Tag Archives: Artificial Intelligence

Speediance Built the Home Gym That Learns Your Body Like a Coach Would

Posted on by

Speediance built its name on a simple idea: swap the bulky machines of a typical gym for real digital resistance packed into one freestanding unit. That idea hasn’t changed, but the ambition behind it has grown. The company now wires its hardware lineup, Gym Monster 3, Gym Nano, and the Speediance Strap, into one learning system that notices more than the workout in front of it.

Gym Monster 3 anchors that system, and the first thing worth noticing about it has nothing to do with resistance numbers. It’s freestanding, meaning it doesn’t need a wall mount, bolted brackets, or a dedicated room. Once it’s set up, it delivers up to 220 lbs of digital resistance across 19 cable positions, enough range to cover a full strength session without swapping equipment.

Designer: Speediance

Gym Nano takes that same resistance technology and shrinks it into something you can carry. At 8.2 kg (about 18 lbs), it still produces up to 220 lbs of digital resistance, along with real-time velocity and power readings. It even measures peak force, rate of force development, and left-right asymmetry, precision that once needed a sports science lab to capture.

The Speediance Strap works quietly in the background, and that’s the point. There’s no screen, no notifications, nothing demanding attention between sets. Using greenteg Calera technology, it reads subtle shifts in wrist temperature over time, building a picture of how the body handles training, sleep, and heat across days, not one session. That signal is what gives the system something to work with.

Gym Monster Ultra rounds out the lineup for anyone chasing a full home rig. It pushes digital resistance up to 260 lbs across 33 cable positions, adding a vertical high-lat pulldown, a multi-grip pull-up bar, and an adjustable thigh restraint. Between these four products, Speediance now has a compact all-in-one, a portable tool, a passive sensor, and a full training station, all connected.

What ties four separate products together is AEVIS, an AI layer built to carry context from one day into the next. Rather than answering a question and forgetting it, AEVIS holds onto training history, recovery signals, sleep, and schedule changes, then lets that shape what comes next. In one demo, a user mentioned a six-hour drive ahead, and the system quietly lowered the planned load, leaning toward recovery.

AEVIS Pre-Trip

AEVIS During Trip

Votus is the part of the system that actually watches you train. Point a single phone camera at yourself, and it builds a 3D read of your motion, tracking rep speed and form while the set happens. That kind of analysis once needed a lab full of sensors, now running through a camera most people already carry. Speediance runs this inside its own ecosystem, not as separate software.

AEVIS Post-Trip

While Votus handles the moment, Oryn thinks about what should happen next. It’s focused on exercise, nutrition, and sleep, and it doesn’t treat each workout like it exists in a vacuum. If the Strap picks up signs of poor recovery, or the calendar shows a demanding week, Oryn can shift the plan toward a lighter load, a different focus, or recovery instead of strength.

Votus

Somni takes this further, modeling how today’s choices might affect tomorrow’s fatigue, sleep, appetite, and capacity to train, without committing to one guaranteed outcome. Speediance calls it a technology preview, not a medical tool, and that same honesty applies elsewhere. AEVIS doesn’t currently adjust Gym Monster’s resistance on its own. That deeper link between AI decisions and the machine’s behavior is still ahead.

Speediance started in 2020 with one connected machine, and now spans training hardware, portable tools, passive sensing, movement analysis, and AI-driven decisions. The company sums up its approach in four words: Sense, Decide, Act, Learn. Each product holds a spot in that loop, and the work now is making those spots communicate more fluently over time. It keeps paying closer attention with every session.

The post Speediance Built the Home Gym That Learns Your Body Like a Coach Would first appeared on Yanko Design.

You Can Finally Pre-Order a Googlebook, Starting at $899

Posted on by

A laptop announced in May with no price, no ship date, and no way to actually buy one is easy to forget by September. Googlebook just stopped being that kind of announcement. Pre-orders opened today across five hardware partners, Acer, ASUS, Dell, HP, and Lenovo, turning a concept built around Gemini Intelligence into something arriving on shelves within two weeks, starting at $899.

That timing matters because it answers the question a spec sheet alone can’t: what does an Android phone actually gain by having a matching laptop nearby? Continue On reopens a task straight from the phone onto the laptop’s taskbar, while a feature called Files reaches into the connected phone’s storage to pull up a photo or document without any manual transfer step in between.

