Tag Archives: Artificial Intelligence

Beni Is What Sony Aibo Grew Up to Be – A $549 Camera Robot That Captures Life While You Live It

Posted on by

Every camera product ever made has one thing in common: it needs you to operate it. A gimbal needs a hand to hold it. A drone needs a pilot with eyes on a controller. An action cam mounts somewhere on your body, which means it films from your body, not at it. Even the smartest AI tracking stand on the market still requires you to set it up, point it at where you’ll roughly be, and go do something interesting within its field of view. All of this adds up to a life spent maneuvering cameras instead of being in front of them, and the shots that matter most, the ones where you’re actually doing the thing, are consistently the ones you’re missing from.

Beni, from Palo Alto startup Mondo Robotics, is the robot that plays camera crew. It rides on four wheels, films in 4K HDR with 3-axis stabilization, hits 17.9 mph, and clears 10-inch obstacles mid-chase, getting itself back upright after a fall without any help from you. The team includes engineers from DJI and Tesla’s Optimus program, which explains how a 3.86-pound consumer robot ends up with hardware like that. It also has color-changing expressive eyes, a swappable hat, a name, and a personality that comes through in little nods and head shakes. That design language places Beni in a tradition Sony’s Aibo started in 1999, building robots people actually want to be around. The engineering is serious, the warmth is deliberate, and the combination puts Beni in territory most robotics products actively avoid.

Designer: Mondo Robotics

Click here to Buy Now: $549 $799 (31% off). Hurry, only 39/230 left! Raised over $2.8 million.

The choice to give a camera robot a face, a name, and a wardrobe is a design statement, not a marketing gimmick. Beni’s eyes change color and express emotions, it nods and shakes its head, and the whole body is customizable through swappable hats, ear attachments, and protection plates, in whatever combination suits you. The character design sits somewhere between a cartoon character and a precision instrument, which sounds like a contradiction and somehow isn’t. Mondo Robotics understood that emotional connection determines how often a device actually leaves the house. A robot you find charming is one you take outside. One that reads like a spec sheet stays in the drawer.

Beni measures 8.5 by 7.1 by 7.1 inches and weighs 3.86 pounds, light enough to throw in a bag. It hits 17.9 mph (28.8 kph) to keep pace with skateboarders and runners, clears up to 10-inch obstacles to handle curbs and stairs, dodges obstacles with onboard sensing, and self-rights when it tips over. The self-recovery system sounds like a footnote until you actually send Beni across gravel and watch it eat dirt, bounce back up, and continue tracking without a second of interruption. A creator who ran it through grass, gravel, trails, and concrete stairs for a full week reported footage that never stopped rolling. That kind of reliability is what separates a robot from a gadget.

The camera shoots 4K with HDR coloring that handles extreme lighting without manual adjustment. HybridFlow Stabilization combines four-axis mechanical and electronic stabilization, which keeps footage smooth even when Beni is clattering across rough terrain at full speed. Storage is 32GB internally with a microSD slot for expansion. Everything core, tracking, filming, obstacle avoidance, runs entirely on-device, and your footage stays on the robot’s local storage. The only cloud connection Beni ever makes is for the optional AI auto-editing feature, and only when you explicitly give permission. For a camera product in 2026, that’s a privacy setup worth paying attention to.

Tail tracking, side tracking, front tracking, and circle tracking give Beni four distinct cinematic modes, each producing footage that most solo filmers never actually get to see. Tail mode keeps Beni behind you at pace, a proper chase-cam shot without anyone chasing. Side mode locks it right beside you, framing you against the rushing landscape the way a Hollywood tracking rig would. Front mode has Beni reversing ahead while filming your face, which turns out to be unusually good for capturing effort and expression mid-activity. Circle mode sends it into a continuous 360-degree orbit for sweeping hero shots. The magnetic Motion Tracker, worn on your wrist or clipped to your bag, keeps Beni locked on even when you briefly move out of frame, and doubles as a manual controller for directing movement without touching your phone.

