited the booth for ShinStar. I mean, it was hard not to stop when he saw the bright orange suit and the bribe to play their game in exchange for a free meal. But these guys are pretty cool. They're not just automating kitchens, they're actually putting entire robotic restaurants into a truck. While we were at TechCrunch, my producer Dave and I actually got to see their also Korean barbecue food truck in action. Inside is a fully robotic kitchen that cooks 24, 7 and requires less space than like a parking spot. As Tord from shinstar explained, we have a fully robotic Korean barbecue restaurant that can cook anywhere, any time. Redefining what dining even means. The whole system is AI powered and electric. It uses predictive models that measure things like humidity to make sure every dish is cooked exactly to spec.

So we have a world renowned chef from Korea who has teach them how to cook like he does. So if you want to cook something else, it will replicate your meals to your liking.

This tech actually cuts restaurant labor by about 80% while speeding things up. And as Tord put it, chefs can spend more time creating new recipes instead of stir frying all day. And we got to film the whole process, the arms grabbing, the wok spinning, the food being dumped into our bowls. It kind of felt like we were in a sci fi movie. And yes, we did actually get to try the food as well. I had the beef, Dave had the chicken, and both were surprisingly good. But what really caught my attention is where they're actually going next. With this technology, shinstar is building micro restaurants for places like airports to starting with LAX in San Francisco, where you'll be able to order food from your phone and pick it up straight from a window. No line, no wait, just hot food ready as you walk by. You can actually be in the security line order and your meals waiting when you get to the other side, Tord told us. No queue. Shinstar's vision is actually pretty simple. Make restaurant quality meals accessible anywhere 247 without the limits of space, labor, or time. So the next time you're craving Korean barbecue and see a futuristic looking food truck rolling up, don't be surprised if a robot chef is behind the wheel. Next up, we visited the booth for a sleep AI, and this one really grabbed our attention because it promises something that sounds almost impossible. They promise full sleep tracking and analysis without wearing anything like no rings, no watches, no sensors strapped to your head. You just need your phone. And we actually talked with one of their team members at TechCrunch and they explained how it works. So when you go to sleep, you just open the asleep app, set your alarm, and place your phone nearby. The app uses your phone's microphone to listen to your breathing throughout the night. Their AI then analyzes those patterns to figure out which stage of sleep you're in. Light, deep, or REM sleep. And in the morning, it gives you a full breakdown of your night with clear explanations and insights right on your phone. The team actually explained to us that while everyone's breathing sound is unique, the pattern of breathing changes the same way across sleep stages. That means their AI can accurately detect your sleep cycles and just from the sound of your breathing without needing any wearable sensors. And their technology's already being used by major companies like Samsung and lg. They've also built one of the world's largest sleep data sets to train their AI models on. They're basically building the foundation for contactless sleep health, something that could eventually be baked right into phones, TVs, or even smart speakers. I personally took A peek at the app, and it's actually quite detailed for something that only uses sound. Plus, it's completely free to try right now. Eventually, they want to move to a subscription model, but as of right now, it's totally free. And if this kind of tech takes off, it could make sleep tracking as effortless as just making sure your phone is on your nightstand by your bed. Something that's actually pretty common for most of us anyway. Next up, we visited the booth for General Neuro, a company that claims to literally tickle your brain to help you learn faster. Yeah, that's right. Their headset uses mild electrical stimulation to boost neuroplasticity, so your neurons fire more easily while you study. In theory, it's like a pre workout, but for your brain. They just launched on Kickstarter, and the founders told us it's designed to lower the threshold for neuron activity in the part of your brain associated with language learning. Basically, you strap this thing on, open their AI powered app, and it zaps your skull while you practice Spanish or Italian. Because nothing says immersion like a gentle jolt to the cortex. Now, I wasn't about to risk frying my synapses for the sake of a YouTube video, so I made my producer Dave do it instead. The rep literally goes, you feel a little tingle. And sure enough, Dave starts wincing and says, yeah, definitely a tingle. Then he asked if it should be harder. The guy actually says, we can make it harder. So, yeah, things escalated quickly.