Designers: Google, Acer, ASUS, Dell, HP, Lenovo

Cast My Apps pushes that further, streaming Android phone apps directly into desktop windows alongside the Chrome browser itself. An app built for a 6-inch screen suddenly runs in a resizable window next to a spreadsheet, treated as a native piece of the desktop instead of a screen mirror bolted on as an afterthought. Passwords, Wi-Fi credentials, and messages move over encrypted, without a manual re-login on either end.

Gemini sits underneath all of it as the connective layer rather than a separate app to open. Magic Pointer turns a cursor into a contextual assistant, summoned to check an email, schedule something, or touch up an image based on whatever it’s hovering over at the moment. Rambler takes a different angle entirely, converting messy spoken notes into text organized enough to hand off without a rewrite.

Create My Widget pushes personalization somewhere most laptops never bother going. Describing a tool in plain language builds a working desktop widget on the spot, no coding background required, no template to hunt down first. It’s a small feature with a large implication: customizing a laptop’s actual function, not just its wallpaper, becomes as easy as typing a sentence.

Developers get an unexpectedly serious toolkit bundled in as well. Google Antigravity ships standard, letting anyone write an app and deploy it directly onto the device without extra setup. A full Linux terminal sits underneath that, running tools like Claude Code for genuine agentic coding sessions, a level of openness rarely handed to a mainstream consumer laptop straight out of the box.

Design stays consistent across all five brands through a single shared signal: the Glowbar, a strip of light Google requires on every Googlebook regardless of who builds it. Acer, ASUS, Dell, HP, and Lenovo each bring their own chassis language and price point, yet that glowing strip functions like a shared handshake, letting someone recognize the platform before ever checking a spec sheet.

Every Googlebook also ships with 12 months of Google AI Pro, unlocking 5 TB of cloud storage and Gemini’s advanced tools alongside three months each of YouTube Premium and Adobe Photoshop. Google backed that with a commitment to feature drops and software updates for up to 10 years, treating a laptop purchase as a decade-long relationship rather than a two-year cycle before the next upgrade feels necessary.

The post You Can Finally Pre-Order a Googlebook, Starting at $899 first appeared on Yanko Design.

The $499 Lapis One Is A Tiny Mac mini Alternative Built Just to Run Your AI Agents

Posted on by

Earlier this year, people were buying Mac minis by the cartload just to give their OpenClaw agents somewhere to live, and Apple Store staff reportedly started calling it the “OpenClaw machine.” Pamir’s Lapis One looked at that trend and asked a very reasonable question: what if the agent’s computer was actually designed for the agent? The answer is a coaster-sized Linux box with a Rabbit R1 orange cap, an iMac G3 two-tone body, and a pixel robot snoozing on its little screen. Your AI assistant gets its own desk, its own internet connection, and its own battery, while your laptop finally gets some peace. Frankly, it’s a nicer setup than my first apartment.

Under that playful shell sits a Rockchip RK3576 chip with a 6 TOPS NPU, 8GB of LPDDR5 RAM, 64GB of storage, and room for an M.2 SSD and a microSD card. The standout feature is Agent KVM, a dedicated USB-C port that captures another computer’s display and fakes a keyboard and mouse, letting your agent operate a second machine the way you would. Tailscale comes baked in, so you can check on your agent from your phone in the grocery store checkout line. Pamir, a San Francisco startup that previously shipped the Distiller dev kit, has pre-orders open at $499, with December shipping to the US. The regular price will be $599 once pre-order pricing ends.

Designer: Pamir AI

Pick the Lapis One up and the design lineage is impossible to miss. The orange wedge housing the screen is doing a very direct Rabbit R1 impression, right down to the color and the cute pixel-art creature that lives on the display. We covered the R1 when it launched, admiring how Teenage Engineering gave it a Tamagotchi-like charm, and the comparison mostly holds here too. Where the two devices split is purpose. The R1 tried to replace your phone and got clowned for it because your phone already does everything a phone needs to do. The Lapis One doesn’t compete with anything you own. It just wants a permanent outlet and a Wi-Fi password, which is a much easier ask.

The other reference point is the iMac G3, and it shows up less in the color and more in the attitude. Apple’s original iMac made a whole generation comfortable with computers by wrapping them in candy plastic and betting the farm on USB when almost nothing supported it yet. The Lapis One runs a similar play for Linux, a platform whose default personality is a bare green board and forty tabs of forum troubleshooting. Four labeled USB-C ports on the back handle power, video out, data, and the KVM function, and the whole thing boots into a clean, readable interface instead of a terminal prompt. That is a genuinely useful decision for anyone who wants Linux without becoming a systems administrator.