Beni’s auto-editing feature builds highlights from your raw footage using AI models that Mondo Robotics has customized, so you’re not manually scrubbing through hours of clips for ten good seconds. The company is clear that this is the only opt-in cloud feature; every other function runs locally. Pilot Mode sits at the other end of the control spectrum, letting you drive Beni manually from your phone for footage from its low-slung ground-level perspective. The vantage point is genuinely unlike anything a handheld camera can produce, just by virtue of where the lens sits relative to the ground.

Battery life runs 1.5 hours per charge, with a swappable design that stretches to 4.5 hours across three batteries. The Standard Combo, including Beni, Motion Tracker, hat, chest protection plate, and one battery, is available at $549 (discounted from the $799 retail price). The Explore Combo adds three batteries, a charging hub, and off-road wheels for $699. Individual add-ons, including extra batteries at $59, off-road wheel pairs at $9, ears and hats packs at $9, and protection plates at $9, are available separately. Every unit comes with a 12-month limited warranty from the date of activation, and global customer support is available at mondorobotics.com. Shipping covers the US, Canada, UK, Germany, France, Spain, Italy, Australia, and New Zealand, with customs fees included on a flat $10 shipping fee.

Click here to Buy Now: $549 $799 (31% off). Hurry, only 39/230 left! Raised over $2.8 million.

The post Beni Is What Sony Aibo Grew Up to Be – A $549 Camera Robot That Captures Life While You Live It first appeared on Yanko Design.

AI Surveillance Cameras Can’t See Through This Jacket

Posted on by

The last time I really thought about what my outfit was doing for me, I was considering whether my summer dress made me look like I was fashionable or it was just perfect for the hot weather. Simon Weckert has me rethinking the question entirely. Do my clothes do something for me other than protect me from the elements or make me look cool? Can they be about something more?

The Berlin-based artist recently unveiled Digital Camouflage, a conceptual garment collection that does one very specific, very remarkable thing: it makes you invisible to AI. Not invisible to the human eye, obviously. You’d still be fully, stylishly present to anyone walking past you. But to the surveillance cameras powered by AI-based person detection systems that increasingly populate airports, streets, and public spaces, you essentially cease to exist. That’s not a metaphor. It’s the engineering.

Designer: Simon Weckert

The collection works through what researchers call an adversarial attack, a method of manipulating visual data so that a machine misreads what it’s looking at. The garments are printed with a continuous, generative pattern that floods AI recognition systems with computational noise. The algorithm looks at you, processes the visual information, and fails to register a person. You’re there. The camera sees you. But the machine watching through it draws a blank.

At first glance, the clothes look like bold graphic fashion. The kind of abstract print that could easily headline a capsule collection or land on an editorial mood board. The technical function is completely hidden in plain sight, which is, when you think about it, a fairly elegant bit of design irony.

Previous anti-recognition designs relied on fixed printed patches, a kind of static visual disruption that quickly lost effectiveness when fabric folded or camera angles shifted. Weckert’s collection uses a continuous adversarial texture across the entire garment surface, designed to hold up against both movement and perspective change. It’s not a sticker. It’s the whole jacket.

The collection is manufactured in Latvia from a blend of 65% recycled polyester and 35% polyester, with the adversarial pattern applied through digital textile printing. For something that sounds like it belongs in a speculative fiction exhibit, it’s a remarkably practical construction.

And yet, the project lives comfortably in conceptual territory. Weckert is not selling these at retail. Digital Camouflage was exhibited at Ars Electronica Festival 2025, which carries a certain kind of cultural weight. Ars Electronica has long been the venue where designers and technologists come to ask uncomfortable questions rather than offer comfortable answers. Showing the work there feels deliberate. This isn’t a product pitch. It’s a provocation.

I think that tension is exactly where the project is most interesting. The garment works, or is designed to work, but its purpose is as much about raising the question as answering it. Who should have the right to walk through a public space without being catalogued? At what point does automated surveillance stop being a security tool and start being a form of ambient control? Weckert isn’t handing out a manifesto. He’s handing out a jacket.

The fact that we even need clothes that hide us from cameras is worth pausing on. Digital surveillance infrastructure has expanded so quietly and so consistently that most of us have simply accepted it as background noise. AI-powered cameras now occupy the same mental category as streetlights: ubiquitous, impersonal, unremarkable. Weckert’s collection interrupts that acceptance, not through argument, but through design.