Apparently the tingling means it's working. Personally, I think it means your neurons are filing a complaint. But to their credit, they were super nice and genuinely into the neuroscience behind it. One of the co founders is even finishing his PhD at NYU. So it's not just some random brain zapper from like a mall kiosk or something. Still, I think I'll stick to Duolingo for now. Now, Dave may have downloaded a few Spanish words directly into his frontal lobe. He can't remember his middle name anymore, but who really needs that info anyway? Either way, he survived, so seems like good science. Then it was time to head over to the Jetbrains booth to learn about everything they were up to. This booth, which was by far the most impressive at the event, had everything from hands on demos with their products to a barista pumping out scorching hot coffee. Their team welcomed us with open arms and walked us through everything the booth and company at large had to offer. First, I got to test their Juni coding agent, which was built into an awesome gamified startup simulation. Now Jooni is JetBrains new coding agent that unlike a simple chat assistant, can understand your project, create a plan and carry out coding tasks such as refactoring or adding tests. You stay in control, reviewing its plans and approving any changes before they're made. The goal isn't to replace developers, but to actually help with routine or time consuming work so coders can focus on the parts of coding that really need human attention. Juni represents JetBrains next step towards practical collaborative AI in everyday software development. And as far as that simulation game is concerned, I did all right. Then I had the chance to sit down with arun Gupta, JetBrains VP of Developer Experience. When we talked, he really drilled down on the customers they're aiming to serve.

Yeah, JetBrains is a global software developer company. We have been into existence for 25 plus years now and I like the.

Way JetBrains refuses to take their focus off the development community.

We build tools across a wide range of programming languages that are meant for team productivity like YouTrack and TeamCity which is your CI CD pipeline. So really integrated across your entire software.

Development life cycle with an emphasis on removing the boring mindless tasks.

We believe that, you know, AI is definitely instrumental technology of the era, but our North Star is productivity, making them super productive but then also giving them the joy while they are coding.

What was also interesting and the focus of Arun's appearance on stage was the way JetBrains challenges some of the popular ideas surrounding vibe coding and whether or not AI is ready for prime time in that realm.

Today the spectrum is early on. It gets you started, gets you an MVP even to a non coder for that sake, what is your next 25.2 miles are going to look like? How are you going to package it up? Is wipe code ready for that? Not right now. Now we have also seen where companies are talking about oh, this is all going to work out agentically. Well, agentically is a bit magical right now, right. I cannot really live on that magic for enterprise software.

And during that same stage talk, Arun made a big announcement as well.

What we want to introduce is a brand new benchmarking platform and we call this as a Developer Productivity AI Arena.

JetBrains was launching their brand new Developer Productivity AI arena, or the DPAI arena for short. The DPAI arena is basically a public testing ground that measures how good different AI coding tools and actually are at real developer tasks. It's like a scoreboard for coding AIs showing which ones truly make developers faster and which ones Just pretend to and any company, research lab or individual can contribute, test their AI agents and see transparent results. Until now, every company has claimed their AI boosts productivity, but there's never been a trusted way to measure it. This actually changes that. And quite Honestly, you know, JetBrains sponsorship aside, it is a benchmark that I've really been wanting to see made myself. Then I had the chance to speak with Denis Sheriev, the AI group product manager at JetBrains, who was really proud of the fact that even with the implications of AI advancements, JetBrains is never leaving coders behind.

We are here to win your trust and we are here to build something that win, you know, you as a user of our agentive system.

Overall, it was pretty easy to see how JetBrains has been in business for over 20 years, serving coders and developers and enabling them to do their best work. So thanks again to JetBrains for bringing us out to TechCrunch Disrupt this year and for sponsoring this portion of today's video. Now, I don't even know what this is because there was never actually a person manning their booth, but I like the name.