The screen is a Sharp memory LCD from the same panel family as Panic’s Playdate, a device we have written about more than once for its willingness to be delightfully impractical. That low-power panel means the Lapis One can sit there all day, idle screen glowing faintly, without murdering its 3,300mAh battery, and it doubles as a desk clock when your agent isn’t busy. It is a small detail, but it is the kind of small detail that separates a product from a dev board somebody forgot to case.

None of this erases the real questions. Battery life tops out around four hours, which turns the “personal UPS” pitch into more of a personal nap timer. Agent KVM hands your agent direct control of a second computer’s keyboard and mouse with zero software safety net on the target machine, so you are trusting your own judgment about what that agent is allowed to click. And $499 for what is, on paper, a mid-range ARM chip in a nice box will keep the Raspberry Pi comparisons coming, same as they did the day this thing launched.

Still, I like where Pamir landed. The industrial design does the job good industrial design is supposed to do, which is make a genuinely intimidating category of hardware feel like something you would not mind on your desk. Whether $499 buys enough agent to justify the coaster remains an open question until October’s first batch actually ships.

The post The $499 Lapis One Is A Tiny Mac mini Alternative Built Just to Run Your AI Agents first appeared on Yanko Design.

Boya Notra Neo AI Note Taker Has 3 Modes, Never Asks Which One to Use

Posted on by

Conversations rarely stay in one place. A lecture bleeds into a study session over Bluetooth earbuds, and a work call turns into a video meeting minutes later, details slipping through in the shuffle. The Boya Notra Neo, the brand’s second-generation AI note taker, was built around this. It senses whether you’re in a classroom, on a call, or listening through headphones, and adjusts how it records without you touching anything.

That kind of adaptability is the real design story here. Notra Neo runs on three recording modes, Ambient, Phone Call, and Bluetooth, switching between them automatically based on how you’re using the device. Most multi-mode gadgets hand that decision back to the user, which defeats the purpose of having options in the first place. Boya’s approach removes the extra step, letting the device read the room instead.

Designer: BOYA

Ambient Recording covers the face-to-face moments, lectures, meetings, interviews, and group discussions where the microphone needs to pick up a whole room, not just the person holding the device. Boya built in 16 feet (5m) of clear 360-degree pickup, enough range to sit at the back of a seminar hall or the far end of a conference table and still catch every word clearly.

Phone Call Recording does what it says, capturing calls without any extra setup. Bluetooth Recording is the more interesting mode, catching audio that flows through connected devices rather than through open air. That covers online classes, virtual meetings, and video interviews, the kind of conversations that happen entirely through a headset. It’s the mode that acknowledges how much communication now happens through a pair of earbuds instead of a room.

Once something’s recorded, Notra Neo transcribes it into text across more than 140 languages and dialects, tagging who said what. That speaker labeling matters more than it sounds; anyone who’s tried transcribing a group discussion knows how quickly it turns into a wall of unattributed text. For multilingual classrooms or international teams, that same feature means reviewing a conversation doesn’t depend on catching every word the first time.

Transcripts are just the raw material, though. Notra Neo’s AI turns those recordings into summaries, to-do lists, mind maps, and insights, plus more than 60 templates suited to different kinds of conversations. A lecture becomes a set of revision notes instead of hours of audio to relisten to. A meeting turns into a list of decisions and follow-ups. A customer call becomes something worth referencing later.

There’s also AI Search, letting you ask for specific information instead of scrubbing through hours of audio, useful for pulling up a decision from a meeting weeks back or a quote from an old interview. Everything lives inside the Boya OS app and web platform, staying synced across devices, with exports available in multiple formats: audio, text, and mind maps, so content moves easily into whatever workflow someone already uses.

None of this works if the hardware can’t keep up day to day. Notra Neo offers up to 20 hours of continuous recording and 8GB of built-in storage, enough to sit through a full semester of lectures or a multi-day conference without hunting for an outlet. The body stays compact enough to slip into a pocket or a pencil case, easy to carry without a second thought.