Fashion has always been a negotiation between the self and the seen. We dress for impressions, for identity, for protection against the cold. Adding protection against algorithmic recognition to that list feels like a genuinely new chapter in a very old human conversation about who gets to see us and on whose terms.

Whether or not Digital Camouflage ever becomes something you can actually buy, it has already done its most important work. It made an abstract threat feel like something you can put on your body. And sometimes that’s all it takes to make people start paying attention.

The post AI Surveillance Cameras Can’t See Through This Jacket first appeared on Yanko Design.

Most Laptops Seal the Battery In, Dynabook’s New One Lets You Swap It

Posted on by

AI on laptops has created a new kind of battery problem. The promise of on-device AI processing, where the work stays local rather than routing through a cloud server, is genuinely useful for anyone handling sensitive data. But running a neural processing unit alongside everything else a laptop already juggles accelerates battery wear in ways that weren’t as much of a concern three years ago. The battery that comes with a new laptop today degrades faster than the hardware around it.

Most manufacturers solve this by designing the battery to be inaccessible. The result is that when a battery eventually loses capacity, it’s either a service center visit or a new laptop. Dynabook’s 2026 fall and winter lineup takes a different position by making the battery user-replaceable in two of its Panther Lake models, the 14-inch XP9/XP7 and the 13.3-inch X9/X7. A sliding panel lets owners swap it themselves without tools or a technician.

Designer: Dynabook

The reasoning is spelled out plainly in the press release: heavy AI use accelerates battery degradation, and forcing users to choose between repair costs and a full replacement creates unnecessary financial pressure and waste. A replacement battery sold separately as an optional accessory means the rest of the hardware, which is still perfectly functional, gets to stay in use longer. It’s a longevity argument that happens to line up with right-to-repair expectations at the same time.

The XP9/XP7, a 14-inch Copilot+ PC, comes in at 999 g, while the X9/X7 at 13.3 inches checks in at 948 g. Both run Intel’s Core Ultra Series 3 Panther Lake processors. The X9/X7 claims up to 17 hours of video playback and up to 32.5 hours at idle, which is a considerable buffer for a machine that doubles as an on-device AI workstation. Both models pass MIL-STD-810H military-grade durability testing.

Connectivity is genuinely comprehensive for an ultra-light design. Thunderbolt 4 and USB 3.2 Gen 1 Type-A ports sit on both sides of the chassis, alongside HDMI output, a LAN port, and a microSD card slot. That’s a full professional port layout without a dongle in sight, which matters for anyone who has spent time hunting for the right adapter in a conference room. Dolby Atmos handles audio on both models.

The on-device AI processing is the capability that sits behind the user-replaceable battery feature in a practical sense. Running local AI keeps private and sensitive data from leaving the machine, which the press release frames as a concrete advantage over cloud-dependent workflows. The AI privacy assist feature also detects attempts to view the screen from unintended angles, adding a physical layer to what would otherwise be a software-only privacy story.

For those who want the same 13.3-inch form factor at the lowest possible weight, the G9 comes in at 899 g with the same Panther Lake processor and MIL-STD-810H certification. It doesn’t include the user-replaceable battery design, but it undercuts the X9/X7’s weight by nearly 50 g, which matters when the laptop travels daily. All models launched in Japan on August 7, with no global availability confirmed yet.

The post Most Laptops Seal the Battery In, Dynabook’s New One Lets You Swap It first appeared on Yanko Design.

Google Built a Translation Device That Doesn’t Need Google’s Own Cloud

Posted on by

Real-time translation has quietly become one of the most useful things AI can do, and most of the solutions doing it well are deeply dependent on cloud infrastructure. The phone apps and earbuds that translate conversations on the fly are continuously routing audio through remote servers, which works fine until the network doesn’t. Travel to areas with spotty coverage, use a device that can’t roam internationally, or simply value keeping your conversations off someone else’s server, and the whole pipeline collapses at the exact moment you need it most.

The Gemma Translator is a direct answer to that. It’s a handheld voice translation device that runs its entire AI processing pipeline locally on a Raspberry Pi 5, with no internet required after the initial setup. There’s no cloud to route audio through, no API key to maintain, and no service that can be discontinued. You speak into it, and it translates and speaks back, all on the same pocket-sized device.