Next up, we stopped by the booth for Tensorfield Agriculture and we never actually got the chance to talk with anyone there either. They were pretty much busy the whole time because but their tech was too interesting. Not to mention, Tensorfield is actually tackling one of farming's biggest headaches, weeds. Instead of spraying chemicals or sending out huge crews of hand pickers, their robots use AI powered vision to spot weeds and then blast them with superheated vegetable oil. Basically, it deep fries the weeds on the spot while leaving the crops untouched. Now, it sounds wild, but it's incredibly precise. About a quarter inch accuracy and and it can cover fields at around a mile per hour. One robot with an operator can reportedly replace an entire hand weeding crew, cutting costs and removing the need for herbicides altogether. It's the kind of tech that feels like a glimpse into the future of agriculture. It's smarter, cleaner, and way more efficient. And even without an interview, this one definitely earned a spot on our list of the coolest things we saw at TechCrunch Disrupt. Next up, we went from a company using AI to spot and kill weeds to one using it to help everything else grow. At the booth for instacrops, we met a team that's basically building a digital agronomist in your pocket. Their platform connects to soil sensors across a farm and uses AI to figure out the perfect time and amount to irrigate and the result? Farmers can use about 30% less water while increasing their profits by around 12%. It's a crop whisperer that never sleeps and lives in your pocket. One that's constantly analyzing your soil data and saying, hey, maybe your blueberries are getting a little thirsty. And if a farm doesn't already have the sensors, instacrops itself will provide plug and play hardware that that connects straight into their platform. Their AI is fine tuned specifically for different crops, so grapes, cherries, apples, avocados, you name it. And instead of adding more guesswork, it helps farmers make smarter, more sustainable decisions. Now, it's not the flashiest AI at the show, but it's one of the most impactful. Less water waste, healthier crops, and maybe, just maybe, cheaper guacamole. Next up, we visited the booth for Unibiotic, a biotech startup that's doing something surprisingly cool with shrimp shells. Yeah, you heard that right. These guys are turning shrimp waste into a magical powder that can help farmers grow more food using way fewer chemicals. Unibio works with a natural polymer found in shrimp shells, which they've engineered at the nanoscale to help plants absorb pesticides and nutrients much more efficiently. Travis from the team told us that a single pinch of this powder, literally enough to fit just in the palm of your hand, can cover about 20 tennis courts worth of crops. That's because it interacts with the plant's own immune system, helping it take in what it needs and waste less. Now, did you know that around 80% of pesticides sprayed on plants normally goes to waste, either washing off or seeping into the soil. Unibios powder helps the plant actually use what's sprayed on it, cutting chemical use by as much as 50% or while still boosting yields. So it's not just better for farmers, it's better for the plant, it's better for the planet and for anyone who eats food, which is you and, well, me too, most days. The company started in Argentina with four women scientists who wanted to upcycle shrimp waste into something useful. And now they're working with major agrochemical companies to integrate their turbocharged powder directly into products. Now, this one doesn't actually use AI, but Travis joked that while everyone here is showing off artificial intelligence, Unibios focused on reducing active ingredients, or as he called it, the bad AI.

We're probably one of the few people who are not doing artificial intelligence, but what they call in the agrochemical industry their special sauce. The thing that makes these pesticides work, they call Them active ingredients, or AIs for short. So we are doing that.

Now. As you've probably noticed by now, I really gravitated towards products that feel like they have potential to make a real difference for people and for the world. And that's why this magical shrimp powder made my list of most interesting things that we saw it disrupt. Next, we visited the booth for Gargoyle Systems, a company building what they call air traffic control for drones. Now, this one grabbed our attention because their approach doesn't just make drone delivery safer, it's actually paying everyday people to help build the network that makes it possible.

Gargoyle Systems are building air traffic control, but for drones, let's say an airport, for example, they can only put things around their own perimeter, but they need situational awareness beyond their perimeter. Same with a stadium, same with a prison.

Right?

So if you have a gargoyle and they're subscribing, we're going to share that revenue with you for having that on your roof.

All right, so here's how it works. Imagine you've got a gargoyle node, which is a little small box, and you stick it on your roof. Once it's online, it becomes part of a distributed system that detects drones flying overhead. Airports, stadiums, even emergency services need this kind of situational awareness beyond their own perimeter. So they subscribe to the network for real time drone data. And if you have a node and it contributes to that coverage, you actually get a little cut of the revenue. So basically, you're helping keep the sky safe and getting paid for it. If you know much about blockchain, it's kind of like helium, but for drones, it's a community powered sensor grid that rewards you in crypto for supporting their infrastructure. And what makes it even cooler is how polkadot powers the whole thing behind the scenes. Because it's built on the polkadot blockchain, Gargoyle can handle instant automatic payments to node owners without a middleman. So no banks or card fees, just real time rewards flowing into your account. Their first deployment is already happening at the University of Tulsa. But the bigger vision is massive. A global mesh network of drone detection nodes that can enable safer delivery systems, smarter emergency response, and even regulated drone zones for cities. So in a world where drones are delivering everything from defibrillators to burritos, Gargoyle Systems could become that invisible layer that keeps the airspace organized. And if you've got a roof, you could literally be part of it and maybe even get paid for it. And finally, we stopped by the Booth for Glide, the company that actually won TechCrunch Disrupt this year. And honestly, after seeing it in action, it's pretty easy to see why Glide built something that feels straight out of a SCI F movie. A truck that can drive up to the train tracks and in under 90 seconds, transform into a train. Yeah, you heard that right. A vehicle that literally becomes a train.