What stands out about the Notra Neo isn’t a single feature but the decision to build a recorder around actual habits instead of asking people to change theirs. Boya will officially bring the device to IFA 2026 in Berlin this September, following a second-generation approach that trims complexity rather than adding more of it. Fewer settings to manage, more attention left for the conversation, which was the point all along.

The post Boya Notra Neo AI Note Taker Has 3 Modes, Never Asks Which One to Use first appeared on Yanko Design.

An AI Gave Me A Full-Body Massage: Hands-on with DaVinci AI Pro Massage Chair at IFA 2026

Posted on by

Somewhere out there is a bingo card of jobs AI would come for, and I guarantee nobody had masseuse on it. AI has already eaten its way through a scary chunk of the low-level workforce, copywriting, junior coding, customer support scripts, even the guy behind the wheel in some cities now. But there was always this quiet assumption that a good deep tissue rubdown needed actual human hands, actual human judgment, someone who could feel where the knot was and adjust on the fly. That felt safe. Untouchable, even. Then I walked into Bodyfriend’s booth at IFA 2026, sat down in a chair, and casually talked my way into the best massage of the show. Apparently the human-to-human touch job just got automated too.

The chair was Bodyfriend’s DaVinci AI, running the same generative AI Bodyfriend had just unleashed in Seoul days before IFA, badged DaVinci AI Pro. Instead of picking from a shelf of presets, it listens to plain language and builds a massage from scratch, mixing 134 basic blocks across body zones, pressure, and duration. Bodyfriend puts the theoretical combinations in the trillions, and having felt the difference, I believe the math. A PPG sensor in the hand massager also reads heart rate, heart rate variability, and blood oxygen in real time, so the chair responds to your actual state, not just your words. I did not know any of this walking in. I just knew I was tired, sore, and willing to talk to a chair like it owed me something.

Designer: Bodyfriend

I had clocked around 15,000 steps that day, running on trade show coffee and basically nothing else, and my feet and shoulders had reached the point of open rebellion. I sat down, and instead of hunting for a remote, I just talked to the tablet mounted on the armrest. “Hey Buddy,” I said, “give me a strong foot massage, and also help relieve tension in my neck and shoulders.” Within about a minute it had built something specifically for that, working my feet, shoulders, and spine while leaving my hands completely alone, because I never asked for them. When I got a flicker of FOMO and asked it to include a hand massage too, it recalibrated on the spot rather than restarting from zero. Talking to it felt closer to briefing a person than operating a machine.

The DaVinci name carries some real design pedigree behind it. Bodyfriend’s earlier DaVinci models were developed with Italdesign, the Turin studio behind decades of concept and production cars for names like Lamborghini and Bugatti, and that automotive language shows up in the chair’s sculpted shell and the confidence of its proportions. This year’s leap is less about the shell and entirely about what is happening inside it. Where the original DaVinci used bioelectrical impedance analysis to profile your body composition, the AI Pro generation swaps that for conversational intelligence, a shift from measuring you passively to actually listening to you. It is the difference between a chair that studies your body and one that takes instructions from it.

Bodyfriend’s booth this year also carried the 733, a wearable healthcare robot that took home an IFA Innovation Award honor for independently guiding your arms and legs through stretching and standing assistance, proof this is a company betting hard on robotics well beyond the living room recliner. Against that backdrop, DaVinci AI reads less like a gimmick and more like the company’s flagship idea about where home wellness tech goes next. We covered Bodyfriend’s Quantum robotic chair at IFA 2024 and called it a bodygasm in leather and Bang & Olufsen sound. DaVinci AI trades some of that theater for something quieter and arguably bigger: a chair that finally understands what you are asking for, in your own words, on the first try.

The post An AI Gave Me A Full-Body Massage: Hands-on with DaVinci AI Pro Massage Chair at IFA 2026 first appeared on Yanko Design.

This $99 Clip-On AI Remembers Everything You’ll Ever Forget

Posted on by

If you live a life full of meetings and conversations, there’s a good chance you’ve forgotten something important that was said to you. Sure, you can take notes or jot things down as you go, but not everyone likes doing that, and not everyone has the time to stop and sort through everything they hear during a busy day.

The SwitchBot AI MindClip wants to do the heavy lifting for you by creating a “second brain.” It’s a lightweight, clip-on wearable AI gadget that can record conversations and meetings while also letting you add voice notes whenever inspiration strikes. From there, it automatically transcribes, summarizes, and organizes everything so the information ends up exactly where you need it, whether that’s your calendar, your to-do list, or a searchable archive you can revisit later.