Designer: Google

The hardware is deliberately approachable. A Raspberry Pi 5 with 8 GB of RAM powers the whole thing, paired with a 480×320 touchscreen, a microphone, a speaker, push-to-talk buttons, and a rotary encoder for selecting languages. Everything fits inside a 3D-printed enclosure in yellow and black, and the STL files for printing it are included in the open-source repository, so anyone with a printer can replicate the physical build exactly.

The software stack is what makes the offline operation genuinely work at useful speeds. Translation runs on Google’s Gemma 4 E2B model via LiteRT-LM, hitting around six tokens per second on the RPi 5 while using under 1.5GB of RAM. Speech recognition uses Moonshine, which is reportedly up to five times faster than Whisper Tiny. Text-to-speech output runs through Kokoro. The result is a complete speech-in, speech-out translation loop that stays on the device from start to finish.

The interface was designed specifically for face-to-face conversation rather than solo use. Two modes let the device handle either one person managing both sides of the exchange or two people independently operating their own microphone and language selection. Physical buttons keep recording controls tactile, which matters when you’re trying to have a conversation rather than navigate menus.

The privacy argument is almost incidental to how useful the offline capability turns out to be in purely practical terms. Field researchers in areas without reliable networks, travelers moving through places where international data is expensive, and anyone working in a sensitive conversation context all get the same thing: a translation device that doesn’t require a connection to keep working. The fact that none of the audio ever leaves the device is simply a consequence of building something that had to work without one.

The post Google Built a Translation Device That Doesn’t Need Google’s Own Cloud first appeared on Yanko Design.

The $599 AI Guitar That Finally Makes Playing Easy

Posted on by

At some point in their life, almost everyone has picked up a guitar, struggled with a C chord for ten minutes, and quietly put it back down. The fingertip pain. The out-of-tune strings. The YouTube tutorial that somehow assumes you already know what a fret is. Learning guitar has always had a steep and slightly humiliating entry curve, and most of us just gave up. The TemPolor Melo-D thinks that’s a problem worth solving, and the design it came up with is genuinely hard to ignore.

Designed by inDare Design Strategy Limited, the Melo-D is billed as the world’s first generative AI guitar, and even that description barely scratches the surface of what it actually is. On the outside, it looks like a guitar that a product designer and a tech enthusiast co-parented together. The body is sleek, the finish is pearl white, and it has a foldable build that makes it compact enough to take almost anywhere. But the real story is what happens when you turn it on.

Designer: inDare Design Strategy Limited

The neck is embedded with LED lights that illuminate exactly which notes to press, removing the guesswork for beginners entirely. A built-in 2.4″ LCD flip screen on the body handles chord guidance and ambient visuals without you ever needing to prop up a phone nearby. These feel like small details, but they represent a real design philosophy: the instrument itself should carry the full experience. You shouldn’t need a separate ecosystem of apps and accessories just to get started.

What makes the Melo-D genuinely interesting is the AI engine running underneath all of it. The guitar operates in three modes: solo, sing-along, and AI creation. In AI creation mode, you can hum a melody and the guitar transcribes it into a guitar solo. You can type a text prompt and it generates a song. Upload an audio file and it converts the music into playable tabs or chord charts. It even generates personalized backing tracks using emotional tags, so the sound starts to feel like yours rather than something borrowed from a practice app.

The “AI everything” wave has made it easy to be skeptical every time a product slaps the word generative onto its marketing. But the Melo-D doesn’t read as a gimmick dressed up in smart-instrument clothing. The six-string silicone system allows for real, note-by-note playing, which means it isn’t just a toy approximating the feel of music. It’s an actual instrument with training wheels you can choose to remove. The goal isn’t to replace the skill of playing guitar. It’s to lower the barrier enough that more people actually get through the door.

inDare’s design work here deserves its own mention. The studio, known for thoughtful human-centered industrial design, brought a lot of restraint to a product that could have easily become overcrowded with features. The foldable form factor, the integrated flip screen, and the silicone strings all feel like deliberate decisions, not afterthoughts. The Melo-D has also earned recognition at the iF Design Awards, which makes sense, because from a design standpoint it reads as a product with a clear and confident point of view rather than a collection of specs looking for a shell to live in.