It has rail gears here, so it goes on the road and on the lane.

I just saw in that video going to the rail. That was really cool.

Yeah.

And it does it under 90 seconds. So we can, like, make the transition under 90 seconds, as a matter of fact.

And that's when it actually hit me. This thing could completely change how freight moves around the world. The vehicle, called Raiden, can carry up to four £40,000 of cargo. It's autonomous, electric, and can lift and move containers all on its own. Their patented hook lift system lets it pick up a container, set it back down, and then seamlessly roll onto the rails when it's time to go long distance. And the genius behind it all is their proprietary AI system, Ezra 16. It pulls data from traffic signals, rail networks, and even companies like BNSF and Union Pacific to figure out the most efficient route for every trip. As their team put it, they're trying to take some load off the freeways and put it back on the rails. Right now, the US only uses about 30% of its rail capacity, and Glide wants to push that closer to 80 or even 90%. That means, like, fewer semi trucks clogging the highways, less pollution, and a logistics system that can literally switch modes of transportation on the fly. It's hard to overstate how big this potentially is. And the business model is just as smart. They're not selling the vehicles, they're offering mobility as a service. They're charging around 25k per month, or roughly 300 grand a year, plus a small upcharge per container. It's the kind of moonshot idea that feels obvious in hindsight. Combine road and rail, automate it, and optimize it with AI. Glide managed to pull that off, and it's no wonder they walked away with the top prize at TechCrunch Disrupt. This one truly feels like a massive game changer to me. If they're able to pull it off at scale, from shrimp based nanotechnology to food that cooks itself and everything in between, like the headgear that fried Dave's brain, we got to see some pretty interesting companies and technologies. And I, for one, am pretty excited about the idea of having automated cooking robots in the airport and sleep trackers that don't require me to wear anything extra to bed. Those represent some pretty cool quality of life upgrades that AI can facilitate. However, I'm even more excited by ideas and that feel like they truly have the potential to change the world. Things like improving farming so that food is healthier and more efficient to produce because it means more people get better access to high quality foods and shipping vehicles that can convert from a truck to a train in 90 seconds and that could leverage our underutilized rail system. Well, that means more efficient shipping, lower pollution and less congested roadways for everyone. AI has finally gotten to a place where where we're seeing real world tangible benefits that are actually improving humanity. And that's why despite all the negatives that come with AI, and don't get me wrong, there are a lot, I'm still insanely excited and optimistic about the future and the AI technology that's helping us build it. Dave and I had an awesome time being surrounded by impressive entrepreneurs, inventors and investors, and we can't wait to see which of these new companies are able to have a significant impact on our everyday lives. So thank you so much again to Jetbrains we for showing us around and making it a memorable experience. Make sure you click the link in the description to learn more about everything they do and to try Juni for yourself. And as always, if you want to see more videos like this in your YouTube feed, make sure you give this video a like and click that subscribe button and I promise to keep you super looped in on all the latest AI news, tools, tutorials. You'll be completely tapped in. I do my best to stay on top of all of it for you so that you never miss what's going on in the world of AI. And thank you so much for hanging out with me today. Nerding out with me on this video TechCrunch Disrupt was a blast and hopefully I'll see you in the next one. Thanks so much. Bye bye. Thank you so much for nerding out with me today. If you like videos like this, make sure to give it a thumbs up and subscribe to this channel. I'll make sure more videos like this show up in your YouTube feed. And if you haven't already, check out Future Tools IO where I share all the coolest AI tools and all the latest AI news. And there's an awesome free newsletter. Thanks again. Really appreciate you. See you in the next one.