Designer: SwitchBot

Plenty of recording devices already exist, mostly built to capture meetings and interviews. But the MindClip does more than just dump everything into one long audio file. It sorts your recordings into categories like Timeline, Priority, Daily Memory, Weekly Summary, and To-Do Overview. You can even sync it with your calendar and reminder apps, turning those “overheard” details into real, actionable schedules and tasks instead of letting them slip away.

As for the design itself, the AI MindClip is a small, button-like, square-shaped clip that looks more like a tiny badge or pin than an actual recorder. It’s built to blend into whatever you’re wearing rather than stand out, which makes it easy to put on every day and take just about anywhere. The device is made from aluminum alloy and ABS plastic, giving it a premium feel while keeping it both durable and lightweight at just 16.8 grams. It comes with a clip-back mount that easily attaches to collars, jackets, bag straps, and lanyards. It’s also sweat-resistant and splash-resistant, so it’s built to handle real-world scenarios like commuting, phone calls, meetings, and everyday wear without missing a beat.

This tiny device is packed with AI-powered features, including automatic speech-to-text transcription, AI summaries built from a variety of templates, automatic to-do generation, and weekly recap summaries. You can even ask it questions about past recorded audio using its “AskAI” conversational search feature, so instead of scrolling through old notes, you can simply ask what was said and get a straight answer.

Of course, a device like this naturally raises questions about privacy and legality. SwitchBot has built in a number of in-app privacy and data management controls, along with end-to-end encryption for both pairing and file transfer. Even so, there will likely be ongoing concerns given the device’s “always recording” nature. SwitchBot will need to make it clear that these features won’t create problems for the wearer or for the people around them, especially since many U.S. states require everyone in a conversation to consent to being recorded.

Personally, I still prefer pen and paper, or typing things out on a keyboard, when it comes to note-taking and remembering things I need to act on later. But no method is foolproof, and there are definitely moments, like a packed workday or a hectic family schedule, when a tool like this could genuinely come in handy.

At the end of the day, the SwitchBot AI MindClip is an interesting bet on where note-taking might be headed. Instead of asking people to change their habits by pulling out a phone or opening an app mid-conversation, it simply rides along, quietly listening and organizing so nothing important gets lost in the shuffle. Whether it becomes a genuine everyday essential or just another gadget that ends up in a drawer will likely come down to how well the AI features hold up in daily use and whether people feel comfortable wearing something that’s always listening.

Still, for busy professionals, parents juggling a dozen things at once, students sitting through long lectures, or anyone who’s ever walked out of a meeting thinking, “wait, what did we just agree to?”, the MindClip is at least trying to solve a problem most of us know all too well. Sometimes the best kind of tech isn’t flashy. It just quietly helps you remember.

The post This $99 Clip-On AI Remembers Everything You’ll Ever Forget first appeared on Yanko Design.

Reolink’s OMVI 2i Ultra Wireless Security Camera can look in two directions at once: Hands-on at IFA 2026

Posted on by

Picture a courier walking up your driveway while someone else quietly tries your side gate. A normal pan-tilt camera picks one of them and forgets the other exists. Reolink’s OMVI 2i Ultra handles it by carrying two cameras at once: a fixed 4K wide lens on top that never stops watching the full scene, and a 5MP pan-tilt ball below that spins 355 degrees to lock onto whatever the wide lens flags. Both feeds record at the same time, and you can tap anywhere on the wide view to slew the turret straight to that spot. It debuted at IFA 2026 in Berlin, ships late October, and costs between $200 and $250.

A detachable 10,000mAh pack claims six months per charge, and the 6W SolarEase panel says it needs about thirty minutes of daily sun to stay topped up, against the one to two hours Reolink quotes for rivals like Eufy’s S340 and the Tapo C675D. Those are Reolink’s own numbers, so hold them loosely until somebody leaves one on a north-facing wall through a British winter. The rest is easier to verify. Wi-Fi 6 across both bands, up to 512GB of local microSD with no subscription attached, colour night vision from twin spotlights, a siren, and IP65 sealing. At a kilogram with the bracket, it is a proper lump of hardware rather than something you tuck on a shelf.