At $599, it’s available for pre-order now with an expected ship date in September 2026. For a piece of hardware this layered, that number asks you to take it seriously, which feels appropriate for what it’s trying to do.

Learning guitar used to mean a real commitment: years of practice, a teacher, calluses, patience, and a certain willingness to sound terrible for a while. The Melo-D isn’t trying to skip all of that, but it is offering a more honest entry point for people who actually want to try. Whether you’re a complete beginner or someone who has quietly given up on the guitar more than once, it meets you where you are. And sometimes, that’s exactly the invitation you need to actually show up.

The post The $599 AI Guitar That Finally Makes Playing Easy first appeared on Yanko Design.

Samsung and Google Are Trying Smart Glasses Again, at the Worst Possible Time

Posted on by

“Glasshole” entered Urban Dictionary as word of the day in March 2013, which tells you roughly how long it took society to make up its mind about face-mounted cameras. Google Glass had barely reached early users when bars, restaurants, and theaters started banning it. The device died from a social verdict so decisive it became a punchline. A decade later, Google is back. At Samsung’s Galaxy Unpacked in London today, the company helped reveal a new generation of AI-powered smart glasses, co-engineered with Samsung, dressed up in Warby Parker frames, and powered by Gemini. Same idea. Different fonts.

The climate into which these glasses arrive is not quiet. Meta spent the better part of 2026 trying to convince the public that AI glasses are aspirational, recruiting Kylie Jenner for the Meta Starfire Kylie Edition at $399 and positioning it as a style statement. What followed was a wave of coverage about covert recordings, a social media tutorial on disabling the LED light meant to prevent covert recording, and a protest campaign near Meta’s headquarters branding the glasses “fascist.” Some owners are reportedly now reluctant to wear them in public, having learned that the street name for their $399 fashion statement is “pervert glasses.” Google could not have picked a stranger moment to try again.

Designers: Samsung, Google, Warby Parker, and Gentle Monster

So what exactly did Samsung and Google show up with today? The glasses, branded Intelligent Eyewear, come in two designs: a slim black frame from Gentle Monster with a refined, fashion-forward silhouette, and a rich brown Warby Parker edition with a subtly upswept brow line meant for everyday wear. Under the hood, the Snapdragon AR1 Gen 1 chip handles the processing, the same silicon inside Ray-Ban Meta’s glasses, which is either reassuring or ironic depending on your mood. Gemini runs as the AI assistant on Android XR, the platform Samsung and Google co-developed, delivering information spoken into your ear rather than displayed on a screen. Battery life sits at nine hours, with a charging case that carries up to seven additional full charges. There is a built-in camera. There is always a built-in camera.

The positioning is reasonably clever and deliberate. By going audio-first with no display, Samsung sidesteps the cyborg-visor aesthetic that made Google Glass so alienating, and by partnering with Gentle Monster and Warby Parker, two brands with genuine fashion credibility, they are betting that good frames can do what celebrity endorsements and $400 price tags couldn’t for Meta. The glasses also pair with both Android and iOS, which removes the ecosystem friction that would otherwise halve the addressable market before a single unit ships. On paper, this is the most considered entry into smart glasses anyone has made.

The paper, unfortunately, has to reconcile with the sh*tstorm that is Meta and Kylie’s brand collab. Kylie Jenner’s job in the Meta Starfire campaign was to make AI glasses feel safe and aspirational, particularly for women. She wore them, posed with them, starred in content shot from the glasses’ own point of view. What nobody at Meta apparently modeled was the possibility that Jenner’s involvement would become the story, and the story would be unflattering. Lorde publicly called the glasses “not sexy.” Women across Brussels and beyond reported being filmed without consent by men wearing the exact product Jenner was paid to glamorize. Meta had to patch their own hardware to stop people from disabling the recording indicator light. The fact that a patch was required is its own headline.

