WHAT'S INSIDE / 01 Leak sensor The whole robot lives inside a sealed hull. Everything that makes it think, move and come back is in there. Water gets in, it stops answering. This board has one job: see the water coming in. Smallest thing on the robot. The one that decides if we get it back... Off to the fab.

36 seconds of airtime per hour. Per robot. Stigmergy, waggle dance, response thresholds, quorum. Evolution shipped first... medium.com/@robotSmallThing/…

Been quietly building this for a while. I kept running into the same question: how do you prove what an autonomous machine actually did? Full write-up soon.

W2: the trap that catches what the ocean hides. Last week, rotating drum. Conveyor. Dual slots. The contradiction was wide open, no solution in sight. A flap. One moving part. That's it. Geometry that works in one direction and locks in the other. The tentacle pulls through, the flap scrapes the microplastics off on the way up, they drop into the collection chamber. Sealed. Contained. Gone from the ocean for good. No net. No filter. Just a flap that understood the assignment. The trap doesn't negotiate. Whatever enters the chamber stays there. This is part of the nervous system of Small Thing. One of three collection systems on ST-001. The one that reads what hides under the surface. Next week, W3. The signal that fires before the damage does... Eyes open, crew ! 🌊 Raw internal exploded view⬇️

W1: a mechanical contradiction in silicone. Microplastics are hydrophobic. So are our tentacles. That's not an accident, it's one of three collection methods we're building into ST-001. Two hydrophobic surfaces meeting in open water. Chemistry does the work. No energy, no filter, no net. The tentacles trail under the hull in a semi-sealed compartment, open to water, exposed to the ocean. That's the whole point. The problem hits when you want to clean them mid-ocean without going back to base. You pull them through a slot in the hull, a fixed blade scrapes the microplastics off into a sealed collection chamber. Nothing lost, nothing back in the water. Except the stiffness that makes the scraping work is exactly what blocks redeployment. Too stiff to pass back through = great scraper. Too loose = deploys fine, collects nothing.😅 Same material. Same slot. Opposite requirements. You can't have both. Until you change the geometry. Sometimes it takes 3 weeks of iterations to find the best possible compromise. Open water doesn't forgive like dry land does, sailor. Rotating drum? Conveyor? Dual slots? Stay close. See you next week to show how we broke the contradiction. 🧵 Side exploded view representing the contradiction⬇️

5 weeks. No whitepaper survives saltwater. W1 a mechanical contradiction in silicone. W2 the trap that catches what the ocean hides. W3 the nerve that tells you when you're sinking. W4 full throttle. taking the helm. open water. W5 not everything comes back from 100m. Deep end only 🌊

No resin, no regrets.

Over the last few months, we've been working intensely behind the scenes. The focus wasn't on communication. It was on structure. We onboarded several new contributors across engineering, operations and strategy. The team is stronger, clearer in its execution, and aligned on short-term delivery. Over these past months, I've met all kinds of people around this project. Curious minds, deep tech enthusiasts, complete beginners, strategists, and like any project, the moment you start talking about what you're building, you always meet the ones who already know everything about everything (it's a universal law at this point 😅). But here's what I noticed: The people who stick around aren't here because the project sounds interesting. They're here because it gives them something they weren't expecting... purpose. Marketers, advisors, engineers, people with zero technical background, nobody was asked to just do a job. They were asked to care. And when people work with heart instead of obligation, everything hits different. From now on, communication will be much more frequent and consistent. I'm entering full founder mode, meaning regular updates, transparent progress, and direct insight into what's being built. On the product side: V1 is ready. This is not a concept iteration anymore. The first operational version has been finalized and tested internally. We'll keep iterating, v2, v10, as many as it takes; but the foundation is solid, validated, and built for what comes next. Regarding deployment: The City of Marseille continues to support our initiative for local deployment. Discussions and coordination are ongoing to ensure regulatory and operational alignment. This remains a key milestone for us. On the token side: $ST utilities will be presented soon. Clear mechanics. Clear purpose. Built around real-world activity, not speculation. Thank you for your patience over the past weeks. We're not here to move fast and disappear. We're here to build something that operates in the real world. Sea you soon ❤️🌊

721 particles/kg/day. 50,000 microplastics entering your body every single day. And the craziest part? It's not the seafood. It's your vegetables, your grains, foods that absorb plastics from contaminated soil and water. The ocean doesn't just stay in the ocean. It cycles back through rain, irrigation, groundwater. Into the food chain. Into you. Previous estimates said 100-900 particles/day. This study says 50,000. We were off by 50-500x because nobody was measuring the right foods. This is why we're building Small Thing. Intercept plastics at the surface before they fragment smaller, sink deeper, spread further. • With the support of governments and cities (the city of Marseille is already backing us: xitter.cf/the_small_thing/status…), we can deploy and scale our robots in real environments. • With the support of companies like @virtuals_io (for example through $ST), we can raise capital and build real utilities around the project. The real alpha: Big cleanup projects: $10M+ boats, crews, fuel, permits Small Thing: backpack-sized, solar, autonomous They measure success in PR. We measure success in particles/$/hour.