Designer: Reolink

Pan-tilt cameras have quietly lied to buyers since the format existed. Sell someone a camera that turns to follow motion and you have also sold them a camera that stops watching everywhere else the instant it does. Reolink’s fix is embarrassingly simple once you see it. Stop asking one lens to be both the guard and the chaser, and give it a second lens that only guards. The wide 4K unit up top runs at 110 degrees horizontal and never rotates, so the scene it opened on is the scene it keeps recording no matter where the PT ball below it wanders off to. Reolink calls this SyncTrack, but the actual idea is closer to division of labour than any clever algorithm. One camera commits to context, the other commits to pursuit, and neither has to compromise for the sake of the other.

This is the third time Reolink has built this exact idea in 2026, and each version has been a slightly different argument about what matters most. The X16 landed at CES in January as a 24MP full metal flagship with 16x optical zoom, but it needs an ethernet run to function, which puts a hard ceiling on where it can live. The 3i followed in June with PoE and wired Wi-Fi options at $319.99, still asking for a nearby outlet even in its most flexible form. The 2i Ultra drops the combined resolution to 8MP plus 5MP and, in exchange, cuts the cord entirely. The trade-off is the clever bit, because being able to pop a camera on a fence or a backyard without worrying about wiring is much more convenient than haggling over megapixels.

A squared-off dual-lens head sits above a separate motorized ball, with a solar panel tilting on its own hinge overhead, and the whole assembly weighs a full kilogram once the bracket is attached. Compare that to Reolink’s own Argus MagiCam, a 108-gram cube you stick to a fridge with a magnet, and the difference in ambition is obvious at a glance. This is not a camera trying to disappear into a bookshelf. It reads more like a small piece of maritime equipment bolted under an eave, and that visible mechanical presence is doing some of the deterrence work before the siren ever gets involved.

That dual-lens design is quite literally baked into the name. OMVI reads like omni-vision, a camera built to watch in every direction at once instead of picking a lane. Arlo’s Ultra flagship still wants a wall plug or a proprietary base station to unlock its best tricks, and Nest’s outdoor cameras have quietly stayed wired-first for years now, which makes the 2i Ultra’s fully wire-free build feel like it is actually catching up to how people install things rather than how manufacturers wish they would. No drilled conduit, no hub tucked in a closet, just a battery, a solar panel, and a spot on the fence you actually wanted the camera to go.

That is the real flex against the flagships twice its price. You are not paying for a wall socket you have to plan your whole install around, and you are not paying for a bulky base station eating an outlet indoors either. You get two lenses doing genuinely different jobs, a battery that outlasts most of the category on paper, and a name that finally means something instead of just sounding techy. At $200 to $250, the OMVI 2i Ultra is the version of omni-vision security that shows up ready to mount anywhere the moment you take it out of the box.

The post Reolink’s OMVI 2i Ultra Wireless Security Camera can look in two directions at once: Hands-on at IFA 2026 first appeared on Yanko Design.

Cloud AI Means Someone Else Has Your Data, MINISFORUM Fixes That

Posted on by

Running a serious AI model at home usually means renting someone else’s cloud server and hoping your data stays private along the way. MINISFORUM built its newest pairing to remove that trade-off entirely. The AI Agent NAS N5 MAX-P495 and AI Mini Workstation MS-S1 MAX-P495 combine for up to 131 TOPS of AI compute performance, 192GB of memory at 8533 MT/s, and up to 160GB of graphics memory, enough headroom to keep the heavy lifting on a desk instead of in a data center.

The N5 MAX-P495 takes the shape of a compact black cube, but what it holds inside matters more than how it looks sitting on a shelf. It combines powerful AI computing with up to 200TB of local storage, enabling users to store, manage, and process large volumes of data within a single AI platform. That capacity turns a spare corner of an office into a private archive large enough to hold years of project files, footage, or research without ever touching a third-party server.

Designer: MINISFORUM

Storage alone doesn’t make an AI system useful, so MINISFORUM built the N5 to run the AI itself. From RAG knowledge bases and AI models to embeddings, databases, and AI agents such as OpenClaw and Hermes, N5 MAX-P495 keeps data and AI workloads local, enabling continuous and private AI operations. A small business could lean on that setup to run its own customer-support agent without routing conversations through an outside company’s servers.

That local-first approach extends to how the N5 fits into a larger setup. Designed to work alongside AI computing devices, N5 MAX-P495 serves as a centralized backend for data, models, knowledge, and long-running AI workloads, bringing storage and computing together at the edge. It’s less a standalone gadget and more a quiet engine room, built to keep working in the background while something faster handles the parts people actually interact with.