Samsung and Google have built in visible recording indicators and user-controlled privacy settings, which they are smart to advertise loudly. But the problem Meta created for the entire category goes deeper than indicator lights. The public now has a fixed mental model of what a camera on someone’s face means, and it was not built by reading spec sheets. It was built by watching someone get filmed on the bus, by the NameTag story, by “pervert glasses” becoming the phrase that stuck when “Intelligent Eyewear” didn’t.

Google killed Glass in 2015 because society rejected the behaviour the device enabled before the technology even had a chance to mature. Everything Samsung unveiled today is genuinely more refined, more considered, and more fashionable than anything Glass ever was. The question is whether any of that is the actual problem. Going by recent evidence, the frames were never really the issue.

The post Samsung and Google Are Trying Smart Glasses Again, at the Worst Possible Time first appeared on Yanko Design.

SAM Might Be the Most Human Thing Samsung Has Finally Built

Posted on by

Most conversations about aging and technology eventually circle back to the same frustrating gap: the tools we build for older people rarely feel like they were designed with older people in mind. They’re functional, clinical, and often carry the visual language of a hospital supply catalog. SAM, a concept robotic mobility aid designed by David Webber for Samsung, feels like a direct response to that problem, and it’s the kind of design thinking I genuinely want more of.

At first glance, SAM looks like it belongs on a near-future film set. The rounded, cloud-white body, the articulating arms, the small expressive face-like display on top. But spend a little more time with it and the aesthetics start to make sense in a very deliberate way. Webber didn’t want SAM to read as medical equipment. He wanted it to feel like a companion, something you’d actually invite into your home and not hide in a corner when company comes over.

Designer: David Webber for Samsung

What SAM does differently from most mobility aids is where the real story lives. A standard walker or rollator solves one problem: moving from point A to point B. SAM looks at the fuller picture of what daily independence actually requires. Picking up a dropped object from the floor. Carrying groceries from the door to the kitchen. Helping someone rise from a chair safely. These are the small, invisible tasks that quietly determine whether an older adult can continue living alone or not, and they are almost entirely absent from the current conversation around mobility technology.

The concept is built around the idea of physical AI, the notion that artificial intelligence should operate in the physical world in a way that supports human capability rather than overriding it. SAM doesn’t take over. It doesn’t assume. The user stays in control, and the technology fills in the gaps, which is a harder design problem than it sounds. Building something that assists without patronizing, that helps without making a person feel like they’ve surrendered something, requires a kind of empathy that most tech doesn’t bother with.

I’ll be direct: the design of SAM is one of the more thoughtful things I’ve seen come out of a Samsung concept in a while. The color options, soft greys and muted sage greens, feel intentional. They’re domestic. They say “living room” rather than “rehab facility,” and that distinction matters enormously for user dignity and adoption. Nobody wants to use a tool that announces their limitations to every visitor who walks through the door. SAM seems to understand that.

The arms and end-effectors are worth a close look too. The gripper design is specific enough to handle real objects but rounded and soft-looking enough not to feel threatening. The integrated display sits at a natural eye level when someone stands and grips the handles, showing simple information like time, a to-do list, and device status. It’s thoughtful UX layered into an already thoughtful form.

Is SAM a product you can buy right now? No. It’s a concept, a vision piece, and it should be taken as one. But the best concept designs do something important: they shift where the conversation is pointed. Right now, too much of the assistive tech space is focused on reactive solutions, devices that help after a fall, after a diagnosis, after independence has already started to erode. SAM is proposing something more proactive. It’s asking what support looks like before the crisis, and that’s a question the industry desperately needs to sit with.

What I hope designers, healthcare companies, and tech brands take from SAM is not the specific form but the underlying philosophy. That independence isn’t just about mobility. That dignity is a design requirement, not a bonus feature. That the people who need the most thoughtful design often receive the least considered version of it. If SAM pushes even one conversation in that direction, David Webber has done something genuinely worth paying attention to.

The post SAM Might Be the Most Human Thing Samsung Has Finally Built first appeared on Yanko Design.

Lenovo Just Made a $44 Kid Phone That Kills TikTok and the Browser

Posted on by

Most phones marketed at kids are really just smartphones in disguise, dressed down with monitoring apps and screen-time limits that parents set and children quietly find ways around. The tension between staying connected and staying focused has pushed many families toward patchwork fixes: a shared phone, an old handset with most apps deleted, or a prepaid SIM dropped into hardware never meant for the job.