Happy New Year to every single one of you. First and foremost: health. Seriously. I hope 2026 brings you and your loved ones strength, peace of mind, and the energy to chase whatever matters to you. Without health, nothing else we build has meaning. Now let me take you behind the scenes of what's been happening with Small Thing. The past weeks have been intense. Not the "posting flashy renders on Twitter" kind of intense; the real kind. The kind where you're knee-deep in CAD files at 2am, recalculating buoyancy margins, debating hull geometries, and testing what actually works in real conditions. Here's where we're at technically: we've moved to a biomimetic manta ray design, and after multiple iterations, we've locked the wingspan around 40cm. Compact, efficient, purposeful. The propulsion system has evolved significantly, we ditched the dual motor setup after running the numbers. Our current architecture gives us approximately 21 hours of autonomous operation versus only 6.5 hours with the previous configuration. That's not a minor optimization, that's the difference between a functional ocean robot and an expensive floating paperweight. The energy management system has been completely rethought. We've designed for near-infinite operational cycles, Thanks, sun. ☀️ The kind of autonomy that actually makes sense for ocean deployment. Because what's the point of a cleanup robot that needs to come back to shore every few hours? On the collection system; this is where we've put serious R&D. We're not just dragging hardware through the water and hoping for the best. We've engineered chemical selectivity into the core design. The system attracts microplastics while allowing marine life to pass through unharmed. Minimizing bycatch isn't a nice-to-have for us; it's a non-negotiable design principle. What's the point of cleaning the ocean if you're harming it in the process? Now let's talk about the team. Because a project like this doesn't happen with just passion; it takes expertise. Over the past months, we've been quietly building something solid. Engineers, advisors, partners who actually understand what it takes to put hardware in the water. Some names you already know. Others you'll discover very soon. What I can tell you is this: we're not figuring things out as we go. We have people who've done this before, and they're all-in on Small Thing. The internal prototypes are already well beyond the concept stage. We're not theorizing; we're iterating on real hardware. Testing, breaking, improving, repeat. That's the rhythm now. And every cycle gets us closer to what we promised you. I'll be transparent: building robots that actually function in harsh ocean environments isn't something you speedrun. Every shortcut comes back to bite you when saltwater meets electronics. We have clear milestones mapped out, and we're checking them off methodically. Not because we're slow, because we're serious. There are countless projects in this space that ship fast, hype hard, and disappear faster. That's not us. We'd rather build something that genuinely works; something you'll actually see floating in the ocean collecting plastic, something that proves this community built more than just another token. We're weeks away from showing you everything. Not months. Weeks, Something's about to surface. Literally... 2026 is yours as much as ours. We've got an ocean to clean, crew. 🌊

Ahoy flibustiers! Hope you're riding the waves well. We named our AI agents after ocean gods. Here's why. When you're building a swarm of autonomous robots to clean the oceans, you need more than code. You need intelligence. Not one AI, a whole pantheon working together. 🔱 POSEIDON doesn't just check the weather. He reads the ocean like an ancient oracle. Wind speed, wave height, sea temperature. He knows when the waters are calm enough for our robots to hunt microplastics. 🐙 KRAKEN has tentacles everywhere. Every robot in the fleet, he feels them. ST-001 at 94% battery, ready at base. ST-007 returning with a full filter. ST-009 charging. Nothing escapes him. 🔷 TRITON is our analyst. Cold, calculating, precise. He takes POSEIDON's weather data, KRAKEN's fleet status, and outputs one thing: mission confidence. 82%? GO. 45%? STANDBY. ⚓ NEPTUNE commands. He listens to all three, then issues the directive. "Deploy ST-002 and ST-004 to collection zone. ST-007 continue RTB." But this? This is just what you see. These 4 agents are the visible layer. Behind them, a massive data pipeline feeds everything back into our core models. Every sensor reading, every collection run, every environmental variable, continuously recalibrating detection algorithms, adjusting filter sensitivity, optimizing patrol routes in real-time. The more the swarm operates, the smarter it gets. This is multi-agent AI coordination. This is how Small Thing cleans the oceans. One swarm. One mind. Many gods. Something big is coming. Stay tuned. 🌊