That faster half of the pairing is the MS-S1 MAX-P495, a compact tower wrapped in a ventilated, diamond-textured front panel that looks more like a piece of desk sculpture than a workstation. It delivers powerful local AI computing for demanding model inference, AI development, and computationally intensive workloads, built for intensive local AI tasks, engineering, and workflows in an ultra-compact footprint. Developers get workstation-class muscle without giving up desk space to a full tower case.

Speed matters just as much as raw power for certain jobs, and MINISFORUM designed the MS-S1 with that in mind. Designed for low-latency AI workloads, MS-S1 MAX-P495 enables responsive local inference for AI agents, computer vision, generative AI, and other real-time applications. A security camera system or a live creative tool built on generative AI benefits directly from that kind of immediate response instead of waiting on a round trip to the cloud.

Connectivity rounds out the workstation’s role in a larger setup. With extensive connectivity and expansion capabilities, MS-S1 MAX-P495 serves as a powerful AI compute frontend, supporting models, applications, and connected devices across flexible edge AI environments. That flexibility means the same box can sit at the center of a home lab or plug into a small studio’s existing hardware without forcing a rebuild.

Paired together, the two devices split the work the way a good team would. The MS-S1 MAX-P495 and N5 MAX-P495 form a complete edge AI computing system, delivering high-performance AI compute at the frontend while integrating compute, massive storage, data, and AI knowledge at the backend, providing a powerful foundation for real-time inference, persistent AI workloads, and scalable local AI applications. It’s a compact answer to a question more households and small teams are starting to ask: whether AI still needs to live somewhere else at all.

The post Cloud AI Means Someone Else Has Your Data, MINISFORUM Fixes That first appeared on Yanko Design.

Plaud One Hands-On at IFA 2026: AI Earbuds That Transcribe Meetings, Then Draft Emails And Decks For You

Posted on by

For three years, Plaud made very good tape recorders with a PhD. You pressed a button, it captured a meeting, and mere minutes later a summary showed up. The Plaud One, launched at IFA 2026 for €249.90, reverses that arrow. Its charging case now holds four beamforming mics, a speaker, an agent button and a built in eSIM, so it stays online without a phone nearby. Hold the agent button and the AI talks back, asks what you meant, then builds the deck or drafts the email while you get on with your day.

That flip from recording to doing needed hardware nobody had built yet. Not a pendant, not a pin, not another app crammed onto a phone you were already staring at. Plaud built a palm sized aluminum box with a fluted lid that looks like a Rimowa suitcase went through a shrink ray, sitting on a table, listening across a room, and talking back like a colleague who never needs coffee. The Explorer Edition opens for pre order now, capped at 2,000 units worldwide, with deliveries landing in Q4. A public version follows in 2027, once Plaud has watched two thousand early adopters put the thing through its paces.

Designer: Plaud

The original Plaud Note clipped onto a phone and did one job, listen to a meeting and hand back a transcript once it ended. The Note Pro added a screen and better mics. The NotePin S shrank the format onto a lanyard clip. Every generation stayed reactive by design, a smart recorder waiting for you to press stop before it did anything useful. That business built Plaud’s 2.5 million user base and reportedly crossed $100 million in annual software revenue by June, which is exactly the kind of runway that lets a company gamble on reinventing its own hardware category from scratch.

Vertical fluting runs down the case’s aluminum lid, the same visual trick Rimowa has stamped into cabin luggage since the 1950s, giving a 62 gram object a sense of density it has no right to carry. Flip it open and the earbuds sit half in ear, silver capped, in a tray that reads closer to a jewelry box than a charging dock. A speaker grille, a record button and a separate agent button line the case’s edge, the three controls that turn a charger into a device you actually talk to. None of it apes a phone. It reads like furniture for a desk, not gear for a pocket.

Earlier Plaud devices leaned on your phone for both processing and connectivity, so the intelligence always sat one hop away from the microphone. Plaud One’s case carries its own 4G LTE connection through a built in eSIM, closing that gap entirely. Capture, transcription and agent response now run in one box, independent of any other screen in your life. Leave your phone in a bag, walk into a cafe, and the case still records the table’s conversation, answers a question out loud through its speaker, or fires off a follow up email on command. Every pendant and pin that came before this quietly depended on a phone doing the real work behind the scenes. Plaud One is the first in the category that genuinely does not.