Lenovo’s answer to that problem is the AI Student Phone L1, which we covered at launch in China in late June 2026 for 299 yuan, or about $44. It looks like a compact candy-bar phone with a 1.83-inch touchscreen and 4G connectivity, but it behaves far more like a purpose-built study companion. There’s no browser, no social media, no games, and based on current reporting, sideloading isn’t an option.

Designer: Lenovo

What makes the L1 interesting from a design standpoint isn’t what’s missing; it’s what Lenovo chose to keep and why. The phone reportedly features a compact body built with Panda glass, a material meant to handle the daily punishment a backpack full of school supplies tends to deliver. That choice isn’t a premium flourish; it’s a quiet signal about who this device is actually built for.

The most telling design decision is the dedicated physical AI button reportedly built into the side of the L1. In most other phones, the AI assistant lives inside software, hidden behind a long press or buried in settings. A hardware button changes that, making help a fixed and visible part of the object. A child doesn’t need to find the feature; the button is it.

Press it, and the L1 reportedly connects to an AI assistant that can help with homework, language learning, and subject lookups. The simplicity is the point. Instead of opening a browser and disappearing down a rabbit hole, the student gets a focused answer and moves on. The phone isn’t trying to replace a teacher; it’s trying to stay out of a teacher’s way.

The feature most coverage tends to bury is the reported QR-code payment system with parent-set spending limits. A child can reportedly take the bus, grab lunch, or pay for a snack without a full smartphone or a physical wallet. GPS tracking and an SOS button reportedly keep parents in the loop. Parents set a cap, the child spends within it, and that’s the end of the negotiation.

That said, the L1 isn’t without its compromises. A 1.83-inch screen is genuinely small, and while that probably discourages mindless scrolling, it also makes extended interaction feel a bit cramped. If handwriting input is included, as some reports suggest, the novelty may wear off quickly. Restraint is a valid design philosophy, but it still has to feel comfortable enough to use every day.

(Image courtesy of @Technology)

For parents in the US or Europe, the L1 is unfortunately a China-only device for now. Products like the Gabb Phone, Pinwheel, and Bark Phone all tackle the same distraction problem, but by restricting a conventional smartphone from the inside. The L1 appears to start from a narrower behavioral premise, building limits into the hardware rather than patching them in through software.

That distinction is worth sitting with. Handing a child their first phone has always been about balancing independence and oversight, and software-based restrictions have always felt like a workaround. The Lenovo AI Student Phone L1 raises the question of whether you can get closer to that balance by designing the object differently from the start, and that’s probably the right conversation to be having.

The post Lenovo Just Made a $44 Kid Phone That Kills TikTok and the Browser first appeared on Yanko Design.

OpenAI and Work Louder Just Built a $230 Physical Remote for AI Agents

Posted on by

Managing multiple AI agents at once has become a surprisingly real friction point in modern knowledge work. Developers using tools like OpenAI’s Codex, which is closing in on nine million users, often run several threads simultaneously. Constantly switching between windows and chat screens to check each agent’s progress breaks the kind of focused flow that agentic workflows are supposed to create in the first place.

That’s the problem the Codex Micro, a collaboration between OpenAI and boutique keyboard maker Work Louder, tries to solve. It’s essentially a compact mechanical macropad purpose-built as a physical command center for AI agents, keeping the most important controls and status readouts within arm’s reach without requiring you to look at your screen. It’s a small but thoughtfully considered piece of hardware for heavy Codex users.

Designer: Work Louder x OpenAI

The heart of the device is its six frosted Agent Keys, each mapped to a separate Codex thread. They light up in different colors to reflect each agent’s status: white for idle, blue for thinking, green for completed tasks, amber when it needs your input, and red for errors. A single tap selects an agent in the background; a double tap brings it to the foreground.

The dial and joystick add two more dimensions of tactile control. The rotary encoder lets you adjust how deeply your agent reasons through a problem, dialing things down for quick tasks and up when the work calls for more careful consideration. The four-directional joystick can be mapped to skills like toggling plan mode or navigating between threads, and it’s never more than a thumb’s flick away.