GM ocean pirates 🏴‍☠️⚓️ Dropping our v3 filtration system refactor The mesh paradox: You'd think "just use the finest mesh possible" right? Wrong. Ultra-fine (10µm): Catches 96% of microplastics but chokes itself to death in 3-4 hours. Your robot spends more time docked for cleaning than actually filtering. Medium (100µm): Only 72% capture rate but runs 16+ hours straight. Here's the robotics math that matters: - Ultra-fine bot: 3hrs × high efficiency = dead battery, constant maintenance - Medium bot: 16hrs × decent efficiency = way more ocean cleaned per charge Battery cycles are expensive. Dock time kills ROI. A clogged filter at 2am doesn't care about your 96% efficiency rating. Built this sim to iterate faster on our filtration redesign. After countless hours testing different configurations, we needed something concrete to validate our approach before committing to the hardware changes. v3 filtration refactor complete. Multi-stage setup hits different when you understand why. ❤️🌊

GM builders 🛠 The RWA market will hit $50B by end of 2025. But here's what we noticed: 99% of "RWA" is: → Tokenized bonds → Digital twins → Virtual ownership All critical. All necessary. But where's the OTHER side? Where's the instant bridge between blockchain and REAL physical action? That's the missing piece of DePIN we've been obsessing over. Human sends tx on @base → Physical robot responds in seconds → No oracles, no delays, no middlemen 30 $ST → Motor spins in 4 seconds. (Check out my latest post) And this is just the surface. We're already deep into full robotic systems. More to come soon. Here's our 24-h challenge: 10000 $ST bounty to anyone showing: ✅ Blockchain tx ✅ Physical robot responds in <5 sec ✅ On Base network ✅ No oracles, no APIs, no delays ✅ Open proof Not because we think nobody can. But because we want to see who else is building this bridge. Show us. Learn together. The infrastructure is ready. The market is exploding. Time to build the missing bridge. This is @the_small_thing 🌊 Bridging the gap between digital and physical, one transaction at a time.

Ahoy, sea rovers! 🏴‍☠️🌊 A story about understanding Real World Assets... Years ago, I worked with a mentor on TV shows. He was tough, often yelling: "You damn geeks, always thinking digital! What about the ANALOG? The PHYSICAL?" He was rough around the edges, but he was also literally serving bulletproof vests to his team. We all loved him for that. At the time, we saw him as "old school", always insisting we couldn't remove physical senses from interactive systems. Years later, I finally understood: this man had figured out RWA long before it became a buzzword. He knew you can't have true interaction without bridging digital AND physical worlds. I'm sharing this so you understand our team's vision of what RWA really is. This isn't just another acronym from a marketing meeting, it's in our DNA at Small Things. Today with my team, we're presenting an R&D video in progress for our new blockchain communication module with the robotic fleet. That's what Real World Assets actually means, From digital to analog. From blockchain to metal. 🤖⚡️ Welcome to The Small Things.

Mesh network stress test running in the Mediterranean 🌊 Testing autonomous routing optimization across 10 nodes with adaptive duty-cycle balancing. The fleet maintains GSM/WiFi-grade resilience while respecting regional duty-cycle constraints critical for long-range oceanic deployments. Key focus: battery performance vs. radio duty-cycle trade-offs. Each node dynamically adjusts transmission power and hop-count routing based on real-time mesh topology. Current run: 25 messages routed through up to 5-hop paths, with autonomous low-battery escalation protocols kicking in when needed. The goal? Keep the mesh alive for weeks, not hours. Deep ocean operations demand extreme power efficiency every milliwatt counts when you're 50km from shore. Real-time telemetry, multi-hop routing, and self-healing topology. This is the unglamorous engineering work that makes ocean cleanup actually scalable. I know many of you want frequent updates and I appreciate the enthusiasm. But as you can see, this is a robotics project with extreme constraints. We can't push daily updates while maintaining quality and safety standards. That said, a proper logbook will be maintained throughout deployment. Engineering takes time. We're building for reliability, not hype. And thank you for all your messages. I genuinely enjoy exchanging with you all. Your support keeps the project moving forward.