Plaud Agent drafts a follow up email the moment a client call ends. Feed it a meeting and it builds a slide deck while you sit through your next call, then drops the finished file straight into Notion with zero copy pasting. The platform connects to Gmail, Google Workspace, Microsoft Teams, Slack and Notion today, with more integrations promised as the software matures. Set it loose on a standing task, a monthly competitor scan for instance, and it runs quietly in the background, waiting with a finished report the next time you check. This is the actual bet Plaud is making, that an assistant worth having has to act, not just remember.

Pre orders for the Explorer Edition are open now at €249.90, with the first 2,000 units shipping in Q4. Buyers get $200 in Plaud Credits toward the heavier agent features on top of a starter transcription plan, so the price buys a running start on the software as much as the hardware. Calling this an Explorer Edition rather than a finished release is the honest move, an open beta wearing genuinely sharp industrial design. Plaud spent three years building the best possible way to listen. Plaud One is its first real attempt at building something worth talking to.

The post Plaud One Hands-On at IFA 2026: AI Earbuds That Transcribe Meetings, Then Draft Emails And Decks For You first appeared on Yanko Design.

Anker MindBase Gives Your Old eufy Cameras a Real Brain and 48TB of Memory: Hands-on at IFA 2026

Posted on by

If you already own eufy cameras, the useful sentence in Anker’s IFA 2026 announcement is buried under the AI talk: your HomeBase S380 footage transfers straight across to the MindBase. Anker built the new hub to work with the eufy security gear people bought years ago, then bolted on a two-bay drive enclosure holding up to 48TB and a 26 TOPS compute card running an in-house language model locally. Your old cameras stay. They get a brain. That is a far rarer offer than a new camera, and considerably cheaper than replacing four of them.

The scale of what it absorbs is the part worth sitting with. MindBase acts as a Matter 1.5 controller with Thread and Zigbee support built in, and it works alongside HomeKit, Alexa, Google Assistant and Home Assistant rather than replacing any of them. It supports 20 cameras and 34 sensors, close to 100 Anker devices, and hundreds of third-party Matter products. Two Gigabit Ethernet ports and three USB 3.0 ports handle the wiring, with 64GB of onboard flash before you add a single drive. Anker wants to be the thing everything else plugs into, not another gadget competing for shelf space.

Designer: Anker

Getting there meant Anker had to kill its own family tree. eufy, soundcore, SOLIX and eufyMake are all being folded into the Anker name over the next couple of years, split into two groups called Anker For You and Anker For Home. CEO Steven Yang put it plainly on stage: category brands made sense when categories were separate, and they no longer are. Anker opened its first standalone store in Berlin the same week. A box asking to be the single brain of your house cannot credibly sit next to four differently branded limbs feeding it data, so the limbs had to go.

Physically, MindBase is trying to solve a problem NAS makers have ignored for two decades. Synology and QNAP enclosures still read as networking equipment, built for a closet, not a living room shelf. Anker’s two-bay chassis takes standard 3.5-inch drives and is styled to sit in the open, which is a real design bet given that spinning disks and a compute card generate heat that needs somewhere to go. Anker Innovations speaker Eric Villines framed the whole product around a gap that has bugged smart home owners for years: our houses are connected, not intelligent. A rule-based sensor cannot tell a raccoon from your kid, and MindBase’s pitch is that local memory finally lets it.

The security agent shows how far Anker is willing to let the box act alone. A yard sensor spots strangers at 15 meters and identifies family by phone credential at 30, then MindBase escalates on its own from a light, to a verbal warning, to active deterrence. For context, an entry-level Copilot+ laptop carries 40 TOPS and mostly uses it for background blur in video calls. Anker is asking for less silicon to make judgment calls about who belongs in your driveway at 2am, which is either the most useful application of on-device AI I have seen this year or the one I would want a very clear off switch for.

None of it has a price or a ship date yet, which is an odd silence for the product Anker built its entire rebrand around. Nest, Ring and Aqara are all circling the same local-hub idea without the drive bay, and that storage layer is MindBase’s real differentiator if Anker can actually ship it at a price regular households will pay. Whether the eufy hardware you already own is enough reason to wait and see is the only honest cliffhanger this launch leaves behind.

The post Anker MindBase Gives Your Old eufy Cameras a Real Brain and 48TB of Memory: Hands-on at IFA 2026 first appeared on Yanko Design.