The largest button, which takes up twice the space, is for push-to-talk. A double-tap starts recording, and the ambient RGB shifts to a snaking green pattern to confirm it’s live. Once you stop speaking, a white effect takes over while Codex processes your voice, and a solid white cue signals it’s ready to send. It makes submitting a prompt feel more like dictating than typing.

Beyond what it does, the Codex Micro is built to feel as good as it works. The body is made from sandblasted polycarbonate that diffuses the RGB lighting into a soft, even glow. The base, a solid aluminum block, keeps the whole unit planted on the desk thanks to an anti-slip ring. Buyers can choose between clicky or silent low-profile switches, with USB-C and Bluetooth connectivity.

What sets it apart from a generic programmable macropad is its direct integration with Codex itself. There’s no extra software to download or vet; key assignments and layouts can be adjusted natively within Codex. For more advanced setups, Work Louder’s Input app unlocks up to six programmable layers, letting you group different workflows into dedicated layouts without needing more desk space. It works on both Mac and Windows.

The Codex Micro ships with a USB-C to USB-C cable and the Codex Icon Keyset, a collection of 32 custom-designed keycaps with icons specific to Codex functions. It’s the kind of peripheral that’ll appeal to developers who’ve started treating their AI agents less like chat windows and more like teammates running parallel tasks. It’s priced at $230, and stocks have been limited since its launch.

The post OpenAI and Work Louder Just Built a $230 Physical Remote for AI Agents first appeared on Yanko Design.

nubia Just Built the World’s First Phone That Does Things on Its Own

Posted on by

Smartphone makers have been adding AI features for a few years now, and at this point, the marketing language around them has become fairly predictable. A phone summarizes your messages, suggests reply text, cleans up your photos, and maybe transcribes a meeting. These are genuinely useful tools, but they’re still responding to individual requests. The phone doesn’t actually do anything on its own. Someone still has to ask.

The nubia NaviX Ultra is making a more ambitious claim. Unveiled at the World Artificial Intelligence Conference in Shanghai, it’s being positioned as the world’s first AI agent smartphone, and the distinction matters. An AI agent doesn’t just respond to prompts; it operates across apps and services on behalf of the user, handling multi-step tasks without being asked to execute each one individually. Nubia is calling this system-level integration rather than a feature layer.

Designer: nubia

That distinction isn’t entirely hypothetical for nubia. The NaviX Ultra’s predecessor, the nubia M153 with Doubao AI assistant, debuted earlier this year in partnership with ByteDance and sold out its initial 30,000-unit batch on launch day in China. The NaviX Ultra is the commercial follow-up to that limited run, now carrying the Doubao assistant at a system level that nubia describes as enabling autonomous cross-platform operation. The phone also received the 2026 SAIL Award ahead of its unveiling.

The hardware specifics haven’t been fully confirmed yet. What nubia has revealed ahead of launch is the design: a metal-framed body with exceptionally narrow bezels, a triple rear camera arranged in a horizontal Pixel-style bar, and a dedicated AI button on the frame. The phone comes in four colorways: Blue Horizon, Black, Dreamscape, and White. Whether the internal hardware matches the ambition of the software positioning is still an open question until full specs are released.

What’s interesting about the NaviX Ultra’s framing isn’t just the technical claim. It’s what the AI agent model implies about how people actually use their phones. The friction of switching between apps, re-entering context, and manually chaining tasks together is real, and it’s something that feature-level AI has only partially addressed. A phone that can hold that context across apps and act on it changes the interaction model rather than just adding a new tool to an existing one.

Nubia’s Doubao partnership with ByteDance gives the NaviX Ultra’s AI ambitions a specific cultural and software context. Doubao is one of the most widely used AI assistants in China, which gives this particular agent integration a large user base to learn from. Whether the system-level AI agent lives up to the label is something that will only be testable after the full device is in hand, but nubia is at least asking a more interesting question than most of its competitors have been willing to pose publicly.

The post nubia Just Built the World’s First Phone That Does Things on Its Own first appeared on Yanko